MVP brief
Ledgerlight
1. MVP summary
What you are building
Ledgerlight is a web app for small accountancy practices that checks clients’ purchase invoices before each VAT return. It connects to the client’s Xero account, reads every supplier invoice attached to the bills in the VAT period, compares what the invoice says with what was posted, checks the supplier’s VAT number with HMRC, and lists anything that looks wrong for a bookkeeper or accountant to review. It changes nothing in the accounts itself.
What the MVP must prove
That the checks are accurate and useful enough for a practice to pay for them on every VAT client. In numbers: the system reads the VAT amount and the supplier’s VAT number correctly on at least 97% of legible invoices in our test set, reviewers judge at least half of its flags worth acting on, and at least three of our five pilot practices sign up at £2 per VAT client a month.
Key numbers
Pre-seed round £250,000. MVP ready for paying practices in June 2027, built by the founders with a freelance developer and designer, with £50,375 of one-off costs. 26 paying practices by February 2028, paying an average of £140 a month each. Revenue in the first 12 months £16,800, gross margin 74%, and a planned loss of £113,040 before tax. Running costs £10,450 a month. £96,585 left in February 2028, about 12 more months at that month’s spending.
What you need
£250,000 from angel investors under the Seed Enterprise Investment Scheme, for 20% of the company. We would also value introductions to small accountancy practices outside Greater Manchester for a second wave of pilots, and to an adviser with experience of selling software to accountants.
2. Problem and users
The problem
Every quarter, small practices prepare VAT returns for their clients from bookkeeping that is often done by the client or a junior. Mistakes in purchase invoices are common and slow to find: VAT claimed on an invoice from a supplier that is not VAT registered, a VAT figure keyed in wrongly, the same invoice entered twice, or a bill posted with no invoice attached. The only reliable check is for someone senior to open the invoices, which few practices have time to do for every client. In our interviews, reviewers spent 30 to 60 minutes on each VAT return and still checked only a sample of invoices.
Target users and early adopters
Accountancy and bookkeeping practices with one to 15 staff, each preparing VAT returns for 30 to 200 clients, mostly on Xero. The buyer is the practice owner; the daily user is the bookkeeper or manager who reviews returns.
Our first 50 practices come from:
- our five pilot practices in Greater Manchester and the practices they refer;
- Sana’s network from 12 years in practice;
- accountancy events and online communities for small practices;
- Xero’s app store, once our app is approved.
How they cope today
- A senior person spot-checks a sample of invoices against the ledger before filing.
- Checklists and spreadsheets built by the practice.
- Data capture tools read invoices when they are first entered, but what is posted can still differ from the document, and many clients enter bills by hand.
- Ledger review tools flag unusual coding and transactions, without re-reading every invoice.
- Some practices trust the client’s bookkeeping and correct errors when the client or HMRC finds them.
Evidence so far
- 22 interviews with practice owners and reviewers in 2026. 19 said purchase invoice errors were their most common VAT return correction, and 15 said they could not check every invoice for every client.
- A pilot of our prototype with five practices since September 2026, under a pilot data processing agreement: 180 VAT returns checked and 1,240 items flagged. Reviewers judged 61% of flags worth acting on, and found VAT claimed without a valid VAT number in 23 returns.
- Three practices have signed letters of intent to pay £2 per VAT client a month from launch.
- We know the prototype’s weak points: photos of crumpled receipts, handwritten invoices, and long invoices with several VAT rates.
3. Value proposition
The promise
Every purchase invoice checked before the VAT return goes in, without opening every one yourself.
Why you, not the alternatives
- We read the invoice itself, not just the ledger: each check compares the document with what was posted.
- Every supplier’s VAT number is checked with HMRC, through HMRC’s Check a UK VAT number API.
- Built for review: each flag shows the invoice with the relevant figures highlighted and the reason, and the reviewer accepts or dismisses it in one click, with an audit trail.
- Works whatever the client used to enter the bill: a data capture tool, Hubdoc or typing it in.
- Data protection designed in: the practice stays the controller and we act as its processor under a data processing agreement. Documents are processed in UK data centres, deleted from our systems 90 days after review, and never used to train AI models.
- Built by a chartered certified accountant and an engineer who has run document checking systems for a lender.
Over time our advantage grows from what reviewers teach us: their accept and dismiss decisions tune our rules and thresholds, without using client documents to train models.
Business model
Subscription
Price and how you set it
£2 per VAT-registered client a month, with a minimum of £79 a month, billed monthly by Direct Debit, excluding VAT. A pilot practice has 70 VAT clients on average, so would pay about £140 a month. Each practice costs us about £37 a month to serve: AI document reading at about 2p a document on our test runs (around 1,400 documents a month for a typical practice), plus hosting and Direct Debit fees.
