Seller records, made reviewable.
An evidence-based specification for turning fragmented UK marketplace activity into a reconciled, categorised and traceable pack that a seller can approve and an accountant can actually use.
Proceed to a bounded, paid validation programme. Do not build a broad tax or bookkeeping platform yet.
Conditional go · evidence requiredA useful service is plausible. A defensible business is not yet proved.
The prior Dweise research is directionally sound, but the operating model needs one important correction: the first offer must be seller-controlled software or practice-supervised processing, not an informal Dweise records-rescue service.
Dweise should own the mechanical job: source ingestion, gross-to-net payout reconciliation, exception detection, evidence lineage and a consistent handover. The seller owns factual classifications; the accountant or bookkeeper owns professional judgement and filings.
What is fact, what is normal practice and what remains a hypothesis.
Four evidence classes prevent Dweise from turning professional conventions or commercial guesses into compliance claims.
The records problem is real
HMRC requires self-employed people to retain records of sales, expenses and supporting proof; traditional accounting also needs year-end stock, receivables, payables, bank balances, capital introduced and drawings. Companies have wider duties covering money in and out, assets, liabilities, stock and goods bought and sold.[1][2]
A payout is not revenue
Marketplace documents themselves distinguish orders, fees, refunds, labels, balances and payouts. Shopify expressly warns that its payout reconciliation report is not a revenue statement; Etsy explains that deposits include sales and account activity after fees, refunds and timing differences.[13][17]
Platform reporting does not decide tax
HMRC says selling personal possessions is usually different from buying or making goods to sell for profit. Platform reporting is not itself a tax threshold, and a personal possession sold for £6,000 or more may raise separate Capital Gains Tax questions.[6][34]
Commercial demand
Competitors prove that sellers and practices buy ecommerce bookkeeping automation, but they do not prove that smaller eBay/Vinted/Depop sellers will pay Dweise. Price, trust and repeat use must be measured in paid pilots, not inferred from search volume or HMRC reporting anxiety.
The accountant-ready pack is a controlled evidence bundle, not a tax return.
Its job is to let a reviewer move from a total back to the marketplace line, payout, bank deposit and seller decision that produced it.
Written equivalent
Source files enter a versioned canonical ledger. Automated controls flag duplicates, gaps and payout differences. The seller resolves factual questions; the professional resolves accounting and tax questions. Dweise then issues a locked, versioned pack and optional import files.Record requirements matrix
| Area | Minimum pack content | Evidence class | Important boundary |
|---|---|---|---|
| Seller identity and scope | Legal name; trading name; address; contact; sole trader, partnership or company; UTR if supplied; company number; VAT number; accounting period; tax year; accounting basis; agent/practice details. | Common Entity and VAT details become legally important where applicable.[2][3] | Dweise should collect only what the workflow needs and must not imply that identity capture itself satisfies an accountant's AML checks. |
| Sales and income | Order ID, item/SKU, order date, dispatch date where available, item price, buyer-paid postage, discounts, marketplace-collected tax, currency, gross amount and source row. | Legal All sales/income must be recorded.[1] | Marketplace-reported consideration, gross buyer payment and accounting revenue may differ. |
| Fees, refunds and adjustments | Marketplace fee, payment fee, advertising, listing fee, shipping label, refund, chargeback, dispute, reserve, hold, credit and non-sale charge, linked to order and payout where possible. | Common Needed to explain profit and deposits; official marketplace reports expose these categories.[10][11] | Do not net everything into “sales”; preserve both gross activity and deductions. |
