For the files no reporting tool will ever read
Your ledger connects in four minutes. It’s the other five files that eat the afternoon: a POS export, a payroll register, a bank CSV, a portal download that only comes as XLSX. A model reads all of them and works out what’s in them once. After that it is compiled code that rebuilds the same numbers every month, with the write-up already drafted.
Built for bookkeepers, fractional CFOs and reporting agencies
Early access. The free tools are live today. The product itself is still being built, and the exports it handles first are the ones people tell us they still rebuild by hand every month.
Which one are you?
All five end up in the same place
All five have the same cause.
Every reporting tool works the same way: one integration per source, and each one has to earn back what it cost to build. Shopify and QuickBooks are worth it. A regional POS with four hundred sites never will be. So those clients stay manual — not because you’ve set something up wrong, but because nobody is ever going to build that connector.
With Databoat
Every source understood once and frozen, so this month’s numbers compare to last month’s and the report rebuilds itself.
Without
A workbook you inherited, re-pasted every month, that breaks loudly or — worse — quietly returns the wrong number.
ILLUSTRATIVE — SOURCES BY ADOPTION. ONLY THE HEAD OF THE CURVE EVER GETS AN INTEGRATION.
So Databoat starts from the file instead.
You spend a few minutes the first month agreeing what the files mean. After that, it’s one step.
01
The same files you already download. CSV or Excel, however messy — headers on row seven, totals at the bottom, four different date formats. One workspace per client.
02
A model says what it thinks each column is, in plain words, and puts up the cuts worth keeping. You correct anything it read wrong. This is the only stage the model reads your files at all — after that the definition is compiled and never guessed at again.
03
Headline numbers with movement against last month, the breakdowns you chose, and a shareable link or PDF. The commentary is drafted by a model over numbers it had no way to change — you edit it, and it goes out in your voice.
04
Drop the new files in and the same report rebuilds itself, computed by code rather than read again, so March genuinely compares to February. If an export changed shape, it tells you before anything goes out.
Files in
Any export
You
Understood
Once, with you
Model reads
Computed
Every month
Code, never the model
Written up
A draft you sign
Model drafts
THE MODEL READS AT SETUP AND DRAFTS AT THE END. THE ARITHMETIC IN BETWEEN IS COMPILED CODE, AND NEVER THE MODEL.
And this is what lands at the end of it.
Headline numbers with movement against last month, the breakdowns you picked, and the detail underneath. The commentary at the top is written from those numbers — that’s the second twenty minutes gone. It’s a draft, never a send: you edit it, you sign it, and every sentence points back at the number behind it.
Monthly pack · Client 004
March
Commentary · drafted
Not sent
Revenue reached $84,210 in March, up 6.4% on February and the strongest month of the twelve. Margin slipped 1.2 points to 41.8% as cost of goods grew faster than income, while payroll held close to flat. Cash closed at $51,380.
Every figure traces to its rows
Edit & sign →Revenue
$84,210
+6.4% vs Feb
Gross margin
41.8%
−1.2 pts vs Feb
Payroll
$22,940
+2.1% vs Feb
Cash at close
$51,380
+11.9% vs Feb
Revenue · 12 months
Same shape every month
| Line | March | February |
|---|---|---|
| Product income | 52,910 | 49,120 |
| Service income | 31,300 | 30,050 |
| Cost of goods | 49,020 | 45,880 |
| Operating expenses | 12,470 | 12,610 |
Heads up. The payroll export gained a column this month (“Employer NI”). It is not in your report yet — include it?
A MONTHLY CLIENT PACK BUILT FROM THREE EXPORTS
Open the four sample packs — hospitality, property, multi-entity and multi-channel retail, each in the practice’s own brand and each reconciled to the penny.
Once it understands your files, it puts up the ways they’re worth seeing, best first. Picking from a list is much easier than describing what you want, which is why nobody needs to know what a pivot table is. Forty charts wouldn’t be an answer, it’d be homework.