For comparison: AutoEntry, a data capture tool, starts at £15.45 a month (excluding VAT) for 50 credits, with an invoice using one or two credits; Hubdoc is included in Xero’s Ignite, Grow, Comprehensive and Ultimate plans; Dext prices its practice plans per client, with a minimum of 10 clients.
The value test is time. If Ledgerlight saves a reviewer 15 minutes on each quarterly return, that is worth more than the £6 a quarter it costs per client at most practices’ staff costs. We will test £2 against £3 per client with the second wave of practices.
4. Competitors and alternatives
| Alternative | What it does | Price | Strengths | Gaps |
|---|---|---|---|---|
| Dext | Captures receipts and invoices for practices and their clients and publishes them to accounting software; add-ons include data health checks | Practice plans priced per client, with a minimum of 10 clients, quoted through Dext’s plan builder | Widely used by UK practices, with mature data capture and integrations | Focused on capturing documents as they arrive; its published review checks look at coding, contacts and transactions in the ledger |
| AutoEntry | Reads invoices, receipts and bank statements and posts them to accounting software | From £15.45 a month (excluding VAT) for 50 credits; an invoice uses 1 credit, or 2 with line items | Simple credit pricing; works with the main accounting packages | A data entry tool: it reads documents as they are entered, rather than reviewing the quarter’s postings before a return |
| Hubdoc (Xero) | Document capture included with Xero subscriptions | Included in Xero Ignite, Grow, Comprehensive and Ultimate (£18 to £70 a month, excluding VAT) | No extra cost for Xero users and already used by many clients | Captures and stores documents; does not review the quarter’s postings before a return |
| Senior review by hand | A manager opens a sample of invoices and checks them against the ledger | Staff time: 30 to 60 minutes a return in our interviews | Judgement and knowledge of the client | Slow, only ever a sample, and the first thing dropped in a busy quarter |
| Practice checklists and spreadsheets | Home-made review lists and reconciliations | Free | Familiar and tailored to the practice | Depend on someone remembering to use them; nobody reads the documents |
Your advantage
Capture tools are strong and we will not compete with them. We sit after them, as a second pair of eyes before the return. Our edge is reading each invoice against what was posted, combined with HMRC’s VAT number check and a review screen designed by someone who has done the job. Larger companies could add this. We will stay ahead by focusing on one painful moment in the practice’s quarter, by publishing our measured accuracy, and by learning from every reviewer decision which flags are worth raising.
5. MVP scope
| Feature | Priority (Must, Should, Could, Won’t) | User story: as a … I want … so that … |
|---|---|---|
| Connect a practice and its clients’ Xero accounts | Must | As a practice manager, I want to connect each client’s Xero file with read-only access so that Ledgerlight can see bills and their invoices without changing anything. |
| Read each attached invoice | Must | As a reviewer, I want the supplier, VAT number, date, net, VAT and total read from each invoice so that they can be compared with what was posted. |
| Compare each invoice with its posting | Must | As a reviewer, I want differences in VAT, totals, dates and supplier flagged so that I see the errors without opening every bill. |
| Supplier VAT number check with HMRC | Must | As a reviewer, I want VAT claimed on invoices without a valid VAT number flagged so that the client does not reclaim VAT it is not entitled to. |
| Duplicate and missing invoice checks | Must | As a reviewer, I want bills entered twice, or posted with no invoice attached, listed so that I can query them with the client. |
| Review queue with accept, dismiss and notes | Must | As a reviewer, I want each flag shown beside the invoice with the reason so that I can decide quickly and leave a record of my decision. |
| Couldn’t read: check by hand | Must | As a reviewer, I want any invoice the system cannot read confidently marked for checking by hand so that I am never shown a guessed figure. |
| Summary report for each VAT return | Should | As a practice owner, I want a one-page summary of the checks and decisions for each return so that I have evidence of review on file. |
| Construction reverse charge prompts | Should | As a reviewer, I want invoices that may fall under the domestic reverse charge for construction flagged so that the VAT is treated correctly. |
| QuickBooks connection | Could | As a practice with clients on QuickBooks, I want the same checks so that I can use one tool for all my clients. |
| Corrections written back to Xero automatically | Won’t | As a reviewer, I might want fixes posted for me so that I save time; we will not do this, because a person should make every change to a client’s accounts. |
| VAT treatment decided by AI | Won’t | As a reviewer, I might want the system to decide the VAT treatment so that I can skip the judgement; it will only point to what needs a qualified person. |
Deliberately left out
- Writing anything back to clients’ accounts: the MVP has read-only access.