| Payout and bank reconciliation | Payout ID, initiation and deposit dates, payout currency, bank reference, opening/closing marketplace balance, component lines, bank match, variance and timing explanation. | Common Bank statements are supporting proof and year-end balances are required under traditional accounting.[1][19] | A zero difference proves arithmetic, not correct tax treatment. |
| Stock and cost of goods sold | Opening stock, purchases, acquisition evidence, item or pool cost, returns to stock, write-offs, closing count, valuation method and COGS bridge. | Legal Stock proof and traditional-accounting year-end stock are explicit HMRC records; companies must retain stocktaking support.[1][2] | Cash basis, traditional accounting, second-hand stock and stock valuation require circumstance-specific treatment.[9] |
| Business expenses | Date, supplier, category, amount, VAT shown, payment source, receipt/invoice reference, business-use percentage, evidence status and notes. | Legal Expenses and proof must be kept.[1] | Allowability and mixed-use apportionment are professional decisions. Stock for resale, raw materials and direct production costs are examples of potentially allowable costs.[8] |
| Personal items versus trade | Seller decision per item or rule-set; reason; date first owned; original purpose/cost if known; evidence; mixed-payout bridge; unresolved high-value disposals. | Conditional HMRC focuses on the facts and intention, not the marketplace name.[6] | Dweise should ask factual questions, never assert trading status or CGT treatment. High-value personal possessions need professional review.[34] |
| Multiple channels, currencies and providers | Channel and account register, complete date range, native currency, GBP amount, exchange-rate source, payment provider, cross-account transfer and duplicate key. | Common Completeness and conversion support. Accounting frameworks generally translate foreign-currency transactions at the transaction-date spot rate, with practical use of averages where appropriate.[33] | Marketplace conversion may be accepted for a settled transaction, but the accountant decides policy and period-end retranslation. |
| Assets and capital spend | Asset description, invoice, acquisition date, cost, VAT, business-use percentage, date available for use, disposal proceeds and prior claims. | Conditional Capital allowances can apply to business assets; claims differ by entity and asset.[7] | Do not automatically expense or claim an allowance. Cars, mixed use, cash basis and disposals need advice. |
| VAT | VAT registration and schemes; tax point; net value; VAT rate and amount; output/input VAT; invoices and credit notes; marketplace-deemed-supplier flags; country and evidence; VAT account mapping. | Legal when VAT-registered VAT records include everything bought and sold, invoices, credit notes and a VAT account; MTD requires specified digital records.[3][5] | VAT registration, rates, place of supply, second-hand margin schemes, imports and marketplace liability require professional advice. The current UK registration threshold is £90,000, subject to exceptions.[32] |
| MTD for Income Tax | Income/expense transaction date, amount, category where required, source and correction history; exportable period summaries. | Conditional Mandatory phasing is based on prior-year qualifying gross self-employment and property income: over £50,000 from April 2026, £30,000 from 2027 and £20,000 from 2028.[4] | Dweise should say “MTD-ready records export”, not “MTD compliant”, until it is compatible software and the full digital-link workflow is tested. |
| Missing records and queries | Exception ID, amount, source, issue, confidence, materiality indicator, owner, due date, evidence request, response, resolution and approval. | Common Accountants commonly request bank records, stock, sales, purchases, debtors, creditors and explanations for gaps.[19][20] | “No evidence” must remain visible; never convert uncertainty into a guessed category. |
A realistic synthetic pack, with every assumption visible.
Public platform documentation provides report concepts and schemas, but reusable first-party example files are inconsistent. The sample pack therefore uses non-personal synthetic lines and clearly states where live-format testing remains necessary.
