Agency Automation ROI: How to Measure It

“What’s the ROI on this automation” is the question every agency owner asks before signing off on a build, and it is a fair question. It is also one that has an honest answer and a dishonest one, and the dishonest one is far more common because it is easier to say in a sales call.

The dishonest version: hours saved times an hourly rate. Someone counts fifteen minutes off a weekly report, multiplies it by fifty-two weeks and a blended rate, and presents a number that looks precise enough to trust. It isn’t. That formula is wrong in both directions at once, and knowing why is most of what you need to measure this honestly.

Why is hours saved times a rate the wrong formula?

Because it counts one side of the ledger and skips the other two.

It undercounts the cost side, because it only ever prices the hours the automation removes, never the hours it took to build it or the hours someone spends keeping it working after the demo is over. And it undercounts the value side just as badly, because the biggest returns from a good automation are rarely measured in hours at all.

What does the formula miss on the cost side?

Two things, and both get left out of the pitch because leaving them out makes the number bigger.

• Build cost. Someone has to map the process, scope the automation, and build it. That is real money whether it is your own team’s time or a consultant’s invoice, and it does not disappear because the automation is now running.
• Maintenance cost. An automation is software with fewer people watching it. APIs change, a field in your CRM gets renamed, a tool ships an update, and the workflow either breaks loudly or, worse, keeps running and produces the wrong output silently. Someone has to own that. If nobody does, the automation eventually costs more than the manual process it replaced, in the exact hours nobody is counting because nobody assigned the ownership in the first place.

Skip either line and the ROI number is fiction dressed as arithmetic.

What does the formula miss on the value side?

The larger point, and the one that gets lost because it does not fit neatly into a spreadsheet cell. A lot of what a good automation buys an agency is not time. It is:

• Errors that no longer reach a client. A rule that pulls the same figure the same way every time cannot make the copy-paste mistake a tired person makes at the end of a long day. The value of that is not the ten minutes it takes to redo the report. It is the client relationship the wrong number would have cost.
• Delays that no longer happen. Work that used to wait for someone to notice it, remember it, or get back from lunch now starts the moment the trigger fires. The saved time is real, but the removed delay is the bigger win, because delay compounds across a project in ways an hourly rate does not capture.
• Consistency. The step that used to happen when someone remembered now happens every time. That is not a time saving at all in the strict sense. It is the difference between a process that exists on paper and a process that actually runs, and it is worth more than the minutes involved suggest.

None of that has a defensible dollar figure attached to it industry-wide, and anyone who hands you one is guessing on your behalf. That does not mean it is not real. It means it has to be measured inside your own agency, not imported from somewhere else.

So what is the honest way to measure it?

Four steps, in this order, and none of them produce a percentage you can quote to a partner meeting before you have run them.

1. Baseline before you build

Before anyone writes a workflow, write down what is actually happening now: how often the specific failure occurs, how it gets caught, and what it costs when it is not caught. Not a vague sense that reporting takes too long. A real count: how many times a report went out with the wrong figure last quarter, how many onboardings skipped this exact step last month, how many days a handoff sits untouched right now on average. Whatever the true numbers are in your agency, write them down before anything changes them. Mapping the delivery week honestly is exactly the work covered in the piece on what an agency automation audit covers, and it is the step every ROI conversation skips, because it is slower than asking a vendor for a number.

2. Count the full cost, not the invoice

Add build and maintenance together, not just the line item you paid once. If you hire this out, my own ladder is one example of what that looks like priced honestly: a fixed-scope build is quoted per workflow, depending on what is being built, and ongoing ownership, when a client wants someone else to hold that responsibility rather than assign it internally, starts at $3,000 a month. Whatever route you take, in-house or out, both numbers exist. The formula that only prices the build and calls maintenance free is the formula that makes every automation look profitable on the day it ships and none of them look profitable eighteen months later, when the true cost finally shows up.

3. Measure the specific failure the automation was built to remove

Not a general productivity feeling. The exact number you baselined in step one, measured the same way, after the automation has been running long enough to be normal rather than novel. If the baseline was “this report went out wrong four times last quarter,” the measurement is how many times it went out wrong this quarter, not how much faster the report gets assembled. Speed is a side effect. The failure you targeted is the actual return.

4. Review at a fixed point, not whenever it feels convenient

Ninety days is long enough for the automation to have run through a normal cycle and short enough that the review still happens. Put it on the calendar when the automation ships, not after someone asks why nobody has looked at it. At the review, compare the number from step three against the baseline from step one. That comparison, your own before and after, is the ROI. There is no other honest one.

Why won’t a percentage benchmark work across agencies?

Because the variables that decide it are different in every agency: what the failure was costing before, how much judgment the step required, how reliably the old process ran to begin with, what the automation actually cost to build and keep alive. Multiply those together across a handful of agencies and you get a wide enough spread that any single percentage claimed as typical is doing more marketing than measurement. That is true of a lot of what gets published on ordering automation work, too, which is a separate question with its own logic covered in the piece on what to automate first. This one is only about how you measure what you already built.

A benchmark also invites the wrong comparison. The question worth asking is never “is my automation as good as the industry average.” It is “did the specific thing I built fix the specific thing I measured, at a cost that made sense once I counted all of it.” The first question flatters whoever quotes the benchmark. The second is the only one that tells you whether to build the next one.

What does this mean for the automation you are about to build?

Write the baseline down before anything ships, not after. Price the maintenance alongside the build, not as a surprise a year later. Measure the failure you targeted, not a proxy that happens to move. Put ninety days on the calendar the day the automation goes live. If you do those four things, you will know whether the build was worth it in terms your own numbers can defend, rather than a percentage borrowed from somewhere it never applied.

That is also most of what a proper automation consulting engagement should be doing on your behalf: building the baseline into the process from day one rather than trying to reconstruct it after the fact, when nobody remembers what the number used to be.

Where to start

If you have not yet built anything and want the baseline done properly before the first workflow ships, that is the whole point of the Agency Ops Audit: $1,500 fixed fee, two weeks, a written 90-day roadmap that names the specific failures worth fixing and the order to fix them in, plus a 60-minute readout call. I have spent twelve years running delivery operations for marketing agencies, across 92 completed engagements, inside ClickUp, Asana, Monday.com, Airtable, and Notion, and the number I trust after all of it is never a percentage handed to me before the work started. It is the one your own before-and-after produces.