DweiseSeller records · solution discovery
Contents
  1. Executive commercial assessment
  2. Evidence discipline
  3. Desired outcomes
  4. Problem-to-outcome map
  5. Solution space
  6. Existing alternatives
  7. Customer and buyer system
  8. Jobs and solution hypotheses
  9. Willingness to pay
  10. Risks and regulatory boundaries
  11. Discovery questions and evidence
  12. Distribution and partnerships
  13. Staged validation plan
  14. Decision gates
  15. Evidence and sources
Solution-space assessment · 1 August 2026

Find the valuable intervention. Do not assume it is software.

A commercial hypothesis report for UK marketplace-seller records: outcomes, competing interventions, buyers, price tests, regulatory boundaries and the evidence required before development.

Jurisdiction United KingdomResearch cut-off 1 August 2026Method amended 11 August 2026Starting point Problem mechanism establishedCommercial status Unvalidated
01 / EXECUTIVE ASSESSMENT

The opportunity is an outcome question before it is a product question.

The earlier research proves that fragmented records can require reconstruction. It does not prove how often that becomes painful, which intervention is appropriate, who owns the budget or whether a repeatable business exists.

Commercial conclusion

Keep seven models in play. The strongest learning vehicles are a bounded records-preparation pilot delivered within a compliant professional structure, an accountant-assisted handover model, and a structured workbook or diagnostic that tests seller behaviour. Guidance may be valuable but difficult to monetise alone. Lightweight tooling is a possible later expression of repeated work. A full platform has no present build case.

Do not choose the most scalable-looking model first. Choose the experiment that can prove a valuable outcome, a reachable buyer and acceptable delivery economics while exposing the least irreversible cost.
ConfirmedSellers must keep accurate sales, expense and supporting records; marketplace payouts can differ from gross activity; seller context and review remain necessary.[2][3][5]
IncumbentsNative exports, spreadsheets, capture tools, ecommerce connectors, ledgers and professional services already solve substantial parts. Any new offer must beat a capable combination, not an imagined vacuum.[10][16][18]
BoundaryCommercial record preparation can fall within accountancy-service-provider AML supervision even when delivered virtually or automatically. Paid interaction with HMRC for clients creates a separate tax-adviser registration question.[6][8]
UnknownPrevalence, urgency, usable file access, time saved, error values, buyer identity, price acceptance, renewal and partner willingness remain unmeasured.
First proofA prospective customer must expose the current workflow, supply real artefacts under appropriate controls, accept a clear scope and pay for a specific outcome.
Outcome firstExplain, evidence, reconcile, hand over and approve.
Buyer secondSeller, practice, or both must own a budget.
Delivery thirdManual, partner-led or tool-assisted remain open.
Build lastOnly automate stable, repeated and paid work.
02 / EVIDENCE DISCIPLINE

Four labels keep feasibility, demand and proof separate.

Commercial statements are deliberately weaker than legal or system facts. Provider claims establish intended scope and advertised price, not independent effectiveness.

Showing all evidence states
Confirmed evidence

Directly supported

Official rules, platform behaviour, published provider scope or an observed artefact. Example: HMRC requires accurate records and marketplace reports separate gross activity, deductions and payouts.

Reasonable inference

Mechanically credible

A conclusion that follows from confirmed facts but has not been measured in the target market. Example: a standard exception pack may reduce repeated accountant queries.

Assumption

Planning input only

A provisional belief used to design a test. Example: growing stock-based resellers feel more acute pain than disciplined single-channel sellers.

Testable hypothesis

Must survive behaviour

A falsifiable proposition with a buyer, intervention, outcome and rejection rule. Example: practices will pay per client for fewer preparation hours.

Provider evidence

Scope, not outcome proof

A2X, Link My Books, Dext, Xero and specialists publish capabilities and prices. These prove active alternatives and set expectations; they do not prove a residual gap or switching intent.[16][17][19]

Not established

No market-size shortcut

Platform-reporting totals, MTD scope and seller population counts do not reveal how many sellers have weak records, need help or will pay. They must not substitute for workflow and purchase evidence.

03 / DESIRED OUTCOMES

Sellers and accountants share totals; they value different forms of certainty.

The intervention should be judged by observable improvements in the work, not by feature count. Use the view control to separate the two sides.

Showing shared and role-specific outcomes
01
Complete source coverage

Know which seller accounts, periods, files, banks and evidence sources are present, missing or out of scope.

ObserveSource register, gap count, undisclosed sources found at review.
02
Explain money received

Move from each payout and bank deposit back to gross sales, fees, refunds, holds and other movements without guessing.

ObservePercentage of payout value explained; unresolved amount and age.
03
Capture evidence once, near the event

Retain receipts, stock cost, item provenance and business purpose before memory and documents disappear.

ObserveEvidence coverage, time-to-capture, follow-up rate.
04
Know what needs judgement

Separate routine records work from personal/trade, VAT, margin-scheme or deductibility questions that require informed decisions or advice.

ObserveExceptions routed; unsupported guesses prevented.
05
Reduce deadline anxiety and rework

Prepare on a predictable cadence, answer questions once and retain the supporting trail for correction or review.

ObserveActive hours, elapsed days, repeat questions, late work.
06
Receive a usable handover

Obtain source reports, reconciliations, evidence, mappings, exceptions and seller assertions in a consistent, inspectable pack.

ObserveTime to first usable working paper; pack acceptance.
07
Chase less, review better

Move effort from locating missing files and rebuilding payouts towards material review and professional judgement.

ObserveQuery count, preparation time, reviewer rework.
08
Protect engagement boundaries

Make the seller's factual responsibilities, unresolved items and approvals visible without implying that preparation software has made tax decisions.

ObserveDocumented approvals, escalations and exceptions.
09
Reproduce the result

Regenerate totals from retained sources and a visible correction history, rather than relying on a final spreadsheet with hidden manual edits.

ObserveReperformance success; unexplained changes; version lineage.
04 / PROBLEM → OUTCOME MAP

The overarching problem breaks into distinct jobs with different owners.

Stage 1 and Stage 9 are not blank product spaces. They are seller-dependent controls: an intervention can prompt, structure, preserve and challenge them, but cannot create undisclosed facts or approve on the seller's behalf.

Figure 1 · Causal problem-to-outcome map

From fragmented evidence to a reviewable handover

The green nodes are desired outcomes. Orange nodes require seller input; purple nodes require professional or commercial judgement. The table below is the authoritative written version.

flowchart LR
  A["Overarching problem
records must be reconstructed across systems"] --> B1["Stage 1
seller context, ownership and intent"] A --> B2["Marketplace activity
gross sales, fees, refunds and timing"] A --> B3["External evidence
stock, receipts and business purpose"] A --> B4["Settlement
payouts, reserves and bank deposits"] B1 --> C1["Prompted provenance
with evidence and uncertainty"] B2 --> C2["Preserved exports
and stable identifiers"] B3 --> C3["Linked source evidence
and cost schedules"] B4 --> C4["Gross-to-net reconciliation
with visible exceptions"] C1 --> D["Controlled classification
and adviser escalation"] C2 --> D C3 --> D C4 --> D D --> E["Stage 9
seller completeness review and approval"] E --> F["Accountant-ready handover
with drill-back and correction history"] F --> G["Desired outcome
less rework, defensible records and informed filing"] classDef seller fill:#fdf0e4,stroke:#ef791f,color:#0e2349; classDef control fill:#e7f0f2,stroke:#3e7a8c,color:#0e2349; classDef judgement fill:#eeeaf6,stroke:#7258a5,color:#0e2349; classDef outcome fill:#e7f3ef,stroke:#2c8e6d,color:#0e2349; class B1,B3,E seller; class B2,B4,C1,C2,C3,C4 control; class D,F judgement; class G outcome;
Written equivalent

Seller context and acquisition intent, marketplace activity, external evidence and settlement information converge into controlled classification and reconciliation. The seller then reviews completeness and confirms the factual picture. A handover preserves the path from source to totals and exceptions. The intended outcome is lower rework and more defensible records, not automatic tax correctness.