- QuickBooks, Sage and other packages, until the Xero version is proven.
- Sales invoices and bank reconciliation checks.
- Bookkeeping or data entry: we do not replace capture tools.
- A client-facing app: clients never log in.
- Tax advice: flags point to what a qualified person should look at, and never tell anyone what to file.
- Training AI models on clients’ documents: we do not do it, and our contracts with AI providers forbid it.
Riskiest assumption
That AI can read real small business invoices accurately enough, with few enough false alarms, that reviewers trust it and keep using it. Systems that read documents make mistakes, and a language model can produce a figure that looks right but is not on the page. If reviewers find even a few confident errors, they will stop trusting every flag. We test this first, on 3,000 hand-checked invoices, and design the system so that an uncertain reading becomes a request to check by hand rather than a guess.
6. Success metrics
North-star metric
VAT returns checked through Ledgerlight each month where the reviewer acted on at least one flag. It shows practices using the product on real returns and finding something worth fixing.
Targets for the first 90 days
For the first 90 days after paid launch (June to August 2027):
- Paying practices: 7 by the end of August, including at least 3 of our 5 pilot practices.
- Usage: 80% of each paying practice’s VAT clients connected within 30 days of signing.
- Usefulness: at least 55% of flags marked as worth acting on.
- Accuracy: VAT amount and VAT number read correctly on at least 97% of legible invoices in the monthly hand-checked sample, and no figure shown that is not on the invoice.
- Retention: no paying practice cancels in its first quarter.
- Pipeline: 40 demo requests from webinars and events by the end of September.
What happens next
- Carry on if at least 3 pilot practices pay, at least 50% of flags are useful and accuracy meets the 97% target.
- Change direction if accuracy is good but practices will not pay per client: test a price per return, or sell through a larger software partner.
- Pause sales and fix the product if accuracy falls below 95% in any monthly sample, or if any reviewer reports a figure that was not on the invoice.
- Stop and agree next steps with investors if fewer than 10 practices are paying by December 2027.
The founders review these numbers every month with our lead investor.
7. Validation plan
| Test | What it tests | Pass if | By |
|---|---|---|---|
| Accuracy test on 3,000 invoices checked by hand | The system reads the key figures accurately enough to be trusted | VAT amount and supplier VAT number correct on 97% of legible invoices, and every unreadable field marked for checking by hand | May 2027 |
| Seeded errors in a test ledger | The checks catch real errors, not just obvious ones | At least 90% of 200 planted errors flagged | May 2027 |
| Prototype pilot with five practices | Reviewers find the flags worth acting on | At least 50% of flags judged useful (result: 61% across 180 returns, passed) | November 2026 |
| Data protection review with the pilot practices | Practices will trust a young company with client data when the safeguards are clear | All five pilot practices sign the data processing agreement after reading our impact assessment and list of AI and hosting providers | April 2027 |
| Paid conversion at £2 per VAT client | Practices will pay per client, not just use it free | 3 of 5 pilot practices pay from launch | June 2027 |
| Two webinars and one accountancy event | We can reach practices beyond our network at a sensible cost | 40 demo requests and 10 trials | September 2027 |
How you will find your first users
- The five pilot practices, three of which have signed letters of intent.
- Referrals: each pilot practice has already introduced at least one other practice.
- Sana’s network of practice owners and managers in the North West.
- Webinars on common VAT return errors, run with the pilot practices.
- Accountancy events and online communities for small practices.
- Xero’s app store, once our app is approved.
8. Founders and team
Founders
Sana Iqbal, CEO. A chartered certified accountant with 12 years in practice, the last five as a manager at a nine-person practice in Stockport, where she reviewed VAT returns for around 140 clients a quarter. She owns the product’s checks, the relationship with practices and sales. Full time since September 2026.
Rob Mitchell, CTO. Ten years as a software engineer, the last four at a Manchester lender building systems that read and check documents in loan applications, handling sensitive financial data under audit. He built the prototype. Full time from the close of the round.
The founders own 100% of Ledgerlight Ltd, in equal shares, and have put in £15,000 of their own money.
Gaps and how you will fill them
- Sales: neither founder has sold software, although Sana has won and kept practice clients for years. We will add an adviser with experience of selling to accountants, and hire a sales lead after a seed round.
- Data protection and AI governance: an external adviser reviews our impact assessment, supplier contracts and accuracy testing before launch.
