A small business that goes looking for AI consulting will be offered a menu: an assessment, a roadmap, a list of use cases, a tool selection, a pilot. Each item is reasonable alone. Bought as a first engagement, most of them leave the business with a document and no change in how work gets done.
We take a narrower view of what a first engagement is for. It exists to change one piece of how your business runs, prove the change with a number, and leave behind someone who can keep it running. Everything in the scope should serve those three things, and anything that does not can wait.
A strategy deliverable has no owner after the consultant leaves. A list of use cases has no baseline, so nobody can say later whether anything improved. A tool selection happens before the work is described, so the tool shapes the workflow instead of the other way round. A pilot without an exception path works on the sample the consultant chose and meets the real mix in week three.
None of these failures is about the quality of the ideas. A business of thirty to two hundred people has plenty of ideas. What it lacks is spare capacity to absorb a long report, and so the report sits. The scope has to be small enough that the business can actually finish it.
One workflow. A named piece of work with a start and an end, such as turning an enquiry into a sent quote, or matching a supplier invoice to a receipt. Not a department, not a theme.
One baseline number. Measured before anything changes: days to send a quote, minutes per invoice, share of orders needing a correction. If it cannot be measured today, the first task is to start measuring.
One accountable owner. A person in your business, not the consultant, who will answer for the number a month after delivery.
An exception path. What happens to the cases the automation declines or gets wrong, who sees them, and how long they may wait. This is the part most pilots skip, and we cover why in our piece on writing down the exceptions rather than the happy path.
A run plan. Who watches it, what it costs each month to operate, and what triggers a review. A workflow without a run plan is an experiment that ends when attention moves on.
If a proposal is missing any of the five, it is not yet a first engagement. It is a conversation about one.
Refuse a strategy deck as the main deliverable. Ask for the change instead, and accept a short written decision about what to do next as a by-product.
Refuse a tool recommendation made before the workflow is described. The workflow decides which category of tool is needed, and the category usually decides the shortlist.
Refuse a pilot with no baseline. Without one, a good result and a bad result look identical, and the business will argue from impressions.
Refuse any scope in which the consultant is the only person who understands the result. If the explanation only works with the consultant in the room, the business has bought dependence, not capability.
A larger company needs a program office to change how a workflow runs. A small business needs one person with the authority to say yes, which is usually the owner or the operations lead, and a week of attention. The people who do the work sit within shouting distance of the person who decides, so a change that would take a quarter elsewhere can be tried in a month.
Small data is also not a problem for the first engagement. A baseline built from forty quotes or sixty invoices is a thin sample, and it is still far better than no sample. The business should not wait for a data project before measuring one workflow. It should measure the workflow in its current, untidy state, because that state is what the change has to survive.
A first engagement is easier to price than it looks, because the scope is small. The honest structure has two phases with a decision between them. The first measures the workflow and designs the change, and ends with a fixed price for the second. The second builds and runs it for a trial period against the baseline.
The estimate for the second phase should show the lines a build always has, and the lines it often forgets. We set those out in what an agent costs to build and run: the build, the operating cost, the cost of exceptions, and the cost of keeping it correct as your systems change. A first engagement quote that shows only the build is a quote for half the work.
Imagine a 35-person services firm where a quote takes an average of four working days to leave the building. The partners suspect that is slow and cannot say by how much, or why.
The first phase measures. Over two weeks the firm finds that the median quote leaves in two days, but one in five waits more than a week, and the wait is almost always for a missing scope detail that nobody asked the client about at the first call. The baseline is plain: a four-day average, a one-in-five long tail.
The change is narrow. The first call gets a short list of questions that must be answered before a quote can start, and a draft quote is assembled from the answers and the firm's past quotes for a person to check and send. The exception path is a list of quotes that could not be drafted, each with an owner and a clock. The owner is the partner who runs sales operations, and the run plan is a monthly look at the four-day number and the exception list.
After the trial period, the average is two and a half days and the long tail is gone, or it is not, and the firm knows which. Either result is useful. The business has a measured number, an owner and a decision, which is more than a roadmap provides.
Pick the one piece of work in your business that people complain about most and describe it in a page: where it starts, where it ends, who touches it, and how you would know it went well. Then try to put a number on it from last month's records. If you can, you have the baseline and half the scope. If you cannot, you have found the first piece of work worth doing, and it needs no consultant.
We begin with a no-risk discovery. We measure one workflow, design the change, and show you the baseline and the plan, and you pay for the discovery only if you decide to go ahead. The build that follows carries a guaranteed estimate: if we estimate low, we absorb the difference. After go-live, Managed Technical Operations keeps the workflow watched and the exception list worked, so the owner you name has someone beside them.
A good first engagement ends with one number that moved and one person who knows why.
In a no-risk discovery we look at your CRM, finance and operations systems and who owns each part, and show what a connected system would do differently. You pay only if you proceed. Or see how we approach it.
More on the same problem:
"Something isn't working." "We're outgrowing our tools." "We're afraid to make the wrong move." No-Risk Discovery is a short, practical conversation that gets you clarity before you commit to anything big. We'll tell you if we're a fit. If we're not, we'll tell you that too.
Say who you are and ask. A person replies within one business day.
