What should an ecommerce marketing agency do each month?

Channels decay. A month with no maintenance is a month where performance quietly falls, so a portion of every month goes on work that produces no headline.

On the feed and shopping side: check disapprovals and fix the causes rather than resubmitting, check price and availability against the store, watch for products that dropped out of the feed, and review the products absorbing spend without producing orders.

On campaigns: pace budgets against the month, review search terms and add exclusions, check bidding targets against actual margin, pause what has stopped working, and make sure nothing is spending on products that are out of stock.

On creative: rotate assets before fatigue rather than after, and put new ones into test on a schedule.

On site and tracking: verify that conversion tracking still fires after any theme or app change, check page speed on the main templates, and watch the funnel ratios for a step that has slipped.

On email and text: check flow performance individually, clean the list, and review anything that is sending to people who should have been excluded.

None of that is glamorous and all of it compounds. A store where these run reliably has fewer emergencies, and the emergencies are the expensive part. This maintenance list is also a fair thing to ask any provider to show you, because it is the difference between managing an account and watching it.

Beyond maintenance, each month should complete a small amount of new work. Small and complete beats large and ongoing, because unfinished work produces no result and cannot be measured.

One substantial fix plus one smaller one is a realistic month alongside trading. Substantial might be rebuilding the product feed, restructuring campaigns around margin, rewriting the top twenty product pages, cutting steps out of the checkout, or building a replenishment flow. Smaller might be fixing a shipping threshold, adding a size guide, correcting schema on product pages, or setting up a back in stock notification.

The order should come from the plan rather than from whatever is topical. A month where the work changed direction because of an article or a new platform feature is usually a month where nothing finished.

Each piece should have a date it went live, recorded somewhere, because that is what makes measurement possible. Without a change log, every number becomes an argument.

And changes should go out one at a time where possible. Two changes to the same funnel step in the same week means neither can be judged.

One useful discipline: at the end of the month, the question is not what did we work on, it is what is now live that was not live before. If the honest answer is nothing, the month produced analysis rather than progress, and two