What the first 90 days with a CRO agency should look like
In a well-run conversion optimization engagement, the first test should be live inside two weeks and the first result should be readable by week six. Month one is audit, research and roadmap. Month two runs tests. Month three is where compounding starts and where you get your first honest read on whether the programme works.
Why the timeline matters more than the deck
Most agency relationships fail in the first ninety days, and almost never because of the strategy. They fail because nothing shipped. A discovery phase stretches into six weeks, the roadmap arrives beautiful and unbuilt, and by the time the first test runs the brand has already decided this was a mistake.
The fix is arithmetic rather than enthusiasm. Test results take time to arrive, so the clock has to start early. In DRIP’s experiment data the median test ran for 42 days. If the first test launches on day 45, the first result lands somewhere near day 90, which is the exact moment you were supposed to be reviewing the programme. Launch on day 12 and you review with data.
Month one: audit, research, roadmap
Weeks one and two are for finding out where the money is leaking, not for redesigning anything.
That means analytics review, funnel-by-funnel: where sessions arrive, where they stop, and what the device split does to the numbers. Mobile is usually where the damage is. Across 486 million sessions and 117 brands, DRIP found desktop converts 1.56 times higher than mobile while mobile carries 78% of the traffic. That gap alone is often the biggest single opportunity on the site.
It also means heatmaps and session recordings on the top five templates, a checkout walkthrough on a real device, and customer research. Not a survey nobody reads. Ten minutes of recordings on a product page will usually explain more than a month of dashboard staring.
The output of month one is a prioritized roadmap ranked by expected return, and the first test brief. Not a slide deck.
Month two: tests in market
By now you should be running, and you should expect most of it to be inconclusive. Optimizely’s analysis of over 127,000 experiments puts the average win rate around 12%. ConversionTeam’s audit of 2,288 tests found 19.1% reached statistical significance per test. Any agency promising you a win every month is either testing trivially safe changes or calling results early.
What you should expect in month two:
| Week | What ships | What you see |
|---|---|---|
| 5 | Test one running, test two briefed | Traffic split confirmed, no result yet |
| 6 | First readable result | A win, a loss or an inconclusive, all three reported |
| 7 | Test two live, test three briefed | Two concurrent tests where traffic allows |
| 8 | Month review | Roadmap re-ranked on what the first results taught |
Losses get reported in the same detail as wins. A loss that stops you rolling out a change that would have cost money is a return, and an agency that only shows you winners is editing.
Month three: compounding and the honest read
This is where a real programme separates from a pretty one. Test velocity rises because the research from month one is still generating hypotheses. The roadmap has been re-ranked twice. Post-purchase and subscription tests enter the queue, because that is usually where the margin sits and it is the part most brands never touch.
Across the engagements we run, one cannabis DTC brand reached 75% growth in subscription take rate, 25% in average order value and 20% in conversion inside three months, while paid spend was scaling underneath. That is what compounding looks like from the outside. From the inside it was a sequence of small tests, several of which lost.
What to hold your agency to
Ask for four things in writing before you sign: the date the first test goes live, the metric that defines a win, the reporting cadence, and what happens to losing tests. If any of those four answers is vague, the first ninety days will be vague too.
One more item to settle in week one, because it is painful to unwind later. Testing tools, analytics and tag manager all sit under your company’s email address, with the agency added as a user. It costs nothing on day one. On day two hundred, when an engagement ends, it is the difference between keeping your historical test data and starting again.
You can see how we structure the audit, research and roadmap sequence at https://www.parahgroup.com
