The Automation Brief an Operator Should Write First
Seven things written down before the first vendor call. They make three proposals comparable and cost an afternoon to produce.
Why the brief comes first
Ask three vendors to quote for automating your order intake and you will get three prices for three different jobs. Not because anyone is being dishonest, but because each of them heard a different half-hour of description and filled the gaps with what they usually build.
The brief closes the gaps on your side of the table. It costs an afternoon, and it turns a set of incomparable proposals into a set of comparable ones.
The seven lines
One. The process, in one sentence. From what event to what outcome. "A rental enquiry arrives by phone, WhatsApp or the web form, and ends when the equipment is on a van with a signed agreement." If it takes three sentences, you have two processes.
Two. Volume per week, from a system. Not a typical day - the count for a period long enough to contain a bad week. This is the line that moves a quote most, and the one most often supplied from memory. What to automate first is decided by this number more than by anything else in the document.
Three. Who touches it, and for how long. Roles, not names, and minutes per item rather than a share of a day. Two people at twenty minutes is a different problem from six people at four minutes.
Four. What finished means. The state the process is in when it is done, written so a stranger could check it. This is the acceptance test, and writing it now prevents the version where finished means whatever the delivered system happens to do.
Five. What must never be automated. Refunds above a threshold, anything clinical, anything that goes to a regulator, anything a customer would experience as a machine deciding their case. This is the line vendors leave out and clients turn out to have meant all along.
Six. What it costs now. Hours per week at a loaded hourly cost, plus the rework you can name. Not the whole company - this process. The arithmetic for producing a number you can defend is in the four questions that break most ROI claims.
Seven. What evidence you will accept. The measurement that will settle it in ninety days, and where the number will come from. Agreeing it now is what stops the after-the-fact comparison against a remembered baseline.
What it changes on the other side
A vendor given this document quotes the same job as every other vendor given it. That is the whole point. Where they still differ - approach, sequence, what they refuse - the differences are real ones you can judge, and reading the quote itself becomes a much shorter exercise.
It also tells you something about the vendor. An experienced one will add to line five and argue with line two. One who treats the brief as scope being taken away is telling you where their margin lives.
The version where you do not build
Sometimes the sixth line comes out small. That is a result, not a wasted afternoon: it arrives before a build rather than after one, and the same page is still a written description of one process, its volume and its real cost. The next attempt starts from it, whoever runs that attempt and whenever it happens.
01Why write the brief before talking to vendors rather than after?+
Because a proposal written from a conversation describes the vendor's understanding of what you said, and three vendors heard three different things. That makes the quotes incomparable in a way that is invisible: they are all reasonable, all confident, and all pricing a slightly different job. Writing the brief first fixes the scope on your side of the table, so the differences that remain between proposals are differences in approach and price rather than in what is being built. It also moves the scoping decision away from the party whose income scales with the scope.
02How precise does the volume number need to be?+
Precise enough to have come from a system rather than a memory, and covering a window long enough to include a bad week. An operator asked how many orders they process will answer with a typical day, which is almost always the number they wish were typical. Pulling the count from whatever system already records it takes twenty minutes and routinely moves the answer by a third in one direction or the other. It is the item that changes a quote most, because volume decides whether a rules-based approach is enough or a model is needed at the front door.
03What belongs in the line about what must never be automated?+
The decisions where being wrong is expensive and cannot be undone quietly: a refund above a threshold, a clinical or legal judgement, anything that goes to a regulator, anything a customer would experience as a machine deciding their case. Writing them down before the build is what turns them into a specification rather than an argument during acceptance. It is also the fastest way to tell whether a vendor has done this before, because an experienced one will add to your list and an inexperienced one will treat it as scope being taken away.
04What if the brief shows the process is not worth automating?+
That is the brief working. The measured cost per week is the same number every ROI claim will later be checked against, and when it is small the honest conclusion is available immediately rather than after a build. The document does not lose its value in that case: it is a written description of one process, its volume and its true cost, which makes the next attempt - by anybody, in any year - substantially cheaper than the first one was.
Free AI Diagnostic
Fifteen minutes, no email required. It maps where your work actually goes and ranks what is worth automating first.
Start the free diagnosticStarts immediately in the browser.
- Fee
- Free
- Length
- 15 minutes
You keep the ranked list of candidates either way.