Know

Seller-only context

Ownership, acquisition, intention, shared activity and undisclosed accounts. Relevant gaps: provenance capture and Stage 1 prompts.

Collect

Parallel evidence

Marketplace exports, supplier proof, stock cost, postage, bank and payment routes. Relevant gaps: source capture and weakly documented channels.

Explain

Identity and settlement

Normalise schemas, retain identifiers, bridge gross activity to payout and expose timing or exceptions.

Approve

Meaning and handover

Controlled mappings, seller review, unresolved queries, source-to-working-paper lineage and Stage 9 confirmation.

Problem-to-outcome map and legitimate intervention choices. Alternatives may be combined; no row implies one preferred product.
Problem / gapWhy it mattersDesired outcomeManual / managedGuidance / advisoryTemplate / workbookAccountant-assistedLightweight toolFull platformEvidence status
Stage 1 · context, ownership and intentPlatforms cannot know why an item was acquired or whether activity is complete.Seller records facts, evidence and uncertainty near the event.Interview and evidence checklist.Explain which facts matter; refer judgement.Provenance register and item/bundle schedule.Accountant sets questions and escalation rules.Prompts, reminders and evidence links.Can orchestrate capture, but cannot originate truth.Mechanism confirmed
Marketplace exports and identifiersDifferent files, dates, schemas and retention create missing or duplicated activity.Complete source register with original files and stable IDs.Collect, inventory and quality-check exports.Channel-specific download instructions.File manifest and coverage checklist.Practice requests standard source set.Validate format, dates, duplicates and missing ranges.Direct integrations and central archive.Structure confirmed
Gross, deductions, settlement and bankA payout is not revenue; timing and reserves create legitimate differences.Explain each payout and bank receipt with visible exceptions.Reconciliation workbook and manual review.Teach the gross-to-net logic.Roll-forward and exception tabs.Reviewer accepts or queries reconciliation.Parse, match, total and flag.Automated feeds, rules and bank connection.Mechanism confirmed
External costs, stock and evidenceMarketplace data omits purchase cost, receipts, off-platform expenses and business purpose.Evidence-linked costs with missing-proof exceptions.Document chase, schedule and matching.Explain record duties and capture habits.Receipt index, stock and COGS schedule.Professional review of treatment.Capture, OCR, matching suggestions and reminders.Receipt, inventory and accounting modules.Need confirmed
Classification and tax-sensitive decisionsPersonal/trade, VAT, margin-scheme and deductibility questions depend on facts and judgement.Routine records separated from escalated decisions; rationale retained.Gather facts but do not make unqualified decisions.Generic education or regulated personalised advice.Decision log and adviser-question list.Qualified professional decides within engagement.Rules may suggest, never silently conclude.Broad tax engine and filing workflow.Advice boundary
Stage 9 · completeness reviewUndisclosed sources or missing periods can remain invisible after tidy processing.Seller sees sources, gaps and exceptions and gives informed confirmation.Guided review and signed statement.Explain responsibility and consequences.Completeness checklist and approval page.Practice obtains client confirmation.Review queue, prompts and immutable approval.Workflow and audit trail, still seller-owned.Responsibility confirmed[5]
Handover and drill-backNo universal accountant-ready format; missing lineage creates repeat work.Reviewer can reproduce totals, inspect sources and resolve exceptions.Prepared evidence pack and walkthrough.Handover checklist.Standard workbook and source index.Co-designed practice format and acceptance criteria.Generate pack, manifest and exception log.Practice portal, ledger posting and submission.Outcome inferred
05 / SOLUTION SPACE

Seven models trade outcome depth against trust, labour and regulatory exposure.

They are alternatives, bundles or stages of learning·not a maturity ladder that every business should climb.

Help people knowGuidance and advisoryEducation, clinics, checklists and referral.
Help people organiseTemplates and structured workbooksRepeatable capture, reconciliation and handover structures.
Do work with or for themManaged, preparation and partner modelsHuman handling, review and professional trust.
Automate proven workLightweight tools or a platformOnly after inputs, rules, exceptions and buyer economics repeat.
Showing all seven models
Model 1

Manual or managed service

A person collects, organises and checks files with the seller, producing a bounded output.

Direct outcomeHigh trustHigh labour
  • Feasibility: possible before software, but scope and training must be strict.
  • Regulatory: commercial record preparation is likely within ASP AML scope; personalised advice adds risk.
  • Support: missing files, anxiety and edge cases can become open-ended.
  • Scale: capacity, process and partner leverage·not code·set the ceiling.
Commercial hypothesis

A high-intent backlog or deadline buyer may pay for a one-off outcome; margin fails if cases cannot be bounded.

Model 2

Guidance and advisory support

Generic education, office hours, diagnostics or personalised professional guidance.

Enabling outcomeTrust variesContent can scale
  • Feasibility: generic guidance is fast; useful personalised advice requires expertise.
  • Regulatory: basic information identical for everyone may sit outside ASP registration; customer-specific tax advice does not.
  • Support: users may expect answers to fact-specific tax questions.
  • Differentiation: clarity and marketplace specificity, but free HMRC material is a strong substitute.
Commercial hypothesis

Guidance may acquire or qualify customers better than it monetises alone; a clinic may sell only when paired with a concrete output.

Model 3

Templates and structured workbooks

A seller-controlled workbook for provenance, files, settlement, evidence, exceptions and handover.

Conditional outcomeEasy to testLow delivery cost
  • Feasibility: high for known formats; upkeep and usability still matter.
  • Trust: transparent formulas help, but users must believe the structure is appropriate.
  • Regulatory: a generic template is lower exposure; completing or reviewing it for a client changes the analysis.
  • Support: formula errors, import changes and classification questions can erase the low-cost advantage.
Commercial hypothesis

A low-ticket product can test self-efficacy and capture behaviour, but free spreadsheets and accounting tools constrain price and differentiation.

Model 4

Reconciliation and records-preparation service

A narrower service focused on source inventory, gross-to-net reconstruction, bank matching and exception preparation.

Strong outcome fitAML likelyQA critical
  • Feasibility: high if channels, period and transaction limits are explicit.
  • Trust: requires secure file handling, visible lineage and professional accountability.
  • Support: exceptions and incomplete history drive cost.
  • Differentiation: marketplace-specific mappings and a useful exception pack, not generic data entry.
Commercial hypothesis

Customers may pay more for “explain and prepare” than for a workbook, especially at backlog or accountant-query triggers.

Model 5

Accountant-assisted or partner model

A supervised practice owns the client, professional decisions and filing; a specialist process or operator prepares the mechanical work.

High trust fitPartner dependencyBatch potential
  • Feasibility: requires co-designed outputs, responsibilities and commercial terms.
  • Regulatory: a narrow all-customers-supervised exemption may apply only if every HMRC condition is met; otherwise supervision is still needed.[6]
  • Support: partner onboarding and file standards replace some seller support.
  • Scale: fewer buyers can refer many clients, but sales cycles and revenue sharing rise.
Commercial hypothesis

Practices pay for capacity, consistency and fewer queries when the offer complements rather than competes with their advice relationship.

Model 6

Lightweight tool

A focused parser, diagnostic, reconciliation assistant, evidence checklist or pack generator.

Narrow outcomeDemand unknownScalable if stable
  • Feasibility: technically plausible for known files; real exceptions and format drift are unmeasured.
  • Trust: users need transparent totals, source drill-back and clear failure states.
  • Regulatory: pure software is treated differently from client service, but data protection, claims and any HMRC interaction remain important.
  • Support: import failures and accounting questions may dominate a low price.
Commercial hypothesis

A tool becomes attractive only when repeated manual work, supported formats and a recurring buyer are already visible.

Model 7

Full software or platform

Direct integrations, evidence capture, ledger logic, collaboration, tax workflow and possibly HMRC submission.

