Which reports should an ecommerce store check weekly?

Revenue and orders against the same week last year and the previous week, from your ecommerce platform rather than from analytics. Average order value alongside them, because revenue moving without orders moving is a different situation from both moving together.

Total marketing spend across every channel, and revenue divided by it. That single ratio catches problems that channel-level reports hide, since a channel improving while the ratio worsens means credit moved rather than money arriving.

New customers and what they cost. Total spend divided by new customer count gives you a blended acquisition cost that no platform will calculate for you and that governs whether the business is growing or churning in place.

Conversion rate split by device and by the two or three sources that matter most. A blended rate moves with the traffic mix, so the split is what tells you whether something changed on the site.

And revenue from email and automated flows, kept separate, because flows are the part that should be growing quietly in the background while campaigns fluctuate.

Product feed and Merchant Center diagnostics first. Disapprovals accumulate silently, and a fortnight of a hero product being unavailable in Shopping is expensive and entirely preventable.

Conversion tracking second. Compare the number of purchase events your analytics recorded against the orders your platform created. A sudden divergence means a tag broke, and the sooner you catch it the less bidding damage it does.

Stock on your top sellers third. Products going out of stock affect advertising performance, feed eligibility and organic listings, and the reporting almost never surfaces it as the cause.

Site errors and speed fourth. A quick look at whether product pages still load quickly on mobile, and whether anything is throwing errors after last week's app installation or theme change, catches the class of problem that shows up in revenue a week later.

None of these takes long. Their value is that each one has a history of costing stores real money while looking like a marketing problem.

The value of a weekly review comes from familiarity. When the report looks the same each time, an unusual number is obvious without anyone having to analyse anything, and that recognition is faster than any alerting system you could build.

So fix the layout, fix the comparisons, and resist adding a new chart every time somebody asks a question. Questions that arise from the weekly review should be answered in a separate piece of analysis, not by permanently enlarging the review.

Write one line of commentary each week about what changed and why. Over a quarter that becomes a record that explains the shape of the business far better than any dashboard, and it stops the team relitigating the same movements.

And keep a change log next to it: campaign changes, theme deployments, app installations, price changes, feed changes. Most unexplained weekly movements turn out to be explained by something in that list.

Keyword rankings. They move daily for reasons unrelated to your work, and a weekly look invites changes to pages that were fine. Organic clicks and impressions monthly, with rankings checked only when investigating something specific.

Small test results. A test read weekly is a test being read repeatedly, which is the fastest way to call a false winner. Look once, at the end, on the date you set.

Lifetime value and cohort retention. These move slowly and need months of data to say anything, so a weekly view shows noise. Review them quarterly, by cohort.

And anything you would not act on. If a number has never once caused a decision, it is decoration, and removing it makes the numbers that matter easier to see. The test for every line in the report is what you would do differently if it moved.