Variation Evidence Gap Checklist
The eight documents that decide whether a variation survives to the final account, and where they usually go missing.
Get it in your diagnostic →Practical, honest tools for the months that decide your margin: variations, payments, notices, evidence and the final account. No sign-up wall. Not legal or QS advice.
Eight questions on how well your live jobs would stand up if challenged. Tick what's true. You'll get a score and see where the gaps sit. Nothing is sent anywhere; it runs in your browser.
Gaps showing up? A Commercial Control Diagnostic reads one of your contracts and turns it into a register where every item is quoted, cited and checked by a person. Book a demo →
The eight documents that decide whether a variation survives to the final account, and where they usually go missing.
Get it in your diagnostic →What to have attached before you submit, so a pay-less notice (the letter that cuts your payment) has nothing to bite on.
Get it in your diagnostic →If your QS leaves mid-job: where the job stands, captured, so the next person doesn't start blind.
Get it in your diagnostic →The one-page monthly rhythm that keeps the proof on a live job current, instead of rebuilding it at the end.
In progressEvery tool here is about visibility and capture: what the contract says, where the source is, and what needs review. None of it tells you a date to rely on or a sum you're owed; those stay with your QS and commercial team.
Written for the people who run subcontracts, from sources we name. No product pitch inside.
Three figures decide a subcontractor's month. The distance between them is where margin quietly leaks away, and most firms never line them up.
For MDs and commercial managers of specialist subcontractorsRetention is deducted automatically and released only when someone asks. A ledger of what's held, where, and when it falls due is the simplest money recovery there is.
For MDs and finance leads of specialist subcontractorsNotice windows start counting from events on site, not from when someone reads the clause. A visible board of live windows is the cheapest insurance a commercial team can run.
For QSs and commercial managers on NEC4 and JCT subcontractsTry it side by side on one contract. The differences show up fast, and they're the ones that cost money.
| On a real subcontract | A general AI chatbot | Senedra |
|---|---|---|
| A clause it can't find | Answers anyway. You may get a confident answer the contract never says. | No quote, no row. If it can't quote the clause, nothing goes in the register. |
| Backing an answer up | Prose, with nothing you can put in front of the other side's QS. | Every line carries its clause and page. Check any item in one click. |
| The term it missed | Simply isn't there, and nothing tells you it's missing. | The register is the checklist: payment, retention, notices, risk, each accounted for. |
| Sixty pages plus amendments | Gets skimmed. Long documents are where answers quietly degrade. | Reads every page, including the schedule of amendments, and a person reviews the result. |
| The deadline buried in clause 61.3 | Mentioned if you ask exactly the right question. | Surfaced on its own row with the wording quoted, for your QS to act on. |
We keep contracts we've marked up by hand for every form we read (NEC4, JCT, FIDIC and more). Every change to the engine is re-run against those sets before it ships, and it has to clear fixed accuracy floors to get through. When a human review catches something on a real job, that correction goes back into the test sets, so the read improves and can't quietly get worse. Those are our own test figures on our own marked-up contracts, not customer results.
How to test it yourself: take one closed job. Ask a general AI three questions about it: the payment terms, the notice deadlines, what was amended off the standard form. Then run the same contract through Senedra and compare what you can actually prove.