No build caseHighest complexityBroad demand unproved
  • Feasibility: technically possible in parts, commercially and operationally unproved end to end.
  • Trust: financial controls, security, auditability and professional acceptance must all be earned.
  • Regulatory: software provision alone has specific exclusions, but filings, client services, banking and advice add separate regimes.
  • Differentiation: established connectors and ledgers already cover supported channels.
Commercial hypothesis

A platform is a later organisational form, not the current solution. Stop unless narrower paid evidence exposes a durable cross-customer core.

Comparative commercial assessment

Qualitative assessment. “Potential” means worth testing, not validated demand.
ModelMeaningful outcomeOperational feasibility nowTrust requirementRegulatory / AML implicationSupport burdenScalabilityDifferentiationCommercial potential
Manual / managedDirect and adaptable; can expose real failure modes.Moderate with hard scope; poor if open-ended.High: financial files, judgement boundaries and human access.Likely ASP supervision for commercial preparation; obtain advice before launch.High and case-sensitive.People/process constrained.Niche expertise and outcome quality.Useful test vehicle
Guidance / advisoryEnabling; weak if the user still cannot do the work.High for generic content; lower for personalised advice.Moderate to professional.Generic basic information lower exposure; specific advice may be in scope.Medium; questions expand.High for content, low for advice.Clear marketplace examples.Lead or add-on
Templates / workbooksGood for disciplined users and transparent handover.High for selected formats.Moderate: formula and scope credibility.Lower for self-service artefact; client completion/review changes boundary.Low to medium; edge-case driven.High distribution, weak defence.Usability, controls and specificity.Cheap behaviour test
Records-preparation serviceDirect for reconciliation, evidence and handover.Moderate to high within file and transaction limits.Very high: accuracy, privacy and liability.Likely ASP; no tax conclusions or HMRC interaction without correct structure.High until exceptions repeat.Productised-service potential.Mappings, exception taxonomy, reviewer acceptance.Strongest WTP test
Accountant-assistedDirect, with professional decisions retained by practice.Moderate; partnership design required.High, partly borrowed from practice.Potential supervised-customer route only under strict conditions.Medium; partner enablement heavy.Better buyer leverage, slower sales.Practice workflow and acceptance.Promising channel test
Lightweight toolDirect for one repeated mechanical job; incomplete elsewhere.Unknown until real-file benchmark.High for outputs; lower human access.Software boundary can be cleaner; data, claims and integrations remain.Medium; format changes and user setup.High if inputs stable.Unsupported niche formats or better exception handling.Earn after pilots
Full platformPotentially broad but still cannot supply seller-only facts.Low before standards and demand proof.Very high.Multiple regimes and professional acceptance.Very high across integrations and workflows.High only after substantial fixed cost.Difficult against incumbent stacks.Do not approve
06 / EXISTING ALTERNATIVES

Customers can already combine free, low-cost and professional options.

Switching will occur only when the current stack fails on an important outcome or costs more in time, risk or professional fees than the change.

Evidence boundary

Published capabilities and prices describe what providers intend to sell. Adoption, satisfaction, switching and real-file accuracy were not independently tested. “Why adopt, reject or switch” below is a reasonable inference to test in discovery.

Current substitutes and adjacent incumbents, assessed fairly against their intended scope.
Alternative / examplesOutcome already offeredWhy adoptWhy reject or deferSwitch / add triggerPrice or commercial signalResidual to testEvidence
Do nothing / year-end reconstructionDelays effort until a deadline or accountant request.No setup cost; may work when activity is trivial.Memory, exports and receipts may disappear; urgency concentrates.First large query, HMRC letter, missed expense or accountant surcharge.Free now; hidden time and risk later.Whether the trigger is common and painful enough to buy earlier help.Inference
Marketplace-native reports
eBay, Etsy, Shopify, Depop, Amazon, TikTok Shop
Authoritative own-system orders, fees, refunds, balances and payouts.Free, closest to source and sufficient for disciplined single-channel work.Schemas, periods and retention vary; no off-platform cost or seller intent.Multi-channel activity, missing IDs, accountant requests or cross-payout exceptions.Included with seller account.Whether users can obtain complete, usable exports in target channels.Official scope[10][12]
DIY spreadsheets and free templatesFlexible sales, cost, stock and reconciliation records.Low cost, transparent and familiar; spreadsheets can form part of an MTD digital-link chain.[4]Setup, formula drift, copy/paste lineage and no maintained schema.Volume, new channel, errors or adviser rejects the handover.Free to low ticket.Whether a controlled marketplace workbook is materially easier and trusted.Conditional
Receipt and evidence capture
Dext, Hubdoc
Document capture, extraction, storage, approvals and ledger integrations.Strong for expense evidence; familiar to practices; mobile capture.Does not by itself reconstruct all marketplace settlement or seller intent.Lost receipts, accountant requests or need for continuous capture.Dext advertises £24.17/month on annual billing; Commerce Lite from £5/month. Xero includes Hubdoc in listed plans.[18][19]Whether marketplace-specific evidence-to-item links add value.Provider claim
Ecommerce connectors
A2X, Link My Books
Supported-channel settlement posting, VAT mappings, payout reconciliation and ledger connection.Strong intended fit for established sellers on supported channels and accounting software.Subscription, setup, tax mapping and supported-channel limits; seller facts still required.Higher order volume, VAT, accountant recommendation or multi-channel complexity.Order/channel-based plans and trials; advertised accountant collaboration and onboarding.Whether Depop/Vinted, mixed activity or low-volume sellers remain underserved after capable setup.Provider claim[16][17]
General accounting software
Xero, QuickBooks, Sage, FreeAgent
Ledger, bank feeds, reconciliation, expenses, reports, adviser access and recognised tax functions.Broad system of record and accountant ecosystem.Marketplace parsing may require apps; setup can exceed a side seller's needs.Business formalisation, MTD, VAT, company status or adviser requirement.Xero advertises regular base plans from £7 to £65/month before promotions and add-ons.[18]Whether a new intervention complements rather than duplicates the ledger.Active incumbent
Bookkeeper or general accountantRecords preparation, accounts, filings, advice and ongoing relationship within engagement.Professional trust and broad context; may already know the client.Marketplace-specific work can be slow or priced as clean-up; seller still supplies facts.Deadline, complexity, tax uncertainty, company/VAT duties or lack of confidence.Fees vary by volume and scope; no single market price.Frequency and cost of marketplace-specific queries and reconstruction.Practice-dependent
Specialist ecommerce accountancyMarketplace bookkeeping, VAT, inventory, accounts and tax support.Specialist knowledge and end-to-end accountability.May be disproportionate for small side sellers; often quote-led and higher commitment.Growing turnover, international activity, VAT, FBA or recurring complexity.One ACCA-regulated provider advertises researched ranges of £195–£775 for multi-channel bookkeeping and £125/month combined-service entry; treat as a provider signal, not a market benchmark.[22]Whether a lower-complexity handover offer is useful or merely fragments service.Provider claim
Connected incumbent stack
Connector + ledger + capture + accountant
Can cover most supported settlement, banking, evidence, bookkeeping and filing work.Best available benchmark for serious supported-channel sellers.Cost, configuration, multiple subscriptions and unsupported seller-only facts.A new offer must be cheaper, easier or better on a narrow residual.Several recurring subscriptions plus professional fees.Whether any material job survives reasonable configuration.Benchmark required
Competitive test: every paid pilot should be benchmarked against the customer's current method and, where relevant, a capable connector-plus-ledger workflow. If the incumbent stack removes most of the effort at acceptable cost, the correct decision is to recommend it, narrow the target segment or stop.
07 / CUSTOMERS & BUYERS

The person with the records, the user doing the work and the buyer may be different.

Discovery must identify role ownership before value propositions or pricing are interpreted.

SubjectSeller / taxpayer / company

Owns the activity, facts, records and consequences.

UserSeller, bookkeeper or practice staff

Collects, reconciles, reviews or prepares the handover.

