Performance review template for Ecommerce

The performance review with the columns a store needs: every line reviewed by catalogue segment, results read on kept orders rather than orders, and causes that include the three things a platform's reporting cannot see — stock, margin and returns.

Everything else is the general review, unchanged. Planned from the plan, variance separated from cause, cause held to evidence, one decision with one owner and a date.

Run it at the end of every season and every promotion, and weekly during a flight. A store's periods are short and lumpy, so a monthly rhythm reviews an average of two weeks that behaved completely differently.

Run it also once the returns from a period have landed, which is usually several weeks after the spend.

A store's campaigns usually span more of the range than their names suggest, and the average across a campaign hides the only thing worth knowing.

The same prospecting campaign can carry a strong result on best sellers and a poor one on the long tail, and the blend will look acceptable. A decision taken on the blend — more budget, less budget, new creative — is then wrong for both halves at once.

So the unit of review is the catalogue segment, matching the unit in the plan. Where the account is not structured that way, the review is still written that way, from whatever breakdown the reporting can produce. The structural change usually follows the first review that shows the gap.

An order that comes back cost money and produced nothing, and in several categories the proportion that comes back is large enough to change every conclusion.

Where that is true, the primary result column reads kept orders rather than orders, and the review has to wait for the returns to land. That means two reviews rather than one: a fast review during and just after the flight, on orders, to make operational decisions; and a slower one several weeks later, on kept orders, that decides what the period actually produced.

Keeping them separate is important. The fast review is for changing weight this week. The slow one is what informs the next plan, and it is the one that occasionally reverses the conclusion of the first.

Two segments hitting the same cost per order have not produced the same outcome.

So the review carries the margin for every segment being judged on a cost per order, taken from the plan. It is what turns a comparison between segments into a comparison of contribution, and it stops the review approving whichever part of the range is easiest to sell.

Where a segment's margin was blank in the plan, it is blank here too, and the decision for that row is usually investigate: find the margin before the next flight, because without it the segment cannot be judged at all.

Three causes are specific to this category and none of them is visible in a platform's report.

Stock. A line that lost its best-selling variant on day nine was interrupted, not outperformed. A campaign pointed at a collection where several items went out of stock has been quietly narrowed. Both look like performance decline.

Returns. A creative or an offer that raises expectations will lift orders and lift returns further, and the effect arrives weeks later against a cohort nobody is still watching.

Delivery cut-offs. Traffic sent after the last date an order can arrive in time converts worse for reasons that have nothing to do with the ad, and the drop is sharp enough to be mistaken for fatigue.

Record all three as events with dates in the notes, and name them in the cause column where they apply.

A period containing a promotion cannot be compared with one that does not, and the weeks either side of a promotion are not ordinary weeks either.

Demand is pulled forward into a sale and taken out of the weeks after it. Auction costs move. The mix of buyers changes. A review that compares a promotional fortnight with the fortnight before it is measuring the offer, not the work.

So the period column carries the promotion window, the notes record where the edges fell, and comparisons are made like for like: this promotion against the same promotion last year, full-price weeks against full-price weeks.

Four decisions cover most rows, and naming them makes the review faster.

Change the weight — usually the response to a pacing variance where the cause is understood. Change the segment split — the response to a margin-adjusted comparison rather than a cost-per-order one. Change the creative — the response to a fatigue signal that survives the stock and delivery checks. And investigate — the response to any row where the cause is not known and the money involved justifies finding out.

Anything vaguer than those four is not a decision.

The accumulated reviews answer questions no single one can.

What each segment has actually cost to sell, across seasons. Which promotions have ever paid for themselves once returns are counted. How much demand a peak pulls forward from the weeks after it. Which creative themes have worked on which parts of the range.

Those are the four questions every store's planning meeting tries to answer from memory. A shelf of consistent reviews answers them from evidence, and it is the only asset in this process that compounds.

Seven checks. Every row is reviewed at segment level, matching the plan. Results are read on kept orders wherever returns are material, with a second review scheduled once they land. Margin is stated beside every cost-per-order comparison. Stock events, returns and delivery cut-offs are recorded as dated events. Promotional periods are compared like for like. Every decision is one of four, with an owner and a date. And the finished reviews are kept by segment and by season.