Public record-source atlas
| Platform | Official records available | Public sample position | Dweise decision |
|---|---|---|---|
| eBay | Seller Hub order CSV; transaction and payout reports; tax invoice; financial statement; Finance API exposes sale, refund, fee, label, payout and exchange-rate concepts.[10][11] | Strong published column descriptions and API examples; no seller data needed for the initial schema. | MVP anchor · Obtain three UK report variants and one full payout cycle. |
| Etsy | Order items, orders, Etsy Payments sales/deposits and monthly statement CSVs; API ledger and payment endpoints.[12][13] | Good public descriptions; live exports still required to test UK fees, ads, VAT statements and deposit timing. | Fallback second platform |
| Vinted | Balance, pending/available funds, withdrawals, order histories and tax-reporting information are visible in the account experience.[15] | No stable, reusable first-party finance CSV sample was located in the public documentation reviewed. This is a research finding, not proof that no export exists. | Week-one gate · Test five consented UK seller datasets before promising automation. |
| Depop | Sales CSV downloadable in three-month ranges with date, buyer, item, quantity, prices, fees, shipping and taxes.[14] | Official field description is public; live file required for exact headers, refunds and payout linkage. | Assisted only until full payout matching is demonstrated. |
| Amazon | Orders, inventory, VAT and settlement reports. Flat File V2 includes settlement, dates, currency, transaction, order, marketplace, amount type/description/value, SKU and quantity fields.[16] | Authoritative schema is public and machine-friendly, but operational complexity is high; legacy settlement formats are due to be removed on 11 November 2026.[35] | Do not lead · Incumbents already serve this segment. |
| Shopify | Orders and reports plus Shopify Payments payout transaction CSV. Published columns include transaction date/type, order, payout status/date, amount, fee and net.[17] | Strong first-party documentation. Third-party gateways, Shopify bills and payout reconciliation remain separate sources. | Phase two |
| TikTok Shop | Finance statement exports with order details, statements, reports, payments, reserves and fee explanations; invoice PDFs and order-level invoice details.[18][36] | Official UK workbook structure is documented; policies and fee types change rapidly. | Discovery only until format versioning and affiliate/ads treatment are proved. |
Four deposits
eBay £118.59 and £102.27; Vinted £58.00; Etsy £38.62.
Thirteen ledger lines
Sales, refund, fees, ads, labels, stock purchase and personal cash movements.
£0.00 variance
Every payout matches the bank while retaining its business/personal split.
Six open queries
No unsupported assumption is hidden inside a polished total.
− refunds £59.50
− fees £23.04
− advertising £8.50
− platform postage £10.17
− stock paid from Vinted balance £12.00
= business marketplace cash movement £170.87
Why the bank shows £317.48
The same payouts contain £146.61 of seller-confirmed personal-disposal cash. Dweise reconciles that amount but excludes it from trading sales, subject to accountant review.
Sample accountant summary
| Category | Amount | Pack treatment |
|---|---|---|
| Net business sales | £224.58 | Gross marketplace sales less full refund; personal sales excluded. |
| Marketplace/payment fees | £23.04 | Net of the fee credit on refunded order. |
| Advertising | £8.50 | eBay promoted listings and Etsy Ads shown separately. |
| Platform postage | £10.17 | Includes a non-refunded outbound label. |
| Provisional COGS | £42.00 | Opening £100 + purchases £82 − closing £140; accounting basis remains an open query. |
| Other evidenced expenses | £33.30 | Postage, packaging and software, with receipt references. |
| Personal cash in mixed payouts | £146.61 | Reconciled, seller-confirmed, excluded from P&L; final posting depends on entity/bank context. |
Sample pack: raw synthetic files, normalised ledger, payout reconciliation, period summary, stock schedule, evidence register, exception queue, approval log and journal preview.
Open sample packThe workflow is an exception-led control process.
Automation should remove repetition. It must not turn uncertain facts into confident accounting decisions.
