Buy fixed price when the scope is genuinely knowable: an audit, a report conversion, a defined build where the definitions are already agreed. Buy day rate when nobody yet knows what is in the data, because a fixed price quoted against unknown data is either padded to cover the worst case or destined for a change request the moment reality shows up.
The mistake is treating this as a negotiation about who bears risk. It is really a question about how much is known, and the answer changes at different stages of the same project. Most well-run Power BI work uses both: fixed price to find out what you are dealing with, then fixed price again for a build that is now knowable.
Which model fits which work?
| Type of work | Better bought as | Why |
|---|---|---|
| Audit or assessment of an existing estate | Fixed | The scope is the checklist, and it doesn't change |
| Converting existing reports | Fixed, per report | The target is defined: match the old report |
| Building on data nobody has examined yet | Day rate, or fixed after an assessment | The unknowns are the cost |
| A defined system, definitions already agreed | Fixed | Knowable, so the risk belongs with the supplier |
| Ongoing changes and improvements | Retainer | Neither party can specify next quarter's work |
| Emergency fix on a broken report | Day rate or hourly | Diagnosis time is genuinely unpredictable |
| Training | Fixed, per day or session | Content and duration are both known upfront |
Notice that most of the list is fixed. The genuinely unpredictable categories are narrower than the industry's habit of quoting day rates for everything suggests.
Why "it depends" is usually about your data
When a supplier will not quote a fixed price for a build, the honest reason is almost always the same: they have not seen your data and they know from experience that its state drives most of the cost.
That is a defensible position. The state of the data is genuinely the largest variable in a Power BI project, larger than the number of reports or the complexity of the visuals. Two builds that look identical on paper can differ threefold depending on whether the source systems agree with each other.
But "I cannot price this" is only half an answer. The better one is to fix the price of finding out. That is what an assessment does: a defined, bounded piece of work that converts unknowns into a specification, after which a fixed price for the build becomes possible for both sides.
Which is why I sell the health check at £950 as a product rather than giving away discovery. Free discovery is never free. It gets priced into the build, in a number you cannot see, along with a margin for the risk that discovery reveals something horrible.
What each model actually protects
Fixed price protects you from the supplier being slow. Once the price is agreed, inefficiency is their problem. This is a genuine and underrated benefit: day rates quietly reward taking longer, and while most consultants are honest, incentives shape behaviour at the margins whether anyone intends it or not.
Fixed price protects you from budget surprise. You can get approval, plan around it, and not have to explain a variance later. For anyone who has to justify spend internally, this is often worth more than a lower headline rate.
Day rate protects you from paying for risk that never materialises. A fixed price against uncertain work includes a contingency. If the work turns out to be straightforward, you paid for a risk that did not happen. On day rate, you would have paid less.
Day rate protects you from a supplier cutting corners. The pressure under a fixed price, when the work turns out harder than expected, is to deliver the minimum that meets the letter of the specification. A good supplier absorbs it. A struggling one starts arguing about what was in scope.
The change request problem
The main way fixed price goes wrong is not the price. It is that the specification becomes a weapon.
You agree a fixed price for a defined build. Three weeks in you realise you also need the figures split by branch. That is a reasonable request, it was not in the specification, and now you are in a change request conversation that neither party enjoys, in the middle of the work.
Two things reduce this. First, a specification detailed enough to be real, which means agreeing definitions before the build rather than during it. Second, a supplier with enough commercial sense to absorb small changes rather than invoicing for every one. That second thing you can only judge from how they behave, which is a good reason to buy something small from someone before buying something large.
The version of this that should worry you is a very low fixed price paired with a rigid specification. That combination is a business model: win on price, recover on changes. The tell is a quote noticeably below everyone else's alongside unusually precise scope wording.
How I price, and why
My own prices are published, which is unusual in this market and deliberate.
Assessments are fixed because their scope is a checklist: a health check is £950, a full estate review £2,450, a Fabric readiness assessment £5,950. None of those depend on what I find, so there is no reason for the price to.
Conversions are per unit, at £600 per report, because the target is defined by the report that already exists. Count your reports, multiply, and you have the number without a meeting.
Builds are fixed once scoped, at £9,500 for a reporting system, after the definitions are agreed. Ongoing work is a monthly plan from £250, rolling with no minimum term, because nobody can specify next quarter's changes in advance and pretending otherwise produces bad contracts.
I do not think this is the only honest way to price. Day rates are perfectly legitimate, and a good consultant on day rate can be excellent value. What I would push back on is the position that nothing can ever be fixed, because plenty can, and the parts that cannot are usually smaller than the industry implies. For market context, UK Power BI contractor day rates had a median of £490 in the six months to 26 August 2026, ranging £320 to £656 across 1,237 quoted rates, according to ITJobsWatch. The full cost breakdown puts that in context.
What to ask before you sign either kind
For a fixed price: what specifically is included, what happens if the data turns out to be worse than expected, how are changes handled, and what does "done" mean. That last one matters more than people expect. Done on the supplier's terms can mean the report exists. Done on yours usually means it works, it is documented, and someone can maintain it.
For a day rate: what is the estimate, what would make it longer, how often will you tell me where we are against it, and at what point do you come back to me rather than keep going. A day rate with no estimate is not a commercial arrangement, it is an open tab.
For both: who owns the output, and is it in writing. The handover checklist covers what should come back to you at the end regardless of pricing model.
Common questions
Is fixed price or time and materials better for a Power BI project?
Fixed price when the scope is knowable, which covers audits, conversions and builds where definitions are agreed. Day rate when the data is unexamined and the unknowns dominate. The best structure for a larger project is often a fixed-price assessment followed by a fixed-price build, so uncertainty is resolved before the big number is agreed.
Why won't consultants give a fixed price for Power BI work?
Usually because they have not seen the data, and its condition is the largest driver of effort. That is a fair reason, but the better response is to fix the price of a short assessment rather than to refuse to fix anything at all.
Is a fixed price always more expensive?
It includes a contingency for risk, so on work that turns out straightforward you will pay more than you would have on day rate. What you buy for that premium is budget certainty and a supplier who carries the cost of anything taking longer than expected. Whether that trade is worth it depends on how much a budget overrun would cost you internally.
What should I do if the supplier says the work is bigger than quoted?
Ask what specifically is different from what was assumed, and whether it was reasonably knowable at quoting time. If the data turned out materially different from what they were shown, that is a genuine change. If they simply underestimated work they could have scoped, that is their commercial risk under a fixed price, and a good supplier absorbs it.
How do I avoid endless change requests?
Agree the definitions before the build rather than during it, since most changes come from discovering that "customer" or "revenue" meant something different than assumed. Then buy something small from the supplier first, to see how they behave commercially before you commit to something large.
Should ongoing support be fixed or hourly?
Neither works well as hourly. Hourly support creates a reason to hesitate before asking for help, which is the opposite of what support is for, and hour-banks encourage stockpiling and end-of-period panic. A rolling monthly arrangement with a clear scope tends to serve both sides better, and the questions to ask before signing one cover what to check.