The five things we look at before quoting an automation

Five questions, in order

Most of the difference between an automation worth building and one that is not is visible before any code is written, and none of it is visible in the technology. These are the five things we look at, and the order they come in matters as much as the list does.

1. where the process is actually written
2. who owns the data it runs on
3. who is allowed to decide
4. what happens to exceptions today
5. what the work costs now

If the first two have no answer, everything after them is guesswork with a number attached to it.

Where the process lives

A process that exists only as habit can still be automated, but the first phase is writing it down, and that phase is longer than anybody expects. The rules turn out to disagree between the two people who run it, and reconciling them is a business decision rather than a technical one.

Who owns the data

Not who stores it — who is allowed to grant access to it, and how long that takes. An automation blocked for six weeks on an approval nobody knew was needed is the most common way a small project becomes a long one.

Who is allowed to decide

Every automation eventually asks a question the specification did not answer, and the project moves at the speed of whoever can answer it.

One named person with the authority to settle a rule is worth more than any amount of documentation. A committee is a schedule risk, and it is fair to price it as one.

What happens to exceptions

How the awkward cases are handled today is the best available description of what the system will have to do, because the exceptions are where the real rules are kept. If the honest answer is that one person handles them by knowing, then that person is the specification.

What the work costs now

A rough figure is enough: hours a week, roughly what those hours are worth, and how often the work has to be redone. Without it there is no way to tell whether a build repays its price, and we would rather say so at the quote than discover it at the review.

Volume, and how varied

A hundred cases a day that are nearly identical is a different piece of work from twenty a day that are all different. Variety, not volume, is what sets the cost, and it is the thing most often left out of a brief.

Why we sometimes say no

Because the honest answer is sometimes that the process should be simplified rather than automated, or that the data will not carry it yet. Quoting for work that will not repay itself is a slow way of losing a client. If you want that conversation, it starts here.