BuyerSeller or practice owner

Controls a personal, business or client-service budget.

Decision-makerTaxpayer, director or partner

Accepts risk, scope and ongoing process change.

Referrer / gatekeeperAccountant, community, platform or adviser

Provides trust and access·or blocks an unsafe or duplicative offer.

Segments are defined by job and complexity, not marketplace membership alone

Segment hypotheses. Priority means learning value, not proven attractiveness.
SegmentLikely jobUserPossible buyerTriggerCurrent alternativePotential interventionCommercial concernDiscovery priority
Personal-item disposerUnderstand whether anything needs recording or reporting.Seller.Usually nobody.Platform reporting message or HMRC anxiety.HMRC guidance, search, accountant question.Generic guidance and referral only.Low willingness to pay; risk of exploiting anxiety.Exclude from paid lead
Low-volume side-trade sole traderKeep simple records and prepare a defensible annual handover.Seller.Seller.Trading allowance, first return, MTD approach or accountant request.Spreadsheet, bank feed, low-cost software.Workbook, clinic or bounded annual pack.Price sensitivity and low recurrence.Test low-cost
Stock-based resale trader
eBay, Vinted, Depop
Link acquisition evidence, item cost, gross sales and payouts.Seller or bookkeeper.Seller or accountant.Growth, mixed personal/trade activity, missing costs or messy year end.Sheets, native exports, accountant clean-up.Records-preparation pilot, structured capture or narrow tool.Receipts and intent may be missing; C2C exports are uncertain. Official Vinted help describes in-app balance and withdrawal history, but no downloadable ordinary-seller finance export was verified from the official pages reviewed.[27]Primary learning
Maker / producer
Etsy, Shopify
Explain materials, fees, ads, refunds, stock and deposits.Seller or bookkeeper.Seller.More orders, second channel or accountant handover.Native CSVs, inventory tools, connector, ledger.Workbook or reconciliation service.Established tools may cover enough.Benchmark first
Established multi-channel / VAT sellerAutomate settlement, VAT mappings, COGS and accountant workflow.Finance staff or practice.Business or practice.Volume, VAT, international operations.A2X/Link My Books + ledger + specialist accountant.Only a proven unsupported exception or partner service.Strong incumbents, high technical and liability expectations.Not initial wedge
Backlog / records-rescue caseTurn incomplete historic files into an exception-led pack quickly.Service operator and seller.Seller or accountant.Deadline, HMRC enquiry, accountant refusal or surcharge.Accountant/bookkeeper clean-up.Bounded supervised preparation service.Selection bias, missing records, open-ended labour and AML.High WTP test
Generalist accountant / bookkeeperReceive consistent seller inputs and reduce non-billable reconstruction.Practice staff.Owner/partner.Recurring marketplace clients, MTD workload or capacity constraint.Checklists, client portals, junior staff, ecommerce apps.Partner pack, batch preparation or lightweight reviewer tool.Workflow variation, integration expectations and trust.Primary channel test
Specialist ecommerce practiceIncrease capacity on supported clients without losing control.Bookkeeping team.Partner / operations lead.Backlog, seasonal peak or unsupported small channels.Own stack and trained staff.Subcontracted processing, format support or overflow capacity.May see little differentiation or competitive conflict.Adversarial benchmark
08 / JOBS & HYPOTHESES

Each proposition names a situation, an outcome, a buyer and a way to kill it.

These are not roadmap items. They are competing explanations of what customers may value.

H1 · SELLER BACKLOG

Records-rescue outcome

Job
“When my accountant asks for records I cannot explain, help me turn the available files into a clear pack of totals, evidence and unresolved questions so I can progress without pretending gaps do not exist.”
Buyer
Growing reseller or side-trade seller.
Value proposition
A bounded, source-linked pack that explains payouts and shows exactly what remains missing.
Test
Paid, supervised one-off pilot on real files; compare time, queries and unresolved value with current method.
Reject if: qualified sellers will not pay the disclosed price, files are rarely obtainable, delivery remains unbounded or the accountant does not find the pack useful.
H2 · CAPTURE HABIT

Structured workbook and review routine

Job
“When I acquire stock or incur a cost, help me record the evidence and context once so year-end preparation does not depend on memory.”
Buyer
Stock-based seller; accountant may refer.
Value proposition
A marketplace-specific capture and completeness workbook with simple monthly review.
Test
Four-week diary and workbook trial; measure use without reminders, evidence coverage and completion.
Reject if: fewer than half of recruited target users continue after onboarding, the work is duplicated elsewhere or seller effort exceeds perceived benefit.
H3 · ACCOUNTANT HANDOVER

Practice-ready evidence pack

Job
“When a marketplace client sends year-end material, help my team see completeness, reconciliations, evidence and client assertions before we begin professional review.”
Buyer
Practice owner, partner or operations lead.
Value proposition
A consistent handover that lets staff review rather than reconstruct.
Test
Blind comparison of current and proposed packs; timed preparation; then request a paid second client.
Reject if: practices demand incompatible formats, time saved is immaterial, or no practice submits a second paid case.
H4 · PREPARATION CAPACITY

Accountant-assisted processing

Job
“When marketplace files create low-value mechanical work, give my supervised practice controlled preparation capacity while I retain client relationship, judgements and filing.”
Buyer
Generalist or ecommerce practice.
Value proposition
Marketplace-specialist preparation under agreed formats, controls and service levels.
Test
Paid batch of three to five clients under written responsibilities, QA and AML review.
Reject if: partner onboarding costs exceed saved delivery time, liability cannot be allocated acceptably or utilisation is too seasonal.
H5 · GUIDED TRIAGE

Guidance plus professional referral

Job
“When I do not know whether I have a simple records job or a tax question, help me organise the facts, identify uncertainty and reach the right kind of help.”
Buyer
Seller; referral partner may sponsor.
Value proposition
Plain-English triage that avoids giving a false personalised answer.
Test
Demo and paid clinic or sponsored referral trial; measure completion, appropriate routing and conversion.
Reject if: free official guidance meets the need, users still demand advice, or acquisition cost exceeds downstream value.
H6 · NARROW AUTOMATION

Reconciliation assistant

Job
“When I receive recurring marketplace exports and payouts, help me normalise, match and flag exceptions without rebuilding the same spreadsheet each period.”
Buyer
Seller, bookkeeper or practice.
Value proposition
Transparent import and exception handling for one proven channel-and-file combination.
Test
Manual wizard or clickable demonstration before code; later a concierge parser on paid files.
Reject if: formats vary too much, incumbent connectors already solve the job, support exceeds saved time or users will not grant data access.
H7 · FULL PLATFORM

All-in-one records and filing

Job
Combine capture, reconciliation, bookkeeping, advice workflow and submissions.
Buyer
Undefined.
Value proposition
Undefined beyond breadth already served by incumbent stacks.
Test
Use only as a negative control in interviews: ask which parts users actually need and already buy.
Current decision: rejected for build. Reconsider only if narrower paid work repeatedly proves a shared platform core and buyers explicitly reject capable stacks.
H8 · NO NEW BUSINESS

Incumbents are sufficient

Job
Customers configure existing reports, sheets, connectors, ledgers and professional services.
Buyer
No buyer for a new standalone offer.
Value proposition
Independent guidance towards the correct incumbent may be the honest outcome.
Test
Benchmark real cases with capable configuration and measure residual work.
Accept this hypothesis if: the capable-stack benchmark removes most material work at acceptable total cost and residual cases are rare or uneconomic.
09 / WILLINGNESS TO PAY

Price must be tested against the avoided work and the buyer's next-best option.

These ranges are experiment designs derived from existing alternatives and earlier internal assumptions. They are not recommended prices or evidence of demand.

