HMRC asked whether disconnected business systems are an admin problem. That is half the question.

Our position, before the detail: the constraint on most businesses is not that integrations are unavailable. It is that nobody has decided which system is right. Connectors are cheap and getting cheaper. Agreement about who owns a record is neither, and it is the thing that determines whether an integration removes work or just moves it.

The government consultation is useful evidence for that argument, which is why it is worth reading properly rather than quoting selectively.

What the call for evidence actually asked

HM Revenue & Customs and the Department for Business and Trade jointly ran a call for evidence titled Business Systems Integration. It opened on 12 March 2026, closed on 4 June 2026, and at the time of writing the government has not published its response.

The framing is narrower than the title suggests. This is a record-keeping exercise: it asks about connections between the systems a business sells through and the software it keeps its accounts in.

What businesses sell through	What it would connect to	The admin it is meant to remove
EPOS and card payment machines	Accounting software	Takings currently re-entered by hand, or reconciled from a Z-report at the end of the day
Ecommerce and booking platforms	Accounting software	Sales, fees and refunds arriving as exports rather than as records
Bank feeds and payment service providers	Accounting software	Money movement that has to be matched back to the sale it came from
CRM systems	Accounting software	The customer record and the invoice record disagreeing about the same customer

One line matters more than the rest, and it is the line most likely to be left out of anything written about this consultation: the government states explicitly that it does not anticipate introducing requirements for businesses to integrate their systems. No mandate is being prepared. If anyone sells you an integration on the basis that HMRC is about to insist on one, they have either not read the document or are relying on you not having read it.

Why we think the connectivity framing is the smaller half

Ask whether integrations exist and the answer in 2026 is overwhelmingly yes. Every serious accounting product has an app marketplace. Bank feeds are standard. EPOS vendors publish APIs. The supply of connectors is not the bottleneck it was five years ago, and it keeps improving without anybody legislating.

Yet the administrative burden the consultation is investigating is still there. That gap is the interesting part, and it is not explained by availability.

Here is what we see instead when we look at businesses that already have connectors switched on and still have the problem. The systems are exchanging data. Nobody has agreed what happens when the data disagrees. A customer's address is changed in the CRM and not in the accounting package. A refund is issued in the ecommerce platform and appears in the ledger as a second transaction rather than a reversal. A booking is amended by phone and the till still holds the original.

In each case the connector worked exactly as designed. The business now has two versions of the truth arriving faster than before, and somebody spends their afternoon deciding which one to believe. That is not a reduction in administrative burden. It is the same burden, relocated and made harder to see.

The question the consultation does not ask

Which system is authoritative for each record, and what happens when two of them disagree?

We would argue this is the single question that separates an integration that pays for itself from one that quietly creates work. It cannot be answered by a software provider, because it is not a technical question — it is a decision about how the business is run, and only the business can make it.

It is also unglamorous enough that it gets skipped. Agreeing that the booking platform owns the appointment, the accounting system owns the invoice, and the CRM owns the contact details is an afternoon of unexciting conversation. Skipping that afternoon is what produces the reconciliation work nobody can later explain.

We set the practical version of this out in our systems integration checklist: define the contract before choosing the connector. The consultation is, in effect, evidence that a great many businesses have done it the other way round.

Where the expensive gap actually sits for most SMEs

Tax record-keeping is where this problem is measurable, which is why HMRC noticed it there. We would suggest it is rarely where the problem is expensive.

Re-keying yesterday's takings is tedious and it is bounded: the work is known, it happens once a day, and somebody does it. The costly disconnection in most small and mid-sized businesses is at the other end of the operation, where an enquiry arrives and dies between systems — a form submission nobody routed, a call that reached voicemail, a quote that was never followed up because the reminder lived in one person's inbox.

Nobody sends you a penalty notice for that. There is no return to file and no deadline to miss, which is precisely why it goes unmeasured for years. The research on response time, read carefully rather than as usually quoted, is largely about how quickly that kind of gap turns into lost revenue.

A business that connects its EPOS to its accounts and leaves its enquiry handling alone has automated the cheapest part of its administration.

What we expect the response to say

Worth putting a stake in the ground, so it can be checked when the response is published.

We expect it to conclude that the barrier is not the availability of integrations but the cost and confidence of setting them up — that smaller businesses report knowing integrations exist, being unsure which to trust, and having nobody internally who owns the decision. We expect the recommended direction to be industry-side: better defaults, clearer pricing, less configuration, rather than anything asked of businesses.

If that is right, it reinforces the point above. The gap being described is an ownership and confidence gap wearing a technical costume.

What we would look at in your business first

Not a list of products. Four handovers, and the question we ask about each.

Handover	The question we ask	What a bad answer sounds like
Enquiry to owner	When an enquiry arrives out of hours, what happens to it before morning?	“It depends who picks it up.”
Sale to record	Which system is right if the till and the accounts disagree about today?	“We check them both.”
Customer to invoice	Where does a change of address get entered, and where does it need to end up?	“Whoever notices updates it.”
Job to follow-up	What creates the next action after a job completes, and who sees it if nothing does?	“We usually remember.”

Every one of those is answerable in a twenty-minute conversation, and the answers decide whether an integration is worth building before anybody looks at an API. Where the answer turns out to be that the process itself is broken, connecting it would only make the breakage faster — we would say so, and the cheaper fix is the process.

What this does not mean

A consultation is not a signal to buy anything.

No requirement is coming. Integration is not automatically correct — we set out when it is the right intervention and when it is not on our systems integration page, and the wider buy, configure, integrate, automate or build decision in custom software vs off-the-shelf. Plenty of businesses reading this should connect nothing at all this year and fix a process instead.

What the consultation does usefully confirm is that the friction is real, widespread enough for two government departments to spend twelve weeks on it, and not something you have invented because your business feels harder to run than it should.

Common questions

What is business systems integration?

Business systems integration is the work of making the separate applications a business already runs exchange information reliably, so a fact entered once appears correctly everywhere else without a person carrying it across. In the HMRC call for evidence the term is used more narrowly, for connections between the systems a business sells through — EPOS, card machines, ecommerce and booking platforms — and its accounting software or bank feed.

Is the government going to require businesses to integrate their systems?

No. The call for evidence states explicitly that the government does not anticipate introducing requirements for businesses to integrate their systems as a result of it. It was gathering evidence on where integration helps and what gets in the way, not preparing a mandate. Anyone using this consultation to sell you software on the basis that rules are coming is misreading it.

Is the call for evidence still open?

No. It ran for twelve weeks, from 12 March 2026 to 4 June 2026, and is closed. At the time of writing the government has not published its response, so what the evidence showed is not yet public.

Does an integration reduce administration on its own?

Not reliably. An integration moves data; it does not decide anything. If nobody has agreed which system is authoritative for a record, connecting two systems replaces the work of copying information with the work of reconciling two versions of it. The saving comes from removing a decision a person was making, not from removing the typing.

Where should a business start?

With the handover that costs the most, not the one that is easiest to connect. In practice that means finding where information currently stops moving, agreeing which system owns each record, and only then choosing how to connect them. If the process itself is the problem, fixing the process is cheaper than automating it.