Written equivalent
Onboard the seller and accounts; acquire files; preserve originals and hashes; standardise; run controls; request factual evidence; obtain seller approval; route professional decisions to the accountant; issue a locked pack; append any correction as a new version; repeat on the chosen cadence.| Stage | Automate | Seller must confirm | Bookkeeper/accountant review | Control/output |
|---|---|---|---|---|
| Onboarding | Question routing, required-file checklist, period validation. | Entity, ownership, platforms, payment accounts, VAT/MTD status as known, previous adviser. | Engagement scope, identity/AML where the practice is responsible, accounting basis. | Signed scope; account register; data roles; exclusions. |
| Import | File fingerprinting, schema/version detection, encoding/date/currency parsing. | Which account and period the file represents; whether it is complete. | Only if a source is non-standard or incomplete. | Immutable original, SHA-256 hash, import manifest and rejected-row log. |
| Normalisation | Map platform fields to canonical orders, balance movements, payouts and evidence links. | Ambiguous account or transaction ownership. | Approve chart-of-accounts mapping template. | Mapping version; confidence; source file and row. |
| Classification | Rules can suggest categories and remember approved patterns. | Personal/trade, business purpose, cost and mixed-use facts. | Allowability, VAT, capital/revenue, accounting-basis and unusual-item treatment. | No auto-approval below confidence threshold; reasons retained. |
| Reconciliation | Group by payout ID; match bank references, amount and date; explain reserves and balance roll-forwards. | Identify missing banks, cash, PayPal, balance purchases and transfers. | Review material residuals and period cut-off. | Matched, timing, explained difference or unresolved difference. |
| Completeness/dedupe | Duplicate hashes and composite keys; missing date spans; sequence gaps; overlapping exports. | Attest all channels/accounts supplied. | Set materiality/tolerance appropriate to engagement. | Completeness statement and duplicate log. |
| Evidence | Create request, reminders, file link, status and expiry. | Upload receipt, stock record or written factual declaration. | Decide whether alternative evidence is sufficient. | Evidence register; never invent a receipt. |
| Approval/export | Totals, control checks, locked version, Excel/CSV/PDF output, journal preview. | Approve factual classifications and completeness. | Approve professional adjustments and import mapping. | Named approver, timestamp, version, unresolved items and export manifest. |
| Corrections | Re-run affected calculations and show differences. | Explain changed fact. | Approve revised professional treatment. | Append-only change log; prior pack remains reproducible. |
Delivery models
Self-service
Seller imports, answers every classification question and approves the pack. Dweise supports the product, not the books.
- Best regulatory boundary and gross margin
- Harder trust and onboarding problem
- Needs excellent exception explanations
Assisted
A supervised practice reviews or completes the work. Dweise supplies tools, file support and a standard pack; the practice owns professional judgement.
- Higher trust and learning
- Clear written allocation of responsibilities
- Wholesale or referral economics
Fully managed
Dweise staff obtain files, repair records and deliver completed books. That can amount to recording, reviewing, analysing, calculating or reporting financial information for others.
- Highest price and fastest learning
- Potential accountancy-service/AML supervision
- PII, quality-control and capacity risk
Sell the handover, not “tax compliance”.
The customer-visible promise is fewer unexplained payouts, fewer evidence chases and less time rebuilding marketplace activity.
Best initial users
UK sole-trader resellers and makers with 200–3,000 annual transactions, one to three channels, no mature ledger and an accountant or deadline. Prioritise mixed eBay/Vinted activity, personal/trade ambiguity and payout confusion over turnover alone.
Pain worth paying for
- The bank shows net deposits but the accountant asks for gross sales and fees.
- Returns, labels, ads and balance purchases land in different reports or periods.
- Stock costs and personal items sit outside platform reports.
- Year-end clean-up consumes evenings and generates repeated questions.
Packages
Pack Builder £129–£149/year: seller-operated imports, reconciliation, evidence queue and export. Partner Review £249–£399/period: the same pack with a supervised practice review. Monthly tidy £19–£29/month: only after recurring behaviour is proved. Prices are hypotheses, inclusive/exclusive of VAT to be confirmed.
Demonstration
- Show a £102.27 eBay deposit that contains a refund, a trading sale and a personal disposal.
- Open its payout components and original rows.
- Ask one plain-English classification question.
- Show the business/personal split, £0 variance and accountant query list.
- Export the pack without claiming the tax answer.
Proof required
- Before/after minutes spent and number of accountant questions.
- Percentage of payout value explained and lines auto-normalised.
- Named accountant acceptance of the structure, not an “endorsement”.
- Security summary, deletion policy and exactly who can see raw files.
- Real correction example proving the audit trail.
Talking points
“Your payout is not your sales.” “Personal sales remain visible in the bank match without being buried in trading income.” “Nothing uncertain is silently guessed.” “You keep the source files and a change history.”
Objections
“I have a spreadsheet.” Keep it; compare whether Dweise explains payouts and source evidence faster. “My accountant does this.” Ask whether a cleaner handover reduces clean-up time or fees. “Will this keep me compliant?” It organises records; your tax position and filing remain yours and your adviser's. “I only sell personal items.” You are probably not the target; use HMRC's checker and retain evidence.[6]
Discovery questions
- Show me the last payout you could not explain.
