HMRC CEST Tool: IR35 Kill Switches Explained

Introduction

For over twenty years we've been helping UK contractors understand and renegotiate their way through IR35. For the most part IR35 is employment law, and we're not lawyers, but with IR35 being such a dominant part of the conversation for the past 20 years, it's something we have come to know very well.

We've listened on the other end of the phone to contractors holding Inside IR35 Status Determination Statements and asked the question every contractor really wants the answer to: is there anything we can do about this? In a lot of cases, the SDS rests on a CEST verdict. And in almost every case, the verdict is treated as the end of the conversation.

CEST stands for Check Employment Status for Tax. It's the free online tool HMRC publishes at gov.uk for working out whether an engagement should be treated as employment or self-employment for tax purposes – in practice, whether a contractor falls inside or outside IR35.

You (or, much more commonly since the 2021 off-payroll reforms, your end client) answer a questionnaire covering substitution, control, financial risk, integration with the client's team, and how business-like the engagement looks overall. CEST then spits out one of three verdicts: Inside IR35, Outside IR35, or Unable to determine.

HMRC has said it will stand by the result "provided the information given is accurate" – which sounds reassuring until you realise the burden of accuracy sits entirely with the user, the verdict isn't legally binding, the IR35 decision engine is more brittle and mechanical than most users realise, and it can't actually do the analysis the case law now requires.

When the PGMOL ruling landed earlier this month it underlined why the wider stage-three analysis (which is looking at whether the contractor is genuinely in business on their own account) cannot be treated as an afterthought. So we decided to stop hand-waving about it, and we got our hands on the source code.

The conclusion isn't subtle. The entire CEST decision engine is two kill switches and a 72-row spreadsheet. There's no machine learning. There's no AI. There isn't even much rule-engine sophistication. And once you can see what it actually does, you can see exactly why contractor after contractor lands in our inbox with either Inside IR35 or Undecided verdicts that don't survive a proper review.

The engine in plain English

Here's the entire CEST decision engine in one paragraph.

You (or, more commonly, your end client) fill in a questionnaire covering seven topics. There's an exit section that asks whether you're an office holder, then five substantive sections: personal service(substitution), control, financial risk, part and parcel (integration), and business on own account. Each section's answers get looked up in a CSV spreadsheet. The spreadsheet returns one of four scores per section: `LOW`, `MEDIUM`, `HIGH`, or a special `OUTSIDE_IR35` kill switch. The five section scores then get combined in a final 72-row matrix, which produces the final verdict: Inside IR35, Outside IR35, or Unknown.

Before the matrix gets consulted, two short-circuit kill switches can produce a verdict on their own:

(1) If you answer Yes to "are you an office holder", the result is immediately Inside IR35, no matter what you say to anything else.

(2) If any single section returns the special `OUTSIDE_IR35` score, the result is immediately Outside IR35, no matter what you say to anything else.

Finding 1: "Unknown" is the result nearly half the time

That 72-row matrix we just described, the one that combines the five section scores into a final verdict, produces an Outside IR35 result in 5 rows, an Inside IR35 result in 34 rows, and a third outcome called `Unknown` in 33 rows.

Across all the combinations of answers, the engine reaches a clear verdict in 39 of 72 cases. Just over half. For the other 33, the decision logic returns the equivalent of "we can't tell."

This matters because HMRC has consistently said it'll "stand by the result" from CEST, provided the information given is accurate. The unstated assumption in that promise is that the CEST engine actually reaches a result. On the engine's own terms, it doesn't, almost half the time.

Finding 2: the office-holder trap

The very first question CEST asks is whether you're an "office holder". The on-screen wording reads:

> Are you an office holder? This can include being a board member, treasurer, trustee, company secretary or company director.

If you tick Yes, the engine immediately returns Inside IR35. The rest of the questionnaire is bypassed. Substitution, control, financial risk, integration, business on own account – none of it gets looked at.

Every contractor running a limited company is a director of their own PSC. The question doesn't specify of which organisation. A contractor reading the examples in good faith can quite reasonably tick Yes.

The legal hook here is Section 5 of ITEPA 2003, which treats income from an office as employment income. The point of the question is to catch contractors brought in to perform the duties of an office, for example a contractor engaged as an interim CFO, where the income arises from the office of CFO itself. That's a narrow and specific carve-out.

The question doesn't capture the narrowness. Tick the wrong box because you're a director of your own PSC, and you've just talked yourself into an Inside IR35 verdict on an unrelated contracting engagement.

If the CEST tool headline question was changed to "Are you an 'office holder' for the end client you will be providing services to", it's a much easier trap to avoid.

