How the two engines work
Both assessments run on one set of answers, so nothing is lost in translation between them. Where they disagree, the disagreement is about the law, not about what you told each tool.
The CEST side is a faithful lookup of HMRC's published off-payroll decision tables at version 1.6.0. Your answers are mapped onto CEST's exact input values and run through its four sections, personal service, control, financial risk and part and parcel, followed by the combining matrix that produces the final result. It reproduces CEST's behaviour, including its early exits and its ability to return no answer at all. It is a reimplementation rather than a live call to HMRC's service, so if you need a determination to rely on with HMRC, run the official tool on gov.uk and keep the reference number.
The Verdict side scores the same answers across six dimensions: personal service and substitution, control, mutuality of obligation, financial risk, integration, and whether you are in business on your own account. Each is scored separately and combined into a probability rather than a binary label, because a finely balanced engagement genuinely is finely balanced and a one-word answer hides that.
Where they diverge, and why
Divergence is not noise. It is concentrated in four places, and knowing which one is driving your result tells you what to do about it.
Mutuality of obligation. This is the largest single source of disagreement, because CEST does not ask about it. HMRC's position has been that mutuality is present in every contract and therefore cannot distinguish between them, so the tool omits it. Tribunals assess the type and extent of the obligation rather than its bare presence, which is why an engagement with no obligation to offer or accept further work can look meaningfully different to a tribunal than it does to CEST.
The reality of the arrangement. CEST records what you tell it. If you say you have a right of substitution, it accepts that you have one, with no mechanism for asking whether it has ever been exercised or would be honoured. Autoclenz v Belcher (2011) established that where the written terms and the true agreement differ, the true agreement governs. Our questions distinguish between a right that exists on paper and one that has actually been used.
Post-2022 case law. The decision tables CEST applies predate the Court of Appeal in HMRC v Atholl House (2022) and the Supreme Court in HMRC v PGMOL (2024). PGMOL in particular changed how mutuality and control are understood, confirming that mutuality can exist inside a single engagement and that a framework of control can be sufficient even without day-to-day supervision. A tool built on earlier logic cannot reflect that.
The undetermined outcome. CEST returns no answer in around 22% of uses, on HMRC's own published figures. That is not a finding of inside or outside, it is the absence of a finding, and it leaves you without the protection HMRC offers for a determinate result. A probability based assessment always produces something you can act on, which is a different kind of answer rather than a more authoritative one.
What to do with a disagreement
If CEST says inside and our assessment says the picture is stronger than that, the useful next step is evidence rather than argument. Identify which dimension is carrying the difference, then establish whether the reality supports it: has a substitute ever worked in your place, do you carry genuine financial risk, can you decline work without consequence. A determination can be challenged, and where your client is the responsible party they must respond to a status disagreement within 45 days.
If both agree, that is genuinely useful information, in either direction. Our guides cover why CEST reaches the answers it does, how the same engagement can produce opposite results at tribunal, and who holds the pen on your determination. If you want your written terms assessed rather than your answers, the contract checker reads the agreement itself.
Common questions
Is this the real HMRC CEST tool?
It is a faithful reimplementation of the same decision logic, not a live call to HMRC's service. Our engine looks up the published off-payroll decision tables at version 1.6.0, the same tables HMRC's own tool applies, so the same answers produce the same outcome. For a determination you intend to rely on with HMRC, run the official tool at gov.uk and keep the reference number.
Why do CEST and IR35 Verdict give different answers?
Mostly because of what each one is willing to look at. CEST does not test mutuality of obligation at all, applies decision tables that predate the Court of Appeal in Atholl House and the Supreme Court in PGMOL, and takes your answers at face value rather than testing them against how you actually work. Our engine weighs all six factors and reports a probability, so a finely balanced engagement that CEST calls inside can come out differently.
What does an "unable to determine" CEST result mean?
It means the tool could not reach an answer, not that you are inside IR35. HMRC's own published figures put this at around 22% of uses. An undetermined result gives you no protection either way, which is precisely when a second opinion that shows its reasoning is worth having.
Which result should I rely on?
Neither, as a final answer. Only a tribunal determines status. CEST carries HMRC's stated commitment to stand by its result where the answers were accurate and given in good faith, which is a real practical advantage. Our result tells you how a case law based assessment reads the same facts, which is what you need in order to decide whether to challenge a determination.