- Which files did your accountant ask for, and what did you send instead?
- Where do you record item cost and whether it was personal or bought for resale?
- What takes longest at year end? What does the accountant send back?
- Would you pay £149 before seeing the result? What evidence would make that safe?
- Who besides you must trust the output?
Acquisition
Accountant referrals first; then problem-led search pages, reseller communities with permission, seller newsletters, eBay/Etsy education ecosystems and deadline-triggered workshops. Never use HMRC fear, scrape seller details or imply platform affiliation.
Ideal partner
A small UK practice or bookkeeping team with 10–100 ecommerce or side-hustle clients, repeated eBay/Etsy work, emerging Vinted/Depop cases and no appetite to maintain bespoke parsers. Avoid leading with Amazon-only firms already satisfied by A2X or Link My Books.
Non-negotiable output
- Stable canonical field dictionary and versioned category map.
- Gross-to-net settlement and bank reconciliation with tolerance disclosed.
- Source file, source row, hash, mapping rule and approval for every line.
- Unresolved exceptions, assumptions and seller declarations in a separate register.
- Read-only locked version plus editable working files.
Software exports
Start with an accountant-neutral Excel/CSV workbook and a journal preview. Add tested profiles for Xero, QuickBooks, FreeAgent and Sage only after three practices validate mappings. Xero accepts bank CSVs with Date and Amount plus optional Payee, Description and Reference; QuickBooks accepts three- or four-column bank CSVs; FreeAgent uses date, amount and description; Sage 50 has a more detailed audit-trail transaction format.[22][21][23][24]
Demonstration
- Select one client and show completeness status.
- Open a payout that mixes trade, personal and refund activity.
- Trace a total to source row and evidence.
- Review the exception queue and seller approvals.
- Change one mapping, re-run and compare pack versions.
- Export a neutral workbook and the firm's preferred import preview.
Partner pack
One-page proposition; sample pack; field dictionary; control methodology; coverage/limitations matrix; security and UK GDPR summary; data-processing terms; responsibility matrix; support/SLA; correction policy; pricing; pilot protocol; case-study template; no-endorsement clause.
Commercial models
Referral: seller pays Dweise; transparent fixed referral fee or credit. Wholesale: £59–£99 per active annual pack with minimum volume. Practice licence: £99–£199/month plus £20–£50 per active client after proof. White-label: later, with clear Dweise processor role, support path and no misleading authorship. Revenue sharing must not compromise recommendation independence.
Service levels
Pilot target: acknowledgement within one working day; parser defect triage within two; correction pack within two after evidence is complete; published maintenance window; named severity levels; export version retained. These are proposed targets, not current commitments.
Objections
“We already use Link My Books/A2X.” Dweise should only proceed where mixed resale platforms or ledger-free handovers remain painful. “We cannot trust black-box categorisation.” Show rule, confidence, source and approval; permit firm templates. “This creates liability.” Keep responsibilities in the engagement and Dweise contract; the seller remains responsible for complete information and the practice for professional work.[20][37]
Discovery questions
- Walk through the last marketplace client whose records you had to rebuild.
- Which totals do you distrust, and what evidence changes your mind?
- What is your minimum acceptable reconciliation and what remains a query?
- Which import shape fits your current ledger and working papers?
- Who approves seller classifications, mappings and corrections?
- Would you put five real clients through a paid batch pilot? If not, why not?
Pilot evidence
Secure two paid firms, five client files, a repeat second period, median time saving above 50%, no material unexplained payout, and written “use again” intent. A logo or testimonial is optional; acceptance of the pack structure is the real proof.
The first cohort should maximise learning, not flatter the product.