Provisional price experiments. Show a real price before asking whether the offer is interesting.
Offer to testCandidate buyerPrice experimentCompeting anchorCommercial assumptionEvidence to collectReject / change signal
Self-guided diagnosticSide-trade seller.Free versus £19–£39 one-off.Free HMRC guidance and generic spreadsheets.Marketplace specificity and a clear output justify a small payment.Checkout conversion, completion, appropriate referral and refund/support requests.Payment collapses or completion is low; use as acquisition only.
Structured workbook packDisciplined stock-based seller.£39–£99 one-off; optional paid onboarding separately.Free templates, low-cost software and accountant checklist.Users pay for control, examples and a better handover rather than formulas.Paid conversion, unaided completion, support minutes and accountant acceptance.Support makes effective margin service-like or users duplicate an incumbent.
Group clinic / office hoursSeller cohort or community sponsor.£29–£79 per attendee; compare sponsored session.Free webinars, forums and accountant consultations.Shared education resolves routine questions without drifting into individual advice.Attendance, prepayment, question mix, advice escalations and downstream actions.Most value requires customer-specific judgement; move to supervised professional model.
Bounded annual records packGrowing reseller or backlog seller.£149, £249 and £399 fixed-scope offers tested sequentially, not discounted negotiation.Bookkeeper clean-up, accountant fee buffer and seller time.A concrete deadline outcome supports a higher one-off price.Qualified conversion, delivery minutes, exceptions, fully loaded cost, outcome acceptance and second-period intent.Median delivery cannot support at least 50–60% contribution after real labour and support, or price resistance persists among qualified cases.
Recurring quarterly preparationSeller or practice.£49–£149 per period within strict channel/order limits.Accounting-software subscriptions and monthly bookkeeping.Regular preparation prevents larger year-end cost and creates retention.Second-period payment, scope growth, support and accountant usefulness.Users only buy at deadline or recurring work is too variable.
Practice wholesale packGeneralist accountant/bookkeeper.£59–£149 per client pack or a paid three-client batch.Junior staff time, specialist subcontracting and write-offs.Practice gross margin improves even after markup or review.Before/after active time, realised fee, review time, error/correction and paid repeat case.Saved time is shifted into partner review or the practice cannot pass cost through.
Practice retainer / capacityPractice owner.£199–£599/month with included bounded cases; price only after batch evidence.Employee or contractor capacity and practice software.Recurring workload and service levels justify reserved capacity.Utilisation, seasonality, renewal, concentration and SLA burden.Demand is episodic, one practice dominates revenue or unused capacity destroys margin.
Lightweight toolSeller or practice user.£9–£29/month seller, or £5–£15/client/month practice; test annual and usage options.Xero from £7/month, Dext from £24.17/month and connector plans by volume/channel.[18][19]A narrow tool saves enough recurring time without becoming another unwanted subscription.Pre-order or paid prototype, activation, repeated use, support minutes and churn reason.Users prefer one-off use, expect broader accounting features or support cost exceeds price.

Commercial assumptions that must be made explicit

A1

There is a costly trigger

A deadline, accountant query, missing-cost concern or reconciliation failure creates urgency beyond general tax anxiety.

A2

The buyer can quantify value

Time saved, fee avoided, delay reduced or record quality improved is visible enough to support a purchase.

A3

Inputs are obtainable

Target sellers can access complete exports, bank records and enough external evidence to produce an honest outcome.

A4

Cases can be bounded

A channel, period, transaction count, currency and exception policy prevent fixed fees from becoming unlimited clean-up.

A5

Trust can be earned

Sellers and practices will share financial files under the proposed security, contract and professional structure.

A6

At least one route repeats

Either sellers renew for another period or a practice submits multiple clients; otherwise the business is episodic consultancy.

Do not forecast SaaS revenue yet: first measure paid conversion, fully loaded fulfilment, support, repeat purchase, channel cost and refunds. Founder time, partner review, compliance and selling effort must be priced as real costs.
10 / BOUNDARIES & FAILURE

The operating model can change the regulatory position even when the screen looks the same.

A self-service artefact, an automated client service, professional bookkeeping and paid HMRC interaction are not interchangeable.

Legal boundary, not legal advice

HMRC says recording, reviewing, analysing, calculating or reporting financial information for others can be accountancy services, including virtual or automated delivery. A commercial preparation or managed-service pilot should not launch until its AML supervisory position, engagement scope, insurance and data roles have been reviewed. Paid interaction with HMRC for a client creates an additional mandatory tax-adviser registration question from 2026.[6][8]

High boundary

AML supervision

Managed bookkeeping or records preparation may be an accountancy service. Controls can include supervision, customer due diligence, risk assessment, records and suspicious-activity processes.[7]

High boundary

Tax advice and HMRC agency

Do not make fact-specific trade, VAT, deductibility or margin-scheme conclusions without the right professional model. If paid work interacts with HMRC for clients, registration requirements apply even if the business does not call itself a tax adviser.[8]

High trust

Data protection and security

Financial exports contain personal and commercial data. Controller/processor roles, contracts, access, retention, deletion, incident response and appropriate security must be designed before real-file pilots.[9][24]

Dependency

Platform access and format drift

Exports, fields, retention, permissions and APIs change. A service can fail before reconciliation if the source cannot be obtained or its scope is misunderstood.

Reliability

Silent inaccuracy

A tidy total can be more dangerous than a visible exception. Every intervention needs rejection states, source lineage, correction history and material review.

Economics

Exception-driven labour

Missing receipts, currencies, partial exports and mixed activity can destroy fixed-fee margins. Scope is a control, not a marketing inconvenience.

Support

Advice-seeking and anxiety

Sellers may treat records help as a promise of tax correctness. Scripts, escalation and response limits must be tested, not added after sale.

Channel risk

Partner dependence and conflict

Practices can lend trust and volume but may resist data access, white-labelling, standard packs, revenue share or a supplier that appears to compete.

Strategic risk

Solving a rare residual

The capable incumbent stack may work well for the commercially attractive cases, leaving only low-value or high-liability exceptions.

Regulatory boundary by operating model

ActivityLikely boundaryBefore a pilotDo not assumeCommercial effect
Publish generic guidance or a self-service workbookHMRC guidance says basic tax information identical for everyone need not require ASP registration; product claims and data collection still matter.[6]Review content, disclaimers, accessibility, privacy and support scripts.A disclaimer cures personalised advice given in practice.Lower fixed compliance cost but weaker outcome and price.
Complete, reconcile or review a seller's records commerciallyLikely accountancy-service-provider activity, including virtual or automated service.Obtain specific professional/legal advice; arrange supervision, CDD, engagement and insurance as required.Calling the work “data processing” removes AML obligations.Raises price floor, onboarding and support cost.
Work only for supervised accountancy practicesA narrow exemption may apply only if all customers are supervised, there is no direct client business, the operator is inside AML controls/training/SAR arrangements, and each relationship has a written compliant contract.[6]Verify every condition and document operating responsibilities.Any accountant referral automatically makes the supplier exempt.Potentially attractive channel; partner governance becomes central.
Give customer-specific tax adviceAccountancy/tax-advice and professional-competence boundary.Qualified delivery model, scope, supervision and professional indemnity.A decision tree can safely determine every case.Higher value possible, but liability and trust rise sharply.
Interact with HMRC for paying clientsMandatory tax-adviser registration applies, subject to stated exceptions and staged timing.[8]Check registration window, conditions and AML status; define who is the agent.Providing software and acting for clients are the same activity.Keep outside initial scope unless a regulated partner owns it.
Store or process seller financial filesUK GDPR/controller-processor duties and security obligations.Data map, lawful basis, minimisation, DPA/terms, retention, encryption, access logging, deletion and incident plan.Email and consumer file sharing are acceptable by default.Trust controls are part of cost of delivery.
Connect directly to bank informationPayment-services/Open Banking dependencies may apply; use an appropriately authorised provider and obtain advice before designing access.[25]Avoid credential handling; test manual statement import first.Bank connectivity is a harmless convenience feature.Defer until clear paid value exceeds integration and compliance cost.

