Keep a logbook: a shared spreadsheet where you record every event that could move a number, the same day it happens. Date, what happened, where, expected effect, who did it. Five columns, two minutes per row. It's the cheapest tool in this playbook, and the one whose absence costs the most.
The benchmark
Try this: ask your team what happened on the site the second week of March. If nobody knows, your year-over-year comparisons rely on everyone's memory.
We've seen this from the other side of the table.
At a clothing brand, a grid test had won by a wide margin, but the founder credited the overall increase to four things that happened at the same time: the grid, a new international payment provider, a filter redesign, and higher-quality traffic. They were right, and that was the problem: without dates, there was no way to isolate what had made the difference.
The table below is the one we opened the next day. Copy it, then rewrite it in your own words.
| Date | What happened | Where | Expected effect | Who |
|---|---|---|---|---|
| 2026-09-21 | Shopify changes how sessions are counted | Reports | Conversion rate not comparable before/after | Shopify |
| 2026-10-03 | New free shipping threshold, €80 instead of €60 | Cart | Average order value up, watch conversion | (first name) |
| 2026-10-10 | Nothing | Site | None | (first name) |
The rule
If it could move a number, it goes in the logbook before the end of the day: promotion, release, campaign, stockout, shipping rate change, tracking change.
The steps
- The spreadsheet. Create it and give edit access to the whole team and to any agencies.
- The review. Every week, when you do the report.
- The annotations. Copy major events into your dashboards. In GA4, annotations are added by right-clicking the curve of a report, with a title of up to 60 characters and one color per type of event. You need the Analyst role.
- The read before any comparison. Before any year-over-year comparison, read the rows for the reference period.
- The line for leadership:
"We don't comment on a number without the logbook row that goes with it."
The trap
Logging only what you do. Outside events move your numbers as much as your own actions do.
At a fashion brand, a change of export carrier pushed the return rate from 25% to 45% two months later, without a single line changed on the store.
Without the carrier's date in the logbook, the increase would have been blamed on the latest product page redesign, and someone would have rolled it back. Also log the weeks when nothing happened. A number that moves when nothing has changed is information.
