The questions worth asking.

That is the objection behind most polite refusals, and it is a fair one. Your organisation spends six hours in total across the review: four conversations of about an hour, and one data request. There is no software to install, no agent in your estate, no procurement exercise and no project team. Everything else happens off to the side and comes back as a finished report.

Read access to billing and cost reports, or simply the exports your team already produces. No production access, no customer or member data, nothing personal. Work is done under NDA. Analysis stays within your tenancy or in UK held storage as you prefer, and nothing leaves other than the report, which contains no personal data. We are happy to complete your third party onboarding and supplier risk assessment before any access is granted.

Keep them doing it. One structural point is worth knowing: a partner whose revenue is tied to the size of your estate has no commercial incentive to shrink it. That is not a criticism, it is how the contract works. An independent review gives you a second opinion with no stake in the answer, and the strongest outcome is taking the plan back to your partner as the work order.

They do, and the report will tell you to use them. They are sufficient at most organisations' scale and we will never recommend buying a platform you do not need. But free tools generate recommendations, and roughly half of recommendations are never actioned. The gap is not visibility, it is the process between detection and remediation. Native tools are also weakest exactly where AI spend now sits.

You may end up doing that, and the report will say so if it is warranted. But platforms in this market typically start in the tens of thousands per year, require integration, and need an internal owner with time to run them. Knowing what you are spending should not require buying something first. Establish the number, then decide. We sell none of them and take no commission from any vendor.

Technically yes, and some do. Two things usually prevent it. The people who could do it are the same people delivering the roadmap, so it slips indefinitely. And an internal estimate of your own waste is a harder document to take to a board than an external one. The plan is written so your team executes it: most first stage actions are configuration, not consultancy.

For a programme running into six figures, a large firm brings governance depth worth paying for. For establishing a number in a few weeks, you get a senior person doing the work directly rather than a partner at kickoff and a delivery team you have not met. Where a large firm is the right answer, the report will say so.

Pilots are the cheapest moment to get this right and the most expensive to leave alone. The costliest item in most estates is not a busy workload, it is a dedicated endpoint left running after a pilot finished. It bills at full rate indefinitely and draws a flat line on a cost chart, which everybody reads as healthy. A review at pilot stage also sets the attribution before there is anything to untangle.

Yes, and this is the common case. Where there is no native attribution, spend is allocated by inventorying every consumer, sampling two weeks of activity, measuring representative calls, and normalising the result against your actual invoice total. Every allocation is labelled high, medium or low confidence with the method shown. A defensible estimate you can interrogate is worth more than a precise figure nobody trusts.

It is usually larger than the finance view suggests, because it arrives in seven or eight places and only one looks like an AI invoice. Video generation on a marketing card, document processing billed per page, AI seats inside software subscriptions. If the total genuinely is small, the report says so, and independent evidence of that has its own value with a board asking the question.

Both, and that is precisely the problem. The budget usually sits with finance and the data sits with engineering, so neither can answer alone. Almost every stalled AI cost initiative stalls on that gap rather than on tooling. The readout is deliberately run with both in the room.

No, and be wary of anyone who does before seeing your data. What is guaranteed is a defensible number: every figure traces to a resource, a utilisation profile, a usage sample or a licence count your own team can open up and check.

The report is built so your own team can execute the plan without further help. If you want ongoing support afterwards it is available, but each review is priced to stand alone and you will feel no pull.

You have a 90-day plan with dated actions and named owners. Most of the first month needs configuration rather than consultancy. If you would like a monthly review to keep it honest, that exists, but it is only ever offered to organisations that have completed a review and never sold cold.