Finding 3: the Outside IR35 'kill switches'

Earlier we said the second kill switch fires when any section returns the special `OUTSIDE_IR35` score. That's true, but it understates what's actually going on. There are seven distinct answer patterns that can cause a section to return `OUTSIDE_IR35`, and each one short-circuits the entire engine to an Outside IR35 verdict regardless of what you said anywhere else, in any other section.

They're worth listing in full, not because you can use them to trick your way to being outside IR35, but just so you can see how the tool works.

(1) "Will you have to buy equipment before your client pays you?"

Answer Yes, and your CEST result is Outside IR35. Always. It does not matter what you said about substitution, control, integration, or anything else. A laptop won't cut it though – see ESM11090 for what HMRC actually means by "equipment" here.

(2) "Will you have to buy materials before your client pays you?"

Yes triggers automatic Outside. But the construction-industry framing ("forms a lasting part of the work, or is left behind when you leave") means most knowledge workers honestly answer No.

(3) "Will you have to fund any vehicle costs?" AND "Will you have to fund any other costs?"

Yes to both triggers automatic Outside. Rare combination for office-based work.

(4) "Yes, unpaid and you would have extra costs that your client would not pay for" on the rectification question, paired with either vehicle costs or other expenses, will trigger automatic Outside for most income types. Requires a specific cluster of answers most users won't give.

(5) You have actually sent a substitute, the client agreed, and you paid the substitute (not the client) – automatic Outside. Almost nobody has ever done this.

(6) The client "would not reject" a hypothetical substitute, and you would pay for them yourself – automatic Outside.

(7) The contract states the client cannot move you off the agreed task without a new contract, and you control your schedule (or it's purely deadline-based), and you control your location (or location doesn't matter), and you either decide how the work is done yourself or follow strict externally-defined procedures – automatic Outside. That fits a fixed-price construction sub-contract or a regulated profession. It does not fit a typical professional services engagement where scope adapts as the project progresses.

Finding 4: the honest middles trap

The next problem is subtler, and in our experience it's the most common reason genuine contractors are getting Inside IR35 verdicts out of CEST.

The Control section asks four questions: who decides what work is done, who decides how it's done, who decides when, and who decides where. Each question offers three or four answers. The pattern across all four is the same: a strong "client decides" option at one end, a strong "I decide" option at the other, and a diplomatic middle option that says, in various forms, "we agree it together."

Now picture a typical professional engagement. A management consultant. A senior interim. An IT contractor doing real work on a complex platform. How the work is done is a genuine collaboration with the client team. When it's done involves some agreement with the people whose time you need. Where is hybrid by mutual consent. The honest, accurate, professional answer to all four questions is the diplomatic middle one.

Here's what the engine does with that.

The diplomatic middle answers cluster in rows that produce `MEDIUM` or `HIGH` control results. There's no path from "I'm a collaborative professional contractor" to a `LOW` score in Control. The only way Control can produce a positive outside signal is via the kill switch, and the kill switch requires you to answer that the client can't move you onto different work without a new contract, and you decide how the work is done with no client input, and the work is by deadlines only with no client schedule, and you choose where you work or it's location-agnostic. All four. Anything less, and Control contributes a neutral-to-inside signal.

So a reasonable contractor giving fair, moderate, accurate answers maps directly to an inside-leaning verdict. The engine gives no signal that this is what's happening. There's no warning that the middle answer carries weight. No indication that "we agree it together" is being scored as partial client control.

This is the trap that catches the most people. It's not exploitable in the wrong direction the way the office-holder kill switch is. It's just quietly punishing the contractors whose engagements look most like ordinary professional service delivery.

Finding 5: The stage-three wall

This is the finding that matters most after PGMOL, and it's the reason we wrote this blog.

Our PGMOL piece argued that the centre of gravity of IR35 has just moved. Mutuality and control aren't enough on their own any more. The case will now be decided at stage three, the in-business-on-own-account (IBOOA) test. That's where contractors have the most say in shaping the outcome, and that's where a tribunal may need to stand back and look at the wider picture.

So a fair question to ask of CEST is: can it do a credible IBOOA analysis?

From above, there are just 5 "Outside IR35" results possible. Out of seventy-two possible results, and those five rows have one thing common. Every single one of them requires the Business on Own Account score to be `HIGH`.

In CEST's scoring scheme, a `HIGH` BOOA score means the contractor is showing strongly business-like behaviour: multiple concurrent clients, retained ownership of work product, the engagement not consuming a significant share of their time, the contract being a one-off rather than part of a series.