Recruit contrasting cases with bounded volumes and obtain explicit consent to inspect, retain and use anonymised format learnings.
| Slots | Seller profile | Why included | Success evidence |
|---|---|---|---|
| 1–3 | eBay + Vinted vintage/resale sole traders; 500–2,000 annual sales. | Tests the intended cross-platform wedge, stock cost and mixed payouts. | Vinted data usable; ≥80% lines mapped; payout variance ≤0.5% before explained exceptions. |
| 4–5 | eBay-only traders, one with refunds/ads and one with foreign-currency sale. | Provides a mature-export benchmark and isolates reconciliation quality. | Full payout cycle reconciles and pack completed within 90 minutes after onboarding. |
| 6–7 | Vinted/Depop fashion resellers with some genuine personal disposals. | Tests the hardest trust and classification journey. | Seller can answer prompts without Dweise making the tax decision; accountant can follow the split. |
| 8–9 | Etsy maker using Etsy Ads; one also uses Shopify or PayPal. | Tests materials, discounts, separate payment providers and a viable fallback platform. | Deposits reconcile and off-platform costs remain clearly separated. |
| 10 | Accountant-referred “messy handover” within volume cap. | Adversarial real-world test of missing files and exception workflow. | Practice reports ≥50% less clean-up time and would reuse the format. |
Five candidate practice conversations
These are research candidates, not endorsements or confirmed partners. Verify professional status, AML supervision, client fit and conflicts before outreach. Two are intentionally incumbent users: a strong product should survive an informed “we already have this” conversation.
| Candidate | Why the conversation is valuable | Opening question |
|---|---|---|
| Rezolution | Bookkeeping practice listed for ecommerce clients ranging from small single-channel sellers to multi-channel businesses.[30] | Where do Vinted, Depop or mixed personal/trade clients fall outside your existing stack? |
| ERAA Consulting | Small, direct-owner specialist describing all-channel ecommerce work.[30] | What would a standard evidence-linked pack need before you would trust a junior to process it? |
| Archimedia Accounts | Works across Amazon, eBay, Etsy and Shopify, typically £100k–£5m turnover.[30] | Is the ledger-free small-seller segment commercially unattractive, or merely poorly served? |
| Tidy Money / Ecom Accountants UK | Advertises specialist work across Amazon FBA, Shopify, eBay, Etsy and TikTok Shop.[31] | Which settlement and client-evidence exceptions consume the most non-billable time? |
| Ecommerce Finance | Positions around marketplace accounting, payout reconciliation and HMRC online-sales letters.[38] | Could Dweise organise records without competing with your advice and filing relationship? |
Accountant buy-in depends more on inspectability than automation.
Dweise needs a control framework before a partner dashboard.
Data structure
Canonical order, balance transaction, payout, bank match, evidence, classification, query, approval and export entities; stable IDs; versioned field dictionary; no overloaded “net sales” field.
Traceability
Original preserved read-only; file hash; parser and mapping version; source row; normalised line; calculation membership; evidence link; who changed what and why.
Review controls
Role-based permissions; thresholds set by practice; seller and professional approvals separated; lock/export only after required gates; unresolved items printed prominently.
Correction process
Never overwrite an issued pack. Append the correction, show before/after totals, invalidate affected export, obtain re-approval and issue a new version.
Security and UK GDPR
Determine controller/processor roles from actual decision-making, not labels. Use a compliant data-processing contract where Dweise acts on a practice's instructions. Minimise buyer details, define retention, encrypt in transit and at rest, enforce least privilege and maintain tested backups and breach procedures.[25][26][27]
Liability boundary
Terms must identify client, purpose, scope, exclusions, seller responsibilities, practice responsibilities, data limitations, correction route and liability position. Professional bodies emphasise written engagement terms and clear allocation of responsibilities.[20][37]
Partner dashboard later
Client, period, platforms, import status, reconciliation percentage, high-value exceptions, seller approval, professional review, deadline, export version and support ticket. No decorative analytics before these operational states work.
Claims policy
Allowed: “organises”, “reconciles”, “flags”, “exports”, “ready for review”. Evidence required before: “saves 50%”, “accepted by accountants”. Prohibited without formal basis: “HMRC approved”, “tax compliant”, “guaranteed accurate”, “replaces your accountant”.
The first paid pilot needs a controlled service desk, not a platform.