- Design: a freelance product designer for 15 days on the review screens.
- Extra engineering: a freelance developer for 40 days on the Xero connection, security and automated testing.
Advisers and partners
- A partner at one of our pilot practices, who chairs a monthly panel of pilot users.
- A data protection consultant who has worked with accounting software companies.
- We are looking for an angel investor with experience of selling software to accountancy practices to join as an adviser.
9. Budget and runway
Key figures
| Figure | Value |
|---|---|
| Build and launch costs | £50,375 |
| Funding | £260,000 |
| Running costs | £10,450 a month |
| Average monthly burn | £9,326 |
| Runway | About 24 months at the month-12 rate of spend |
| Lowest cash balance | £96,585 in Feb 2028 |
| Cash at month 12 | £96,585 |
| Year-one revenue | £16,800 |
| Break-even | Not reached in year one |
Build and launch costs
| Item | Cost |
|---|---|
| Freelance developer: Xero connection, security and automated testing (40 days at £500) | £20,000 |
| Freelance product designer: review screens and onboarding (15 days at £425) | £6,375 |
| Test set: 3,000 invoices checked by hand by freelance bookkeepers | £3,000 |
| Cyber Essentials certification and an independent penetration test | £5,000 |
| Legal fees for the round: subscription agreement, articles and shareholders’ agreement | £7,000 |
| Customer terms, data processing agreement and privacy notice (technology solicitor) | £3,000 |
| Independent review of the data protection impact assessment and AI testing | £2,000 |
| SEIS advance assurance and accountant’s help with the round | £1,500 |
| Brand and website | £2,500 |
| Total | £50,375 |
Funding
| Source | Type | Amount |
|---|---|---|
| Pre-seed round: SEIS shares | Investment (equity) | £250,000 |
| Founders’ own money still in the company account | Your own money | £10,000 |
| Total | £260,000 |
Early revenue
| Plan or product | Price | Cost per sale | Sales in year one | Revenue |
|---|---|---|---|---|
| Subscriptions (paying practices, average £140 a month) | £140.00 | £37.00 | 120 | £16,800 |
| Total | £16,800 |
Running costs
| Cost | Per month | Per year |
|---|---|---|
| Founder salaries (two at £2,750 a month, with 3% pension) | £5,650 | £67,800 |
| Cloud hosting (UK region), AI services and monitoring: fixed part | £600 | £7,200 |
| Software tools (code hosting, design, CRM, help desk, accounts) | £350 | £4,200 |
| Marketing: webinars, accountancy events, content and LinkedIn adverts | £1,800 | £21,600 |
| Travel to practices and events | £300 | £3,600 |
| Accounting, payroll and legal | £400 | £4,800 |
| Insurance (professional indemnity and cyber) | £250 | £3,000 |
| Coworking (two desks in Manchester) | £400 | £4,800 |
| Security upkeep: monitoring, certification renewal and a testing reserve | £300 | £3,600 |
| Accuracy checks: hand-checking a sample of invoices each month | £400 | £4,800 |
| Total | £10,450 | £125,400 |
12-month cash flow
| Item | Mar 2027 | Apr 2027 | May 2027 | Jun 2027 | Jul 2027 | Aug 2027 | Sept 2027 | Oct 2027 | Nov 2027 | Dec 2027 | Jan 2028 | Feb 2028 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Opening balance | £0 | £199,175 | £188,725 | £178,275 | £168,134 | £158,199 | £148,470 | £139,050 | £129,939 | £121,137 | £112,541 | £104,357 |
| Receipts from customers | £0 | £0 | £0 | £420 | £700 | £980 | £1,400 | £1,820 | £2,240 | £2,520 | £3,080 | £3,640 |
| Funding received | £260,000 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 |
| Total cash in | £260,000 | £0 | £0 | £420 | £700 | £980 | £1,400 | £1,820 | £2,240 | £2,520 | £3,080 | £3,640 |
| Build and launch costs | £50,375 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 |
| Direct costs | £0 | £0 | £0 | £111 | £185 | £259 | £370 | £481 | £592 | £666 | £814 | £962 |
| Running costs | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 | £10,450 |
| Loan repayments | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 |
| Founder’s drawings | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 | £0 |
| Total cash out | £60,825 | £10,450 | £10,450 | £10,561 | £10,635 | £10,709 | £10,820 | £10,931 | £11,042 | £11,116 | £11,264 | £11,412 |
| Net cash flow | £199,175 | −£10,450 | −£10,450 | −£10,141 | −£9,935 | −£9,729 | −£9,420 | −£9,111 | −£8,802 | −£8,596 | −£8,184 | −£7,772 |
| Closing balance | £199,175 | £188,725 | £178,275 | £168,134 | £158,199 | £148,470 | £139,050 | £129,939 | £121,137 | £112,541 | £104,357 | £96,585 |
Prices exclude VAT. Costs exclude VAT if you are VAT registered (you reclaim it) and include it if you are not. Build and launch costs and funding fall in the first month and any loan repayments start in month two. Runway counts whole months before cash goes below zero.