Failure modes to design into experiments

Failure modeEarly warningExperiment controlStop or change rule
Seller supplies incomplete accounts but signs a confident pack.Sources discovered after review; unexplained date gaps.Account register, date coverage, exceptions and explicit Stage 9 prompts.Never label complete without evidence and informed seller confirmation.
Operator makes unsupported personal/trade or tax classification.Free-text case judgements and seller requests for certainty.Fact gathering, uncertainty labels and professional escalation.Stop case at boundary; do not guess to finish a pilot.
Reconciliation forces balance through a plug.Unexplained adjustment, rounding or suspense grows.Tolerance policy, immutable source totals and exception queue.Output remains unreconciled until explained or accepted as an exception.
Support consumes the price.Long onboarding, repeated missing-file contact, platform-specific troubleshooting.Time log by intake, processing, exceptions, QA and support.Narrow scope, raise price, partner-route or stop.
Practice likes the demo but never sends a client.Compliments, no files, no paid batch and no named owner.Ask for one paid case with date and scope.Do not count letters of interest without operational commitment.
Seasonal one-off demand masquerades as recurring SaaS.All purchases occur at year end; no second-period use.Offer both one-off and recurring paths and request renewal payment.Model as episodic service/product, not subscription.
11 / DISCOVERY & PROOF

Ask for recent behaviour, artefacts and trade-offs·not opinions about a proposed app.

Interview questions should expose the current job before demonstrations test a specific intervention.

Seller discovery

  1. Walk me through the last time you prepared marketplace records. What started the work?
  2. Which accounts, marketplaces, payment routes and banks were involved?
  3. Show me the files or screenshots you actually used. Which were hardest to obtain?
  4. How did you move from gross sales to each payout and bank deposit?
  5. Where do stock cost, receipts, postage and other external costs live?
  6. When did an accountant or HMRC ask a question you could not answer immediately?
  7. What did you do, how long did it take, and what remained uncertain?
  8. What tools or people have you already tried? What worked well enough to keep?
  9. What would make you refuse to share files or change the current process?
  10. What did you pay last time, including software, professional fees and your own time?
  11. If this outcome cost £X under a fixed scope, what would you compare it with?
  12. Who besides you would need to approve the purchase or output?
Avoid: “Would you use software that automates this?” It invites politeness and hides the present workflow.

Accountant / bookkeeper discovery

  1. Walk me through the most recent marketplace-client handover from first receipt to accepted working paper.
  2. Which sources are routinely missing, compressed or unreliable?
  3. How many queries and active hours did the last case require, by staff grade?
  4. Which connector, ledger, portal or checklist did you use, and what remained?
  5. What does “ready for review” mean in your practice? Show an anonymised checklist or pack if permitted.
  6. Which decisions must remain with the seller, bookkeeper, accountant or tax specialist?
  7. How do you price messy records today, and who absorbs overruns?
  8. Would a standard pack fit your working papers, or create another reconciliation?
  9. What security, AML, insurance and subcontractor conditions would block a partner?
  10. Would you supply one paid client case? If not, what concrete condition is missing?
  11. Who owns procurement, data protection and implementation approval?
  12. What result would justify submitting a second client?
Ask for: a paid case, timed comparison or operational commitment·not only a letter of support.

Evidence required to validate or reject each live hypothesis

HypothesisDiscovery evidenceDemonstration evidencePilot / paid evidenceValidate provisionally whenReject or change when
H1 · Records rescueRepeated recent backlog episodes, artefacts, hours, fees and unresolved sums in a defined segment.Seller understands pack, scope and exception wording and still wants the outcome.At least five non-friendly qualified buyers pay a disclosed test price; accountants accept output.Measurable rework falls and delivery economics are positive within scope.No urgency/payment, missing inputs dominate or outcome duplicates accountant work.
H2 · Capture habitEvidence is currently lost because capture is delayed, not because users do not care. Record whether each participant was independently acquired or practice-sourced.Directly recruited sellers complete representative tasks without facilitator interpretation; practice-sourced sellers are measured as a separate cohort.Four-week continued use, improved evidence coverage and willingness to pay/renew, reported separately by acquisition cohort.Behaviour persists with limited reminders within the relevant cohort and accountants see cleaner inputs.Drop-off is high in either target cohort, a practice is effectively compelling completion, or existing capture tools are preferred.
H3 · Handover packPractices show recurring preparation/chasing cost and compatible acceptance needs.Blind reviewers locate sources, totals and exceptions faster than baseline.Two practices pay and each submits a second client or paid batch.Review time and queries fall without extra partner risk.Formats fragment by practice or saving is absorbed by extra review.
H4 · Partner processingPractices have a capacity problem they would outsource under defined controls.Roles, AML, data and QA model passes partner review.Paid multi-client batch meets service level and fully loaded margin.Partner renews and cases remain standardisable.Governance cost or liability exceeds capacity value.
H5 · Guided triageSellers cannot distinguish records work from advice and currently take poor actions.Users reach correct next steps without receiving personalised conclusions.Paid/sponsored conversion or valuable referral economics.Completion and appropriate routing are high with manageable support.Free guidance is sufficient or most questions need professional advice.
H6 · Narrow toolSame repeated mechanical steps and file formats appear across paid cases.Prototype is understood and trusted; source drill-back matters.Pre-order or paid use, repeated periods, low support and stable accuracy.Automation removes a material share of paid labour with acceptable maintenance.Incumbents win, format variance is high or support economics fail.
H8 · Incumbents sufficientTarget users can configure existing products or professional help at acceptable cost.Capable-stack walkthrough covers the desired workflow.Benchmark leaves little material residual across real cases.Recommend/incorporate incumbents and stop new-product work.A narrow repeated residual remains valuable and paid.
12 / DISTRIBUTION

Direct seller and accountant-led routes test different buyers, trust and economics.

Run them in parallel at small scale. Do not infer that an accountant referral proves practice willingness to pay, that direct interest proves low-cost acquisition, or that practice-supervised seller behaviour predicts independently acquired seller behaviour. Practice clients may be more established, deadline-driven and responsive under professional pressure. Tag every case by acquisition cohort and do not transfer completion, Stage 9 approval, payment or repeat-use evidence between routes.

Direct seller

Trigger-led acquisition

  1. Triggers: first accountant request, year end, MTD transition, platform-reporting message, missing costs, rejected handover or HMRC contact.
  2. Entry assets: plain-English source checklist, payout explainer, diagnostic and anonymised before/after pack.
  3. Channels to test: search, seller communities, reseller educators, bookkeeping content, marketplace-specific groups and referrals.
  4. Offer: a fixed-scope outcome with supported channels, period, file list and exclusion rules visible before payment.
  5. Risk: anxiety traffic, low budgets, community trust and high education/support cost.
Proof: source-attributed qualified leads, disclosed-price checkout/deposits, CAC including founder time, completed cases and repeat purchase.
Accountant-led

Trust and portfolio leverage

  1. Partners: generalist accountants, bookkeepers, ecommerce specialists, professional networks and practice-software consultants.
  2. Models: referral only, co-branded workbook, practice-paid batch, supervised subcontractor or later per-client tool.
  3. Offer: fewer preparation hours, standard inputs, visible client assertions and a pack that fits the practice's working papers.
  4. Governance: client ownership, AML, data controller/processor roles, QA, liability, branding, support and no tax judgement by the preparation layer.
  5. Risk: longer sales cycle, custom formats, margin share, concentration and fear of disintermediation.
Proof: a named owner, written operating scope, paid client batch, timed saving and a second client·not demo attendance.