Manual work is acceptable when it is documented, reproducible and kept on the correct side of the professional-services boundary.
| Capability | Deliver manually now | Automate first | Do not build yet |
|---|---|---|---|
| Onboarding | Secure form, file checklist, scope letter, platform/account register. | Conditional checklist and completeness dashboard. | Automated entity/tax advice. |
| File intake | Encrypted portal with named folders; record hashes and manifests. | Format/version detection, validation and safe rejection messages. | Direct APIs for every marketplace. |
| Normalisation | Reviewed spreadsheet transformations and small scripts. | eBay parser; second-platform parser after file gate; canonical schema tests. | Amazon depth, scraping or uncontrolled AI extraction. |
| Reconciliation | Spreadsheet payout roll-forward reviewed by a second person. | Payout grouping, bank match suggestions, tolerance and unmatched queue. | Automated bank access until consent/security and economics are proved. |
| Classification | Seller-operated questionnaire; practice makes professional decisions. | Rule suggestions with confidence and explicit approval. | Autonomous personal/trade, VAT or capital-allowance decisions. |
| Pack/export | Controlled workbook/PDF generated from templates; manual QA checklist. | Reproducible report generation, version lock and export manifest. | Direct ledger posting or HMRC filing. |
| Support | Ticket log, decision register, published pilot response targets. | Diagnostics bundle without raw buyer data. | Open-ended WhatsApp/email handling of financial files. |
Roles and minimum controls
Founder / product operator
Own scope, tooling, file support, issue log, source preservation and metrics. Must not make client accounting decisions unless appropriately supervised and engaged.
Qualified bookkeeping/accounting partner
Designs category map, reviews sample packs, sets tolerances and owns professional decisions for partner-led cases.
Data engineer or analyst
Builds deterministic parsers, tests schemas and reconciliations, handles currency/date quirks and maintains version fixtures.
Privacy/security adviser
Confirms roles and lawful basis, processor terms, retention, secure transfer, incident response and whether a DPIA is appropriate.
Written equivalent
Month one resolves the legal/data model and gathers real files. Month two sells and delivers five bounded pilots while measuring the work. Month three tests repeatability, practice batches and exports before a funding decision.Prove the inputs and boundary
- 15 seller interviews; 8 practice interviews.
- 10 full, consented file sets; five Vinted sets.
- Written AML/accountancy, liability and data-protection review.
- Accountant field dictionary workshop.
- Seller-operated prototype using the synthetic pack.
Run the bounded paid pilot
- Five paid seller packs or one equivalent practice batch.
- Independent second-person arithmetic/lineage check.
- Track active minutes, follow-ups, exception rate and correction rate.
- Accountant compares output with current working papers.
- No direct posting; journal preview only.
Test reuse and distribution
- Second period for three sellers.
- Two paid practice pilots covering at least five client files.
- Three practices accept one stable core pack.
- Measure support cost, acquisition cost and renewal intent.
- Publish only evidenced claims.
Go, revise or stop
Build the narrow MVP
- ≥5 sellers pay £149, or two practices pay ≥£500 for batches.
- ≥80% line automation and ≥95% payout-value reconciliation.
- Median human processing ≤90 minutes per bounded annual pack after first setup.
- Two practices commit repeat clients; ≥4/5 usefulness.
- Boundary, insurance and data controls are resolved.
Change segment or workflow
- Payment exists but only below £99.
- One platform causes most manual work.
- Accountants value the pack but require the same structural changes.
- Time saving is 25–50% or follow-up is 15–30% of lines.
- Seller-direct acquisition is weak but practice batches convert.
Do not fund the platform
- Fewer than three paid commitments after 40 qualified conversations.
- <60% lines normalise or median manual work exceeds three hours.
- No accountant will reuse the pack.
- Legal/AML/PII requirements make the chosen model uneconomic.
- An incumbent closes the resale-platform and ledger-free gap before Dweise earns distribution.
Unit economics should be measured as a service before they are forecast as software.