Assumptions behind the numbers
- All figures exclude VAT. Ledgerlight is registered for VAT, as its customers are VAT-registered practices that reclaim it. The forecast runs from March 2027, when the round closes, to February 2028.
- Paying practices: 3 in June 2027, when paid plans start, rising to 26 by February 2028. No practice cancels in year one, although we expect some to.
- Price: £2 per VAT client a month, minimum £79; an average of £140 a month per practice, based on the pilot practices’ 70 VAT clients each.
- Cost to serve: about £37 a month per practice, mostly AI document reading at about 2p a document on our test runs, plus hosting and Direct Debit fees (Stripe charges 1%, capped at £4, for Bacs Direct Debit).
- One-off costs of £50,375 in month one: a freelance developer for 40 days at £500 (£20,000; IT Jobs Watch gives a £500 median contract day rate for React developers outside London, November 2026), a freelance designer for 15 days at £425 (£6,375, the median for UX designers outside London), hand-checking 3,000 invoices for the test set (£3,000), Cyber Essentials and a penetration test (£5,000), legal fees (£10,000), an independent review of the impact assessment (£2,000), SEIS advance assurance and accounting (£1,500) and brand and website (£2,500).
- Running costs of £10,450 a month: founder salaries of £2,750 a month each with a 3% pension (employer National Insurance is covered by the Employment Allowance, as both founders are on the payroll), marketing £1,800, coworking £400, and the rest for cloud and AI services, tools, travel, insurance, accounting, security upkeep and monthly accuracy checks.
- No R&D tax relief, grants or loans are included.
10. Funding
How much you need
£250,000
What it will pay for
- One-off costs of £50,375 in month one: freelance development and design £26,375, the hand-checked test set £3,000, security certification and testing £5,000, legal fees for the round and our customer contracts £10,000, impact assessment review £2,000, SEIS advance assurance and accounting £1,500, brand and website £2,500.
- Running costs of £10,450 a month, of which founder salaries are £5,650 and marketing £1,800.
In the first 12 months the business uses £163,415 (the one-off costs and a planned loss of £113,040), leaving £96,585 in February 2028: about 12 more months at that month’s spending.
What you are putting in
Sana and Rob have put in £15,000 of their own money. £5,000 paid for prototype hosting, AI testing and legal advice for the pilot, and £10,000 is still in the company account and included in this forecast. Rob built the prototype unpaid over nine months alongside his job, and Sana left her practice role in September 2026. They will own 80% of the company after the round.
Terms and the next milestone
We are offering 20% of Ledgerlight Ltd for £250,000, a pre-money valuation of £1 million, in ordinary shares. This is the most a company can raise under SEIS. We have applied to HMRC for advance assurance: the company has gross assets under £350,000 and fewer than 25 employees, and started trading less than three years ago. Qualifying investors can claim income tax relief of 50% of their investment, on up to £200,000 a year, if they keep the shares for at least three years. We will share this plan only with investors who qualify as high net worth or sophisticated under the financial promotion rules.
The next milestone is 26 paying practices, measured accuracy of at least 97% and an approved Xero app by February 2028. That supports a seed round under the Enterprise Investment Scheme to hire in sales and engineering. As with any early-stage company, investors could lose all the money they invest.
11. Roadmap
| Milestone | Target date | Who |
|---|---|---|
| Apply to HMRC for SEIS advance assurance | December 2026 | Sana and accountant |
| Close the pre-seed round | March 2027 | Sana |
| Impact assessment complete; data processing agreements signed by the pilot practices | April 2027 | Rob and data protection adviser |
| Accuracy and seeded-error tests passed | May 2027 | Rob |
| Cyber Essentials certification and penetration test complete | May 2027 | Rob |
| Paid launch with the first three practices | June 2027 | Sana |
| Apply for Xero app approval | July 2027 | Rob |
| 90-day review: 7 paying practices | August 2027 | Sana and Rob |
| Webinars and first accountancy event | September 2027 | Sana |
| Construction reverse charge prompts and return summary report released | October 2027 | Rob |
| 26 paying practices; start seed round conversations | February 2028 | Sana and Rob |