MTD turns one records request a year into five. Has anyone costed that?

Multiply your annual records requests by five. Four quarterly updates plus the year-end return. That is the number that decides whether Making Tax Digital hurts your practice, and it is not the number anyone has been writing about.

The rules have been explained to exhaustion — thresholds, start dates, compatible software, exemptions. The workload has been explained almost nowhere. Which is backwards, because the rules are HMRC’s problem to communicate and the workload is yours to absorb.

Five touchpoints, and four of them are new

Under the standard quarterly periods, updates cover 6 April to 5 July, 6 April to 5 October, 6 April to 5 January and 6 April to 5 April. The deadlines are 7 August, 7 November, 7 February and 7 May respectively. Calendar quarters — 1 April to 30 June, and so on — are an option, but they land on the same four deadline dates (GOV.UK, Use Making Tax Digital for Income Tax: send quarterly updates, checked 17 August 2026). The tax return itself must then be submitted by 31 January following the end of the tax year, after adjustments and after every other income source has been added and checked (GOV.UK, submit your tax return, checked 17 August 2026).

Five submission events per client, per year. Each one needs something from the client: the records themselves, or at minimum a confirmation that what you already hold is complete for the period.

There is an honest caveat, and it matters for your modelling. If a client is on monthly bookkeeping and their records already flow in without being asked for, your marginal cost is four extra review-and-file cycles, not four extra chases. If a client hands you a folder in November for the whole year, the multiplier really is five, and four of those five are chases you have never run before.

So the first count is not “how many clients are in scope”. It is how many in-scope clients currently supply records only once a year. That subset is where the entire workload increase lives.

The February deadline is the one to model first

Look at 7 February.

Self Assessment online filing and payment both fall on 31 January (GOV.UK, Self Assessment tax returns: deadlines, checked 17 August 2026). The third quarterly update deadline lands seven days later. The January you already struggle with now has a week-long tail with its own hard deadline attached — and the records for that quarter need to be in your hands before the 31st, not after it, because nobody is starting a chase cycle on 1 February.

7 May is the second problem. It is the fourth update for the year just ended, and it arrives while finalisation work for the previous year is still open on your desk.

If you only have the appetite to model one quarter properly, model the quarter ending 5 January, due 7 February. It is the one that will tell you whether the plan works.

Which of your clients, and when

Qualifying income	Tax year it is measured in	Must use MTD from
More than £50,000	2024 to 2025	6 April 2026
More than £30,000	2025 to 2026	6 April 2027
More than £20,000	2026 to 2027	6 April 2028

Source: GOV.UK, Find out if and when you need to use Making Tax Digital for Income Tax, page last updated 26 March 2026, checked 17 August 2026. These dates and thresholds have moved more than once. Re-check the page before you plan against it.

The first band is already live. The second band is the one that changes the shape of a typical small practice, because £30,000 of qualifying income catches a great many sole traders and landlords who currently send you a shoebox once a year.

Do the arithmetic on your own book

No published average is any use here. Five numbers from your own records are:

1. In-scope clients per band. Count them against the table above, by the tax year that determines their entry.
2. Of those, how many are annual-records clients. These are the ones whose request count goes from one to five.
3. Your real chase ratio. Not your estimate — count it. Pick twenty files from last year and count the actual emails, calls and reminders it took to get a complete set. If the honest answer is four contacts per client, then five requests is twenty contacts.
4. Multiply. Annual-records clients × 5 × chase ratio. That is your new contact volume.
5. Divide it into four windows. This is the step that hurts. Today that volume is smeared across the calendar and absorbed by whoever has a quiet afternoon. From April it compresses into four fixed fortnights, one of which is the first week of February.

If step 5 produces a number your current staffing cannot clear in a fortnight, you have found the real cost of MTD, and it is not a software licence.

Recurring work should be scheduled, not remembered

The saving grace is that the new work is identical every quarter. Same clients, same list of items, same dates, known a year in advance. That is the profile of work that should be issued by a schedule rather than by someone remembering.

Three routes, and you should price all of them yourself — I have not verified current pricing for any of them:

• Practice management suites — Karbon, IRIS Elements, BrightManager and similar. Job templates and recurring workflows sit alongside everything else you already run in them. More configuration up front, and you are buying a whole system to solve one problem.
• Dedicated request tools — Content Snare, FileInvite, ClientLoop. Narrower scope: issue a defined list of items, chase automatically until it lands, sit alongside whatever practice system you already have. ClientLoop supports recurring request schedules and automatic reminders, which is the specific shape of this job; so, broadly, do the others.
• What you already own. A saved item list, four calendar entries per client, and a mail merge. Free, unglamorous, and genuinely sufficient for a small in-scope book. Do not buy software to solve a forty-client problem you can solve with a spreadsheet.

Whichever route, the thing to change before April is the artefact: the records request stops being an email someone writes and becomes a defined list that reissues on a schedule. That change is worth more than the tool you make it with.

The judgement

The software licence is the cheap part of MTD, and it is the part the profession has spent three years discussing. The expensive part is four additional chase cycles per client per year, against headcount you cannot flex, on dates HMRC chose rather than dates that suit you — one of them a week after the worst day in your year.

That cost is entirely knowable now. It takes an afternoon with your client list and last year’s email history to produce the actual number for your practice. Practices that run that afternoon in August have eight months to fix what it tells them. Practices that discover it on 1 February 2027 will be discovering it while the clock runs.