The existing Dweise projection of £99–£149 annual licences and practice distribution is a reasonable test range, not market evidence.
| Metric | Target | Why it matters | Measurement |
|---|---|---|---|
| Paid conversion | ≥20% of qualified seller conversations at £149, or ≥25% of qualified practices into a paid batch. | Separates anxiety and compliments from purchasing intent. | Price shown before demo; deposits recorded; no founder-friend freebies in denominator. |
| Delivery labour | ≤90 minutes median per bounded annual seller pack after first setup. | At £149, prolonged manual clean-up destroys contribution margin. | Time log by import, exception, evidence, QA and support. |
| Contribution margin | ≥60% at target price after support, processing and partner fees. | Leaves room for development, sales and compliance overhead. | Actual labour at market cost, payment fees and software costs; founder time not treated as free. |
| Accountant value | ≥50% preparation-time reduction; ≥4/5 usefulness; repeat client committed. | The practice channel is the strongest scalable trust route. | Before/after timed task and structured acceptance form. |
| Retention signal | ≥60% of pilot sellers choose a second period or annual renewal at disclosed price. | One-off tax anxiety may create sales without a durable business. | Ask for payment, not stated intent, at pack delivery. |
| Correction quality | 100% issued packs reproducible; no material silent changes. | Trust and liability depend on a durable audit trail. | Regenerate from manifest and compare hashes/totals. |
How this plan sits against the companion reports
Three Dweise documents now describe the same business from different angles. They mostly agree, but two numbers do not, and the disagreement matters more than the agreement.
| Question | This operating plan | Business projections report | Reconciliation |
|---|---|---|---|
| Decision horizon | A 90-day validation with three hard gates. | A 12-month plan with a scale decision at months 9 to 12. | Compatible if the 90 days sit inside months 1 to 3. Gate 3 here should be the same event as the projections report's month-3 checkpoint, not a separate one. |
| Seller price | £129 to £149 a year self-service; £249 to £399 partner-led. | £99 to £149 a year, with the model implying about £200 of revenue per paying record. | Inconsistent. The projections assume an average well above the licence price, only reachable if partner and pilot revenue carry a large share. Fix the projection, not the price. |
| Practice pricing | Wholesale £59 to £99 per pack; practice licence £99 to £199 a month plus £20 to £50 per active client. | Practice base plus a per-client charge, but modelled at a much lower blended value. | This plan's practice pricing is the stronger of the two and should replace the projection's assumption. |
| First offer | Seller-controlled software, or practice-supervised processing. Fully managed only after written advice. | Same conclusion, reached through the AML boundary. | Agreed. This is the most settled decision across the whole research set. |
What could become defensible
Weak moat
CSV parsers, generic category suggestions, dashboards and one-click PDFs are straightforward to copy. Platform coverage can disappear when an incumbent adds one connector.
Stronger assets
Versioned UK marketplace schemas; a proven exception taxonomy; accountant-approved working-paper structures; repeat partner distribution; correction lineage; and measurable historical reliability across changing formats.
Proceed, but fund learning, not breadth.
Dweise can deliver a valuable service if it makes fragmented records explainable without pretending that reconciliation is tax advice.
The opportunity is commercially credible as a seller-controlled, accountant-enabled records layer for smaller multi-marketplace traders. It is not yet defensible as a broad platform, and it may never be attractive at low-ticket pricing if human evidence chasing remains high.
The claims that still need real-world evidence.
These are not conclusions and should not appear as marketing claims before the named test passes.
Target sellers will pay £129–£149 annually.
Test with paid deposits before a complete product exists.
Vinted records can support reliable gross-to-withdrawal reconciliation.
Test five full cycles and decide against the pre-set Etsy fallback.
At least 80% of lines can be normalised without manual restructuring.
Measure across distinct report versions, not repeated files from one account.
Accountants prefer the proposed core pack and can import it efficiently.
Require three practices to use the same version on real client files.
Seller factual confirmation keeps the initial model outside regulated bookkeeping.
Obtain written professional advice based on the exact screens, support scripts and contracts.
Accountant distribution lowers acquisition cost enough to offset a slower sales cycle.
Track qualified lead source, days to decision, partner margin and client activation.
Primary guidance first; commercial recommendations clearly labelled.
Sources were checked on 31 July 2026. Marketplace formats, software imports, thresholds and guidance can change; re-check before pilot launch and before every material product claim.