Automation as a New Revenue Stream: Three Kinds of Income and When Each One Arrives

An automation revenue stream is the income an MSP earns from designing, building and running automations for its clients, and it is really three streams, not one. They arrive at different speeds, they have different margins, and an MSP that treats them as the same thing usually ends up with the least interesting of them.

The first stream: project fees

This is the one everybody sees. A client has a process that hurts, you design and build the fix, you invoice for it. It arrives first, usually within a quarter of starting, and it is lumpy. A good month is followed by a quiet one, because every project ends.

Project income is worth having, but it has a ceiling that sits at the number of projects your engineers can deliver. It is also the stream most exposed to the pricing problem in the post on selling outcomes instead of hours: quote it by the hour and it shrinks as you improve.

The second stream: the running

The second is what happens after go-live, and it is the one that turns a project business into a recurring one. An automation that runs needs somebody to look at it in the first month, to own it, and to fix it when a connection breaks or a field changes. Clients will pay for that, if you have written it down as a service.

Two things have to be true. The running has to be a defined line in the retainer, with the automations it covers named. And you have to be honest about what you are watching. A workflow built in abi. Agent Studio runs in the client's own n8n, and abi. does not monitor it. If you want the running to be a revenue line, you are the one who watches. The post on the first month after an automation goes live sets out what to check, and the post on the retainer that includes AI covers how much to set aside for the running cost.

Recurring income takes longer to build, because it only starts when the first automation goes live and it grows one client at a time. It is also the stream that makes the business worth more, because it renews without a sales conversation.

The third stream: the next thing

The third comes from the conversation you have once the first two are running. A client with three automations live is a client whose business you now understand better than their last supplier did. Their map shows the next process, the one that costs most hours, and you are the obvious person to fix it.

This is the stream that costs least to win. There is no pitch, only a quarterly review with a ranked list of what is next and what each item is worth. We described what that meeting looks like in the post on the virtual CEO conversation.

It is also the stream most MSPs never collect, because nobody had the list.

What to expect and when

In rough order: project fees in the first quarter, the running from the second, the next thing from the third or fourth. Do not plan a year around the third stream before you have the first two.

It also helps not to put numbers on any of this that you have not earned. Each MSP's mix will differ, and the honest figure for your own business is the one you get from your first three clients, not from anyone's average.

Where the margin is

Projects carry your engineers' time and the discovery work. The running carries a smaller amount of recurring effort, and so usually has the better margin as a share of the fee, once the checks are routine. The third stream has the least selling cost of the three.

That is the argument for starting the running as soon as the first automation is live, instead of treating it as an afterthought. It is also why the service line post argues for writing the whole offer down before you sell any of it.

When this is the wrong move

If your engineers are fully booked on managed services and you have no capacity to build, the first stream will hurt the stream you already have. Automation is a poor reason to let existing clients wait. The honest answer may be to start with one client, learn what it costs you in time, and decide from there.

It is also not right for every client. A very small client with a handful of processes that rarely change has little to automate. Say so. A client who is told that, and then needs you in a year, remembers it.

A last practical point. Write down which stream each client is in. A client on one automation and no running agreement is a project client, and the right next step is to offer the thirty-day review and make it a line in the retainer. A client with three automations live and a quarterly meeting is ready for the third stream. The list takes ten minutes to make, and it tells you where the next conversation should start, which is more than most pipelines do.