Before Mavrin Labs builds anything, you and the team agree four things in writing: where the work stands today, what should be different, how you will read the outcome, and the date you will look at it together. No build starts without a number to beat.
The written agreement is what turns a technology project into something you can judge, and the page is for owners and nonprofit leaders who need a result they can check, such as the time saved by automation.
Key takeaways from this page:
- Four parts go in writing before any build: a baseline, a target, one measure and a review date.
- The measure is simple enough for you to take the reading yourself.
- Mavrin Labs commits to the method and the measurement, not to a number promised in advance.
Why do projects end without a verdict?
Projects end without a verdict, in Mavrin Labs' view, because nobody wrote down what success would look like while there was still time to measure it. The software ships, the invoice clears, everyone moves on, and later nobody can say whether it paid off. Nobody captured the figures that would answer it.
The cause is rarely bad faith. Somebody has to take a baseline before the build, sitting with the person who does the work and timing it. Once the new system is running, the old week has gone and no amount of analysis brings it back.
How a result is agreed
The agreement has four parts, baseline, target, measure and review, and all four go into the scope before any build begins.
-
Baseline, step 1
The team times the work as it runs today and counts how often it runs, with the person who does it, before anything changes.
-
Target, step 2
You set what should be different, in hours back each week or in revenue, in words your own team would recognize.
-
Measure, step 3
You both agree one way of reading the outcome, simple enough that you can take the reading yourself, and who takes it.
-
Review, step 4
You fix the date you will look at the result together, written into the scope before the build begins.
What counts as a measurable result?
A measurable result is a change you could notice without a report. Examples are the time a task takes each week, time from first enquiry to paid invoice, the share of orders that need a human correction, revenue from customers who come back.
Each one can be read the same way before and after, which is the only property that matters.
The table below sets out four kinds of measure and how each one is read.
| Measure | How it is read | Typical unit |
|---|---|---|
| Hours back | The same task, timed the same way, before the build and on the review date | Hours a week |
| Elapsed time | How long work waits between the stages of a process that crosses several people | Hours or days |
| Error and exception rate | How often an item needs a person to fix it before it can carry on | Share of items |
| Revenue held or repeated | Money that arrives because a follow-up or an invoice went out when it should have | Revenue in dollars |
To set a dollar value beside the hours, the guide to the public wage for an hour of office work shows the published figure for Sarasota and Manatee counties. For reference, the Bureau of Labor Statistics put the mean wage for office and administrative support work in the North Port-Bradenton-Sarasota area at $23.98 an hour in May 2025, across 39,490 jobs (BLS wage estimates by area, checked Wednesday, November 18, 2026). A firm trying to automate client onboarding for appointment-led work might instead count the days from a first call to a first visit. The automation being measured is described on the page about AI automation that gives staff hours back, and how an in-person project runs covers a local start.
A number that nobody can take twice is not a measure. If a target cannot survive that test, Mavrin Labs says so during the first step rather than after the build.
A worked example of a measurable result
What follows is an illustration written for this page. It names no client and contains no figures. A firm processes supplier documents by hand and wants that week back.
- Before the build
- The team sits with the person who does the work, times the task, counts how often it runs, and writes the baseline into the scope with an agreed target and review date.
- On the review date
- The team times the same task the same way, and you read the difference together, including the part of the workflow that still needs a person.
What a result depends on
A result depends on things outside this arrangement as well as inside it. Some of your systems may limit what a workflow can reach, and a supplier can change an interface.
Demand, staffing and the market move for their own reasons. An estimate made before a build is an estimate, and the range around it is part of the agreement rather than a footnote to it.
The facts the firm states about itself, such as the team that takes these measurements, are listed with their basis and the date each was confirmed in the evidence register for this site.
Mavrin Labs commits to the method and the measurement, and states where the outcome depends on systems it does not control.
What happens at the review?
At the review you take the measurement again and read it together, on the date agreed before the build. Mavrin Labs brings the baseline, the target and the new reading, and says what the team reads into the difference, including the parts that did not work. Nothing about the conversation depends on who remembers what.
You then agree the next step in the same written way: more of the workflow, a different workflow, or nothing further. The build process covers the five steps the build itself follows, and the engagement terms cover how scope and the investment are settled around this agreement.
Where the reading itself has to come out of live systems rather than a hand count, reporting assembled from live systems is the service that builds it.
If you have not picked the task yet, the project finder shows how to identify AI use cases worth measuring. When you are ready to set a baseline for one task, start a conversation and describe how it runs today. What buyers usually ask before that point is answered in the FAQ.
Frequently asked questions
Who measures the result, you or us?
Mavrin Labs measures the result, and you can repeat the reading yourself, so the measure has to fit in one sentence. The team records the baseline in the first step and takes the same reading again on the review date. If your own systems already report the number, the agreement says to use your reading.
What if we have no data to set a baseline from?
No data is the usual starting point, so the team creates the baseline from scratch. The team times the work as it runs, with the person who does it, and counts how often it runs in a normal week. That timing becomes the written baseline, and it is often the first measured view of the task.
Can the target change once the work has started?
The target can change once work has started, and the change goes in writing like the original. A build sometimes uncovers a step nobody mentioned, or the business shifts what matters most. At the end of any step you can revise the target, the measure or the review date, and the review judges the latest version.
What does Mavrin Labs commit to, and what stays uncertain?
Mavrin Labs commits to the method rather than to a number. The method is a measured baseline, an agreed target, one way of reading the outcome, and a review on a fixed date whatever it shows. No builder can promise a result in advance, because parts of any outcome sit with systems, suppliers and markets nobody here controls.
What happens at the review if the target was missed?
At the review you look at the same measurement together and agree the next step from what it shows. A missed target usually has a readable cause: a step still done by hand, a system that blocks part of the workflow, or a target set too high for the baseline. The next step goes in writing too.
How does this differ from a fixed scope of work?
A fixed scope of work describes what gets built. This agreement describes what should change in your week because of it, which is a different question and the one you care about. The scope still exists and still names deliverables, and the measurable result sits beside it as the test both sides apply afterwards.
Last updated
Reviewed by Mavrin Labs on
Enquiry form
Start with one workflow
Tell Mavrin Labs what is stuck and which systems it touches. A person at the firm reads every enquiry and replies by email.
Which hours would you take back first?
Name the one workflow that eats the most of your week. Mavrin Labs reads it, and the reply comes from the people who would build the fix.
Start a conversation