Partnership hypotheses

Partner typeWhy they may engageWhy they may refuseFirst experimentEvidence of channel fitCommercial model to test
Generalist accountant / bookkeeperMarketplace cases create irregular preparation and client chasing.Small case volume, liability, incompatible working papers or existing staff process.One paid blind pack comparison and one live client.Second paid client and measured staff-hour reduction.Per-pack wholesale or small paid batch.
Specialist ecommerce practiceOverflow capacity, unsupported C2C channels or lower-value segment.Strong in-house stack, competitive threat or quality concern.Adversarial benchmark on a difficult but bounded case.Partner identifies a residual and commits capacity use.Subcontracted service with explicit client boundary.
Seller educator / communityUseful record-keeping content and member benefit.Tax anxiety, reputational risk or promotional fatigue.Sponsored source-checklist workshop with no product pitch.Qualified follow-through, not views or registrations.Sponsorship, referral or member discount only after evidence.
Marketplace / ecosystem providerSeller success and fewer reporting questions.Low priority, legal caution, long procurement and no reason to endorse one supplier.Interview ecosystem or partner managers after seller proof.Data access, distribution commitment or funded pilot.Integration/education partnership; not initial dependency.
Accounting software / app adviserImplementation help for a specific seller segment.Already has connector partners; narrow demand.Show residual after capable configuration.Partner recognises non-duplicative job and refers cases.Implementation add-on or referral.
13 / STAGED VALIDATION

Use the cheapest credible evidence at each stage; add software only after repetition.

Synthetic files can rehearse operations and demonstrations, but only consented real workflows and payments validate a market.

STAGE 0 · BOUNDARY

Make testing lawful

  • Before retaining any Stage 1 artefact, put discovery-data controls in place: consent, minimisation, secure transfer, access, retention and deletion. Prefer fully anonymised artefacts or no-copy screen walkthroughs where they answer the question.
  • Professional review of ASP/AML, tax-adviser and insurance position may continue alongside early discovery, but must pass before any Stage 3 client-record preparation.
  • Define excluded advice, supported files, no-HMRC scope and incident controls.
Exit evidenceDiscovery-data controls are active; a written operating boundary and safe real-file route are approved before Stage 3.
STAGE 1 · PROBLEM

Interview with artefacts

  • 15–20 target sellers across two candidate segments.
  • 8–12 accountants/bookkeepers across generalist and specialist practices.
  • Pre-register recruitment routes, founder hours, elapsed time and cash budget; treat them as early distribution evidence rather than free research inputs.
  • For each platform advanced to a pilot, verify the ordinary-seller input route, file or screenshot coverage, identifiers, date completeness and historical access before recruitment. Vinted remains unverified for a downloadable finance export.[27]
  • Observe recent workflow, files, hours, fees, exceptions and attempted alternatives.
Exit evidenceOne repeated costly job in a named segment, role ownership, recorded recruitment cost and a usable input route for every platform advanced.
STAGE 2 · DEMO

Compare interventions

  • Show workbook, sample handover and service scope·not an app concept.
  • Use existing synthetic sample pack for safe rehearsal only.
  • Pre-register the capable incumbent stack, configuration, operator skill and setup-time allowance, or have an independent practice operate it.
Exit evidenceUsers understand trade-offs and choose a concrete outcome at a disclosed price; the incumbent comparison is reproducible rather than self-graded.
STAGE 3 · PAID PILOT

Deliver manually

  • Five to ten bounded seller cases or two paid practice batches.
  • Log active time, missing files, exceptions, QA, support and corrections.
  • Measure before/after outcome and request a second purchase.
Exit evidencePayment, outcome improvement, acceptable errors and viable fulfilment.
STAGE 4 · STANDARDISE

Earn automation

  • Identify repeated inputs, rules, exceptions and outputs.
  • Test a no-code wizard or concierge parser behind the service.
  • Pre-sell the narrow automated outcome.
Exit evidenceStable core removes material labour and users pay to repeat it.
Sequence rule: if a stage fails its exit evidence, narrow, change the buyer/intervention or stop. Do not compensate for weak demand by adding features, more channels or a more polished demonstration.

Experiment register

ExperimentQuestion answeredMinimum artefactPrimary measureWhat does not countNext decision
Workflow interviewIs the problem recent, repeated and costly?Fully anonymised or controlled redacted export, checklist, email query or no-copy screen walkthrough.Observed steps, time, exceptions, alternatives, recruitment source and founder acquisition effort.General frustration, hypothetical interest or retained personal data without Stage 0 controls.Choose or drop segment/job.
Pack comprehension testDoes the output communicate completeness and exceptions?Synthetic source-to-pack example.Unprompted task completion and interpretation errors.Visual preference.Revise output or discard concept.
Price page / depositWill a qualified buyer commit money?Scope, exclusions, security model, price and delivery date.Paid conversion from qualified prospects.Email sign-up, “would buy”, discount-only sale.Keep price, change value, or stop.
Manual seller pilotCan the outcome be delivered safely and profitably?Controlled workbook, checklist, exception log and QA.Outcome delta, fully loaded delivery and acceptance.Founder heroics or free friend cases.Standardise, raise price, narrow or stop.
Practice batchDoes the accountant channel create repeatable portfolio value?Written responsibilities and three-client batch.Review time saved, second paid batch and margin.One design-partner meeting.Partner-led model or direct-only.
Incumbent benchmarkIs there a meaningful residual?Same files through a pre-registered capable stack and current method, preferably operated or reviewed by an independent practice.Coverage, setup, exceptions, drill-back, total cost, operator skill and configuration record.Feature comparison without files or a founder-selected weak configuration.Complement, differentiate or stop.
Concierge automationWhich repeated steps deserve code?Manual front end with scripts or no-code behind it.Labour removed, accuracy, maintenance and repeat payment.Technical demo on synthetic happy paths.Approve narrow prototype only.
14 / DECISION GATES

No software build is approved by this report.

Proceed only through explicit gates. Thresholds below are provisional management rules for small-sample validation, not market facts. Before seeing results, put them in a dated protocol and share it with a named independent stakeholder or adviser so a solo operator cannot quietly redefine success.

Problem mechanismConfirmed

Official evidence establishes the need for accurate records, external evidence, gross-to-net explanation and seller review.[3][4][5]

Commercial painUnconfirmed

Frequency, active hours, error values, urgency and incumbent failure are not measured in a defined target segment.

Willingness to payUnconfirmed

Earlier price ranges are hypotheses. No current evidence reviewed establishes paid conversion, renewal or price elasticity.

Best operating modelOpen

Service, partner, workbook and lightweight-tool models must compete on outcome, lawfulness, trust, support and economics.

Gate A

Problem and segment

  • At least one narrowly defined segment shows the same recent job across multiple independent cases.
  • Artefacts confirm the workflow; the problem survives a pre-registered or independently operated capable-incumbent benchmark.
  • A buyer and trigger are identifiable, and every platform advanced has a usable input route.
Proceed: choose one job/segment. Narrow: if pain clusters by channel, input route or circumstance. Stop: if exposure rarely becomes costly failure.
Gate B

Outcome and trust

  • Demonstration users correctly understand sources, exceptions, boundaries and responsibility.
  • Sellers/practices are willing to share real files under proposed controls.
  • The output saves measurable work or materially improves record quality.
Proceed: paid pilot. Change: simplify intervention or move to partner delivery. Stop: if trust/access blocks the outcome.
Gate C

Payment and delivery

  • Five or more non-friendly qualified buyers pay a pre-declared seller price, or two practices pay for batches.
  • Recruitment source, elapsed sales effort, founder hours and direct spend are recorded for every buyer even though channel repeatability is decided at Gate D.
  • Fully loaded fulfilment supports at least a provisional 50–60% contribution margin within scope.
  • No material silent error; exceptions and corrections are reproducible.
Proceed: repeat and standardise. Change: price, scope or channel. Stop: if value cannot cover acquisition, labour, compliance and support.
Gate D

Repeat and channel

  • Direct-route repeatability requires independently acquired sellers to complete and pay for a second period; practice-sourced seller behaviour cannot satisfy this test.
  • Practice-route repeatability requires a second paid batch from the practice, with seller response measured separately from direct cases.
  • Acquisition source and fully loaded sales effort are measurable by route.
  • No partner supplies a dangerous share of revenue without alternatives.
Proceed: choose direct, partner or hybrid model only for the route whose evidence passes. Change: treat as one-off product/service. Stop: if repeatability is absent and one-off economics are weak.
Gate E

Automation readiness

  • Paid cases share stable inputs, rules, exceptions and outputs.
  • A narrow automated step removes a material part of delivery without moving errors into support.
  • Buyers pre-order or pay for the tool-assisted version.
Proceed: build the smallest supported step. Narrow: keep service wrapper. Stop: if variance and maintenance exceed labour saved.
Gate F

Platform approval

  • Several narrow paid workflows converge on a common platform core.
  • Incumbent stacks cannot meet the same outcome at acceptable total cost.
  • Security, regulation, support, distribution and funding are evidenced.
Current state: fail. A full platform remains stopped. Technical feasibility does not move this gate.