A SAMPLE OF THE VIEWS PROPOSED FROM A MONTHLY PACK
It goes out under your name, so it has to hold.
A spreadsheet template points at cells. Databoat learns that this column is the net amount, and that row is a total rather than a transaction — so when the file shifts, it still knows what it’s looking at.
The definition is frozen after the first month. That’s what makes a trend a trend, instead of two different answers to the same question — and it’s the one thing a chat window can’t give you.
The dangerous failure isn’t a template that breaks. It’s one that keeps working and quietly returns the wrong number. A renamed column, a new field, a summary row swept into a total. Databoat spots it, says so in plain English, and waits for you before anything goes out.
A mapping you can’t inspect is a guess with your signature on it. So every formula is written out in words before the first run, every figure traces back to its rows, and every sentence of the commentary points at the number behind it. Checking a report should take a minute, not a rebuild.
None of this means ripping out what you already pay for.
They connect your ledger and leave you to conform on everything else. That reformatting is the step that eats the afternoon, and it’s the step Databoat takes off you. You don’t have to take our word for it — it’s in their own documentation.
“Users cannot simply upload raw accounting system exports; they must restructure their data to match Fathom’s predefined template format.”
Joiin has you map the chart of accounts by hand. Spotlight and Reach anchor on the ledger and treat a spreadsheet as the fallback you conform to. Every serious product in the category connects the ledger and leaves the rest with you. We have written up Fathom, Spotlight and Joiin properly — including when you should stay where you are.
Databoat reads the awkward exports, works out what’s in them, and hands over clean mapped data in the shape your tool expects. You keep your templates, your branding, your process. The manual step in front of them just goes away.
For the clients with no accounting system worth connecting to — a regional POS, a wholesale portal, an industry ERP, a spreadsheet somebody maintains by hand — Databoat produces the finished report itself. Same work underneath either way.
And the files never have to leave your machine.
Databoat works out a process, not just an answer, so it can run where the data already is — in your browser. Only the finished numbers sync, which is what lets next month compare to this one. If you handle other people’s financials, that’s the difference between a tool you can just start using and one you have to ask permission for.
WHAT MOVES, AND WHAT DOES NOT
Which leaves what it costs.
One workspace is one client, entity or business. No tiers to compare, nothing held back, nothing to outgrow — you pay for the clients you’ve got, and the price moves when your practice does.
Only the clients you actively report on count here. If you lose one, archive it — billing stops that month, nothing is deleted, and the history is waiting if they come back.
Your price
$300 / month
$39 first workspace + $29 × 9
$30.00 per client
At 10 clients, against published list pricesWhere most bookkeeping practices sit
Not the cheapest, deliberately. The column that matters is the middle one. For roughly the same money, the difference is whether you still open Excel first — and pricing under them would suggest this was the same product.
The free tools below are genuinely free and always will be. The product isn’t, because a report going out to someone’s client isn’t something to try on a whim. Ask for a trial and you’ll get a real one, on your own files.
You can judge the reading engine today.
Each one does a single job on a single export, and runs entirely in your browser. It’s the same reading engine the product uses, pointed at one file type — so it’s the fairest way to find out whether Databoat would make sense of yours.
Drop the whole export zip from Search Console and get the write-up first: ranked findings with the evidence under each, then the trend, every query plotted against the click-through it should be earning, and tables that sort and filter.
OPEN →Paste a column and the total it should hit. It tells you whether the difference is a transposition, a missing row, a sign the wrong way round, or a pair — and which rows to look at.
OPEN →Validate an IBAN’s length and check digits in your browser before a payment goes out, and read the sort code and account number back out of a UK one.
OPEN →Check a VAT registration number against both of HMRC’s checksum schemes without sending it anywhere. Catches the mis-keyed digit on a supplier record before it reaches a return.
OPEN →The things people ask before they send files.
Early access
Early access opens practice by practice. Tell us the report you rebuild every month and the files behind it. If it’s close to something we already handle, we’ll ask for a sample and set you up on your real data. If not, we’ll say so plainly rather than park you on a list.