What a fractional AI engagement should deliver in the first 30 days
6 minute read. Updated 2026-08-08.
By day 30, a fractional AI engagement should have produced a written recommendation with a rough cost and timeline attached, specific findings on data quality, and an honest answer on whether the original brief is achievable, including a recommendation to stop if that's the truth. If instead you have only meetings and reassurance, set a 7-day deadline for the missing deliverable, then end the engagement if it doesn't appear.
What day 30 is actually for
30 days is long enough to find out whether the problem you hired for is the problem you actually have, and short enough that if nothing concrete exists by then, you haven't lost much. A good specialist treats day 30 as a checkpoint with named outputs, not a general sense of progress.
What follows isn't a wish list. Each item is something you can hold in your hand on day 30, a document or a working prototype, not a feeling. If a specialist can't point to something specific, ask why.
Week 1: access, not conversation
By the end of week 1, the specialist should have access to the systems and data they need to actually assess the problem, not a series of meetings about getting access. If week 1 closes with only meetings booked for week 2, the engagement has already lost a fifth of its time to your own process.
You should also have a short written scope, no more than a page: what problem is being solved, what's explicitly out of scope, and how success will be measured. If this doesn't exist by day 5, write it yourself and send it across for agreement, don't wait for it to arrive.
If the work touches more than one team, by week 1 you should also know who signs off the final recommendation. Projects stall most often not because the technical work is hard, but because nobody agreed in advance who gets to say yes.
What 'assessed the data' actually means
A generic status update saying the data has been looked at tells you nothing. By day 30 you should have specifics: which fields are missing or wrong in ways that matter for the model, and how much of the target dataset is actually usable as it stands.
This matters because most AI projects that fail, fail on data quality discovered too late, not on model choice. A specialist who surfaces this in week 1 or 2, rather than week 11, has paid for themselves regardless of what happens next.
By day 15: a straight answer on feasibility
Around the halfway point, you should get an honest read on whether the thing you hired someone to build or fix is actually achievable with your data and your systems, and what would need to change if it isn't. This is often uncomfortable, and it's the single most useful thing a fractional specialist can tell you, because someone employed full-time and under pressure to justify their own role rarely delivers it straight.
If instead you're getting reassurance with no specifics, that's the first warning sign. Someone 15 days in who can't yet describe a concrete blocker either hasn't looked closely enough, or is managing your mood rather than the project.
By day 30: a document with a cost on it
The deliverable for day 30 is a written recommendation: what to build, what it will take to build it, roughly how long, and roughly what it will cost, whether in more of this specialist's time or in headcount you'd need to hire. It should name the risks plainly, including the ones that make the project not worth doing.
20 working days at the median advertised Artificial Intelligence day rate of £551 comes to £11,020. That's roughly what a month of a fractional engagement costs before any output is counted. For that money you should have a document you could hand to a board, not a slide deck of possibilities.
If the recommendation is to stop, that's a valid day 30 outcome too, and often the most valuable one. A specialist who tells you plainly that the project isn't worth pursuing, before you've spent 6 months on it, has still done the job.
The signs of stalling, named
Watch for a scope that keeps growing without a matching change to the deliverable date. Watch for updates that describe activity, meetings held, models tried, without describing findings.
A specialist who hasn't yet said no to anything you've asked for, a month in, is probably managing your expectations rather than the project. None of these signs is damning alone. Together, by day 30, they mean you're paying for presence, not progress.
If none of this has happened
Say so directly, and set a hard deadline, 7 more days, for the missing deliverable to appear. If it still doesn't, end the engagement. This isn't harsh, it's what a fixed-term arrangement is for: the decision to continue or stop is supposed to be cheap and clear.
Don't renew a contract past day 30 on the strength of good meetings alone. Renew it because you've read a document that tells you plainly what's true about your data, your systems and your options, written by someone with nothing to gain from telling you otherwise.
Look back in 6 months at whether that day 30 document turned out to be right. If it wasn't, that tells you something about how carefully it was written, not just about how the project turned out.
What to do about it
- Insist on system and data access in week 1, not meetings about access.
- Get a short written scope, with sign-off named, by day 5.
- Expect specific data quality findings by day 30, not general reassurance.
- Expect an honest feasibility answer around day 15, not comfort.
- The day 30 deliverable is a written recommendation with a cost attached, including a recommendation to stop if that's the honest answer.
- If none of this exists by day 30, set a 7-day deadline, then end the engagement.
Questions people also ask
What if the specialist says everything is fine but there's nothing in writing?
Ask for it in writing anyway, and treat the reluctance as information. A specialist who has done a proper 30-day assessment has notes, findings and a recommendation already, because that's the natural output of the work. Verbal reassurance with nothing written down usually means the assessment hasn't actually happened yet, whatever's been said in meetings.
How much should 30 days of a fractional AI engagement cost?
It depends on the role and rate agreed, but as a reference point, 20 working days at the median advertised Artificial Intelligence contract day rate of £551 comes to £11,020. Use that as a rough anchor for what a month should cost, and judge the deliverable against that figure, not against goodwill.
Is it normal for the specialist to say the original brief isn't achievable?
Yes, and it's a good sign rather than a bad one. A specialist willing to say the brief needs to change, this early and this plainly, is doing the job you hired them for. One who agrees with everything you originally asked for, unchanged, a month in, either hasn't looked hard enough or isn't willing to be the bearer of bad news.
What should I do if day 30 arrives with nothing concrete delivered?
Say so directly and give a short, fixed extension, around 7 days, for the specific missing item to appear. If it still doesn't, end the engagement. A fixed-term arrangement exists precisely so that this decision is cheap; don't let politeness turn a 30-day checkpoint into a 6-month drift.
Where the figures come from
Every rate and salary quoted in this article is a median or percentile of figures advertised in UK job postings over the six months to 8 August 2026. They are not rates paid, and the gap widens at the top of a range.
- IT Jobs Watch, UK contract rates, 6 months to 8 August 2026, read 2026-08-08.
- IT Jobs Watch, UK permanent salaries, 6 months to 8 August 2026, read 2026-08-08.