If your BOOA score is `LOW` or `MEDIUM`, which describes essentially every contractor who's full-time on a single engagement, however flexibly that engagement is structured, however good the substitution clause, however little the client controls, the CEST tool cannot give you an Outside verdict. Never (well, unless in the unlikely event you hit on the Outside IR35 kill switches above).

Take a contractor who answers honestly to this one question – 'have you done any self-employed work for other clients in the last 12 months?' – and says No. Maybe they're new to contracting. Maybe their last engagement ran for 15 months. Maybe their previous work was through PAYE before they set up their PSC. Whatever the reason, they tick No.

That single answer locks them out of every matrix path to an Outside IR35 verdict. No matter what else they answer, if they tick No to this, the result will never be outside IR35.

What CEST doesn't even ask

It's worth listing the factors CEST never asks about. Each of these is a case-law staple. Each is something PGMOL has just made more important, not less.

The engine has no field for mutuality of obligation. Not partial, not approximate, none at all. The Supreme Court in PGMOL re-emphasised mutuality as a separately necessary element of an employment relationship. CEST embeds HMRC's institutional position that mutuality is automatic given any contract, a position the Supreme Court didn't endorse.

The engine has no field for professional indemnity insurance, or public liability insurance, or any form of business insurance. PI cover is one of the cleanest financial-risk markers there is for a knowledge-economy contractor, and the engine can't see it.

The engine has no field for the right to refuse work. That was the factor the PGMOL tribunal placed real weight on. The referees could decline matches. CEST never asks the equivalent question.

The engine has no field for your broader career portfolio beyond a 12-month window. A 15-year contractor with 30 prior clients but who happens to have done one long recent engagement scores the same as a brand-new entrant.

And the engine has no field for business identity. No LinkedIn, no website, no branded invoicing, no separate business bank account, no PII insurance, no marketing spend, no recruitment-firm relationships, no work-acquisition documentation. Of the five IBOOA strengtheners we walked through in the PGMOL piece, CEST captures one partially and four not at all.

The bigger picture

So we've got a tool whose first question can force an Inside verdict on a misread. Whose substantive decision logic returns "we can't tell" for nearly half of the answer combinations it sees, with the scale of that uncertainty largely hidden from the ordinary user journey. Whose Control section punishes the most accurate answers a professional contractor can give. And which doesn't ask about half a dozen of the factors case law has just told us are decisive.

This is what your end client is using to assess your IR35 status. This is what a lot of Status Determination Statements are built on. This is the tool HMRC has said it'll stand by the results of, "provided the information given is accurate", while shipping a tool whose underlying engine doesn't reach a result in nearly half of the cases it's asked about.

The kindest thing you can say about CEST is that it was designed for a 1960s tradesperson view of self-employment (buys tools, drives a van, supplies materials) and has been bolted onto twenty-first-century knowledge-economy contracting in a way that produces inside-leaning verdicts almost by structural default.

What to do about it

If your end client has just landed you with an inside-IR35 SDS that leans on a CEST output, the practical implication of all of the above is clear. The CEST verdict isn't the end of the conversation.

The first move is to get a proper status review done by a firm that specialises in IR35 employment status work. We're accountants, not employment lawyers, and formal status determinations aren't a service we offer. But this is one of the situations where the cost of a proper review is trivial compared to the cost of accepting an inside determination for the rest of an engagement. We've got a number of specialists we trust and can point you to one.

In parallel, do the work the PGMOL piece walked through. The factors the case will actually turn on (engagement structure, team separation, business identity, work-acquisition trail, substitution and right-to-refuse) are the factors CEST can't see. Strengthening that picture is the work that gives a status review something to actually defend.

If the client won't move and an umbrella arrangement becomes the practical landing place, do it with your eyes open. A clean umbrella beats a messy inside-IR35 PSC arrangement on most measures, but it's still a downgrade from outside-IR35 if outside was genuinely available.

If you're an end client running CEST as part of your SDS process, the harder question is whether a CEST output alone meets your reasonable-care obligation under the off-payroll working rules. After PGMOL, our view is that it doesn't. The engine can't do stage three. Your SDS process needs to.

The Bottom Line

If your IR35 status has been decided by CEST, the verdict should really just be a starting point for a real assessment.

If you're sitting under an inside-IR35 SDS that was driven by a CEST output, get in touch with No Worries Accounting. We've been guiding Kiwi and Aussie contractors through the UK tax system for over twenty years. We won't do the formal status determination ourselves but we'll give you an honest read on whether the SDS is worth challenging, and we'll point you at the specialists who can do the formal work.