Final commercial assessment

DirectionCurrent assessmentWhyNext proof
Accountant-supervised records-preparation pilotTest firstHighest direct outcome and strongest learning, with professional trust and boundary discipline; labour and partner dependence remain risks.One compliant practice partner, paid batch, timed baseline and second case.
Structured workbook / capture routineTest alongsideFast, reversible and useful for testing Stage 1/Stage 9 behaviour; price and differentiation may be weak.Four-week paid/committed use and accountant acceptance.
Direct bounded reconciliation packTest only after boundary reviewPotential high-intent WTP test, but likely ASP supervision, data handling and support cannot be treated informally.Written legal/AML route and five paid fixed-scope cases.
Guidance / triageUse as test or channelCan reduce confusion and qualify needs; free alternatives and advice creep constrain standalone economics.Paid or sponsored completion and appropriate referral conversion.
Lightweight reconciliation toolEarn laterPlausible only after paid cases prove repeated formats, rules, exceptions and recurring use.Concierge automation removes material labour and receives prepayment.
Full marketplace records platformStopNo broad residual, buyer, WTP, distribution or operating case is validated; capable incumbent stacks already exist.None before Gates A–E pass.
Recommended next action

Recruit discovery participants in two deliberately different groups: stock-based resale sellers with recent accountant or backlog friction, and generalist practices that handle marketplace clients. Complete the regulatory boundary review in parallel. Use interviews and artefacts to choose one outcome, then compare a workbook demonstration with a supervised manual pack. Ask for payment before any parser, integration or platform work.

15 / EVIDENCE & SOURCES

Official evidence establishes duties and boundaries; commercial conclusions remain tests.

Original research cut-off: 1 August 2026. The Vinted input check and validation-method amendments were added on 11 August 2026. Refresh provider scope, input routes and prices before a pilot and at least quarterly while testing; record the exact check date. Regulatory conclusions require case-specific professional review before a live service.

Starting material

The specified problem-architecture report was treated as the causal baseline, with the incumbent landscape, operating-validation plan and projections report used to identify prior assumptions. Earlier product recommendations and prices were not carried forward as validation.

Source hierarchy

HMRC/GOV.UK, ICO and FCA for duties and boundaries; official platform and provider materials for intended scope and current advertised terms; specialist-provider pricing only as a labelled commercial signal.

Commercial method

Every live proposition includes a job, buyer, intervention, outcome, evidence requirement and rejection rule. No population prevalence, demand percentage or revenue forecast is inferred from desk research.

Incumbent fairness

Existing categories are assessed against intended jobs. A boundary is not a defect. The benchmark is a capable combination, not an isolated feature list.

Regulatory caution

The report identifies likely boundaries, not legal advice. The operating facts·who handles files, who decides, who contracts and who interacts with HMRC·determine the position.

Limitations

No representative survey, consented client files, paid experiments, accountant working papers, independent provider performance test or production security review was available.

Direct sources

[1] Problem Architecture and Evidence AssessmentDweise research · 1 August 2026Causal architecture, lifecycle, controls, incumbent scope and unresolved commercial evidence.
[2] Online-platform income: when to tell HMRCHMRC / GOV.UK · checked 1 August 2026Personal possessions, profit-seeking goods, trading allowance and calendar/tax-year distinction.
[3] Self-employed records to keepHMRC / GOV.UK · checked 1 August 2026Sales, expenses, accurate and identifiable business transactions, proof and stock.
[4] MTD for Income Tax: create digital recordsHMRC / GOV.UK · updated 16 July 2026Digital records, links, spreadsheet/CSV transfers, supporting documents, bank-feed limits and taxpayer responsibility.
[5] MTD for Income Tax: submit the tax returnHMRC / GOV.UK · checked 1 August 2026Correct-and-complete declaration and written client confirmation for agent submissions.
[6] Accountancy service providers: AML registrationHMRC / GOV.UK · checked 1 August 2026Recording/reviewing financial information, virtual and automated delivery, generic-information boundary and supervised-customer conditions.
[7] Money-laundering supervision responsibilitiesHMRC / GOV.UK · checked 1 August 2026Customer due diligence, risk assessment, controls and record keeping.
[8] Mandatory tax-adviser registrationHMRC / GOV.UK · updated 29 May 2026Paid interaction with HMRC for clients, staged timing and software-provider distinction.
[9] Controller and processor contractsInformation Commissioner's Office · checked 1 August 2026Data roles, contracts, security, sub-processors, end-of-contract and audit provisions.
[10] eBay: reconcile sales transactionseBay UK · checked 1 August 2026Transaction IDs, gross/net, fees, refunds, statements and reports.
[11] Etsy sold-transaction spreadsheetsEtsy Help · checked 1 August 2026Separate order, sales and deposit CSVs.
[12] Shopify payout reconciliationShopify Help · checked 1 August 2026Balance activity, deductions, holds, adjustments and warning that the report is not a revenue statement.
[13] Depop sales downloadDepop Help · checked 1 August 2026CSV sales, fees, shipping and taxes.
[14] Amazon SP-API report typesAmazon developer documentation · checked 1 August 2026Specialised report families and schema variety.
[15] TikTok Shop finance report guideTikTok Shop UK Seller University · checked 1 August 2026Order, statement, payment and reserve report layers.
[16] A2X integrations and pricing modelA2X official materials · checked 1 August 2026Supported channels, payout reconciliation, accounting integrations and order/platform-based plans. Provider claims only.
[17] Link My Books pricing and scopeLink My Books · checked 1 August 2026Supported sales channels, order/channel pricing, accountant users, onboarding, historical data and stated advice boundary. Provider claims only.
[18] Xero UK plans and ecommerce app collectionXero official materials · checked 1 August 2026Current advertised regular base prices, bank/document functions, Hubdoc inclusion and connector ecosystem. Provider claims only.
[19] Dext business pricingDext · checked 1 August 2026Document capture, extraction, storage, integrations, base price and Commerce Lite add-on. Provider claims only.
[20] QuickBooks UK MTD softwareIntuit · checked 1 August 2026Digital records, bank and adviser collaboration claims. Provider claims only.
[21] Choose MTD for Income Tax softwareHMRC / GOV.UK · checked 1 August 2026Record-keeping, bridging and submission functions vary by product.
[22] E-commerce accountant fee guide and service pricingE-commerce Accountants UK / Tidy Money Ltd · checked 1 August 2026Provider-published April 2026 fee signals and cost drivers; not independent market evidence.
[23] Draft MTD for Income Tax engagement scheduleICAEW · June 2025; checked 1 August 2026Professional draft on client source-information, documents and completeness responsibilities; not legislation.
[24] Guide to data securityInformation Commissioner's Office · checked 1 August 2026Risk-based technical and organisational security measures; confidentiality, integrity and availability.
[25] Account-information service security and registrationFinancial Conduct Authority · checked 1 August 2026Regulatory and security expectations for account-information services.
[26] VAT margin-scheme recordsHMRC / GOV.UK · checked 1 August 2026Purchase/sales evidence and scheme-specific record requirements.
[27] Vinted: balance and withdrawal historyVinted Help · checked 11 August 2026Official help describes in-app withdrawal history by year, amount and date. No downloadable ordinary-seller finance export was verified from the official Vinted pages reviewed; treat that absence as an input-feasibility gap, not proof that a feature cannot exist.
← All reports