What Making Tax Digital is really costing your practice in hours

The value an accountant adds is advisory - interpreting a client's position and signing off a return under the practice's name, not keying their transactions into a ledger. That's why accountants sit inside a professional-conduct framework that pure data-entry work doesn't touch in the same way. ICAEW members work under the PCRT (Professional Conduct in Relation to Taxation), which sets out that professional judgement, human oversight and transparency to HMRC are required elements of tax work, not something a tool can take off your hands.

That advisory work is where the practice earns its fee - reviewing a set of accounts, advising on structure, catching what a client's own bookkeeping missed. It's judgement, not throughput. But the quarterly cadence created by Making Tax Digital for Income Tax is multiplying the volume of admin sitting underneath that judgement, and that admin is what this post maps.

The admin MTD didn't cause, but did multiply

Making Tax Digital for Income Tax (MTD ITSA) went live on 6 April 2026 for sole traders and landlords with gross income above £50,000. The first quarterly submission window - April to June - closed at the start of August, with the deadline for filing it falling on 5 August 2026. A second phase brings everyone earning above £30,000 into scope from April 2027, roughly doubling the affected client base again, and a third phase drops the threshold to £20,000 from April 2028, pulling in most of the remaining sole traders and landlords still outside MTD.

For the client, MTD is a compliance change. For the practice handling their affairs, it is something closer to a structural shift in the shape of the year. An annual return that landed once, in a predictable window, has become four filing cycles a year, every year, for every client in scope. The admin did not get more complicated. It got more frequent - and frequency is what breaks a practice's capacity, not complexity.

None of the underlying tasks are new. Chasing records, answering "is this deductible" questions, checking a client's software is actually capturing what it needs to - practices have always done this. What has changed is the cadence. A task that used to happen once a year, in a window you could plan resource around, now happens four times, on a rolling basis, for a growing share of your client bank.

That is the specific pressure behind this post: not "AI for accountants" in the abstract, but the practical question of where a quarterly filing cycle actually eats practice hours, and which parts of that cycle are safe to hand off.

Where the hours actually go, every quarter

Based on how the quarterly cycle plays out across a typical UK practice, four tasks account for most of the added load. The ranges below are illustrative - in a practice this size, we typically see figures in this territory, not a verified industry benchmark - but they are the shape of what shows up when we map a practice's workflow in an assessment.

• Chasing quarterly records. Bank statements, receipts, mileage logs. Where a year-end chase might once have run to three or five follow-ups per client across a whole year, that same chase pattern now repeats every quarter. For a practice with 30-40 clients in MTD scope, that can mean several hours a week in the run-up to each deadline, purely on records-chasing emails.
• Query triage around each deadline. "Is this expense allowable this quarter?" "Why does this figure look different from last time?" Volume spikes in the ten days either side of a filing date, and each query that used to take five minutes can stretch to fifteen or twenty when it is written from scratch under time pressure.
• Verifying clients are actually MTD-compliant, quarter to quarter. A client who switched from a paper diary to Xero, QuickBooks, Sage or FreeAgent still needs checking - are receipts being captured in real time, is the categorisation consistent, is the bank feed connected properly. That verification step, done properly, is not a one-off onboarding task under MTD. It recurs.
• Pre-submission reconciliation checks. Before a quarterly figure goes to HMRC, someone at the practice needs to look at it. That review step, multiplied by four a year instead of one, is where a lot of the added capacity pressure actually sits - and it is the one task on this list that should never be shortened, only made faster to prepare for.

Add those four together across a client bank moving through MTD, and the capacity ceiling problem gets sharper. New clients you would ordinarily take on now compete with a filing cycle that runs four times as often as it used to.

What is safe to automate here, and what is not

The temptation, faced with a quadrupled filing cycle, is to reach for automation everywhere. That is the wrong instinct in a regulated profession, and the ICAEW's guidance on AI and the PCRT (Professional Conduct in Relation to Taxation) is clear on why: judgement, oversight and transparency to HMRC have to come from a person, not a tool standing in for one.

What that means in practice, task by task:

Safe to hand off, with review: drafting the records-chasing sequence (a three-stage reminder - gentle nudge, direct ask, final chase before the deadline - written once per client type and personalised via merge fields rather than composed fresh each time); drafting first-pass responses to common client queries for you to check before sending; document and receipt ingestion through tools like Dext or AutoEntry, which have been doing AI-powered extraction for several years and are well suited to the categorisation-and-query layer that still sits on top of them; and the built-in AI reconciliation suggestions inside Xero, QuickBooks, Sage and FreeAgent, provided someone is still checking the output rather than accepting it by default.

Not safe to hand off: the technical judgement behind whether a specific transaction is allowable, any figure that goes into the quarterly submission without a professional review step, and anything client-facing where the advice carries weight beyond a routine reminder. The efficiency gain on that last category is never worth the professional risk or the relationship cost of getting it wrong.

The practices handling MTD well are not the ones automating the most. They are the ones who have been precise about which half of that list is genuinely safe to template, and which half stays firmly with a qualified person.

One quarter, before and after

Take the records-chasing task on its own, for a practice with around 40 clients now filing quarterly. Before any change, that chase typically runs three to five follow-up emails per client across the quarter - each one drafted from scratch, personalised by memory, and tracked by eye in an inbox or spreadsheet. Across 40 clients, that is a genuinely large volume of near-identical drafting, and in a practice this size we typically see it running to a working day or more of admin time concentrated in the fortnight either side of the deadline.

After restructuring the same task, the practice writes the three-stage sequence once per client type - sole trader, landlord, mixed income - with merge fields for name, missing item and deadline date. A shared tracker flags who has responded and who needs the next stage. The person running the chase is now reviewing and sending a drafted message rather than composing one, and triggering the next stage from a list rather than remembering who is overdue. The task does not disappear. It shrinks from a working day of scattered admin to a couple of focused hours, and the actual reconciliation work behind it is unchanged - the saving sits entirely in the chasing layer, not in anything that required professional judgement.

That is the pattern worth generalising: MTD did not create new work that needs a person's expertise. It multiplied the frequency of work that mostly did not need it in the first place.

Find your own number

The ranges above are a starting shape, not your practice's actual figure. Client mix, how far along your team already is with digital record-keeping, and which software your clients are spread across all change where the hours really sit.

Our admin cost calculator for accountants gives a rough starting estimate based on your practice's size and billing rate - a quick way to see the scale of the number before going further.

For the fuller picture, the AI workflow assessment for accountants maps your practice's actual quarterly cycle - not a generic one - against the tasks above, and comes back with a prioritised, named plan for what to automate first and what stays with a qualified reviewer. It sits alongside our post on where AI actually saves UK accountants time (and where it doesn't), which goes deeper on the onboarding and query-triage side of the same problem.

If you would rather start smaller, the free AI readiness quiz takes about two minutes and flags your highest-friction area before you commit to anything.

Either way, the number worth knowing is not a national average. It is how many hours your practice specifically is now spending four times a year on work that used to happen once - and that is the question worth answering before the next quarterly deadline lands.

Sources: GOV.UK, Making Tax Digital for Income Tax step-by-step guidance (gov.uk/government/collections/making-tax-digital-for-income-tax-for-businesses-step-by-step), for the 6 April 2026 go-live date, the £50,000 threshold, the 5 August 2026 first-quarter filing date, the April 2027 second-phase threshold of £30,000, and the April 2028 third-phase threshold of £20,000. ICAEW guidance on AI and the PCRT (Professional Conduct in Relation to Taxation), icaew.com, for the professional-judgement and oversight requirements referenced above.