Short answer
Alternatives split three ways: other incrementality platforms, running match-market tests yourself, or replacing testing with a properly validated model. Platforms buy you test design and speed. Doing it yourself is very doable with an analyst. A model answers a broader question with less certainty.
Market testing measures cause directly. Everything else infers it. So the first question is not which vendor — it is whether you need a test at all.
Which of these is true for you?
Run the tests yourself.
The methods are public and the tooling is free. What you need is discipline, not software.
Watch out · Size the test before you run it. A test too small to detect anything gets reported as "the channel does not work".
Matched on how sales behave over time, not on size. The gap that opens between the two groups is the effect. Nothing is inferred.
What does a match-market test do?
- Match-market test
- A controlled test where a channel is turned up, down, or off in one set of markets while similar markets carry on unchanged. The difference in total sales is the real effect. No attribution model needed.
The hard parts are not the idea. They are matching markets, sizing the test, and reading the result. That is what a platform sells you. It is also learnable.
Running tests yourself: what goes wrong
- Tests too small to detect anything. Someone reports that as "the channel does not work." Expensive misread.
- Markets matched on size or demographics instead of how sales behave over time.
- Stopping early because the numbers look good. That turns an experiment into confirmation.
- Measuring channel-reported sales instead of total sales.
When should you build a model instead?
When the question is a whole budget, not one channel. You cannot test every channel at every spend level.
“A test tells you one true thing. A model tells you many uncertain things. Checking the second against the first is the whole job.”
How long does a test need to run?
Long enough to cover your purchase cycle and collect the sample you need. Commonly four to six weeks.
What if I have no regional sales data?
Then market testing is off the table, and you are looking at time-based holdouts or modeling. Worth fixing — regional data helps the modeling too.
Are platform lift studies trustworthy?
Useful directionally, and structurally conflicted. The platform designs the test, defines the control, and reports the result.
Can a test find my saturation point?
One test gives you one point on the curve. Several at different spend levels start to trace it.
Where we’re different
Click a claim. We’ll show our work.
Marketing mix modeling is a time-series problem. So we hide the last three to six months from the model, then make it predict them. If it cannot get under 10% error on months it never saw, we do not ship it — and you get your money back. Ask anyone else on your shortlist for that number. Most of them do not have it.
We are data scientists, not a software company. We do not sell a tool that competes with the vendors on these pages. That is why we will tell you when one of them — or a free open-source framework — is the better answer for you.
Book the two-week auditFind your wasted third
Want this run on your own numbers?
The two-week audit shows you where each of your channels sits on its own curve — what to cut, what to grow, and how much budget is sitting in the wrong place.
Book the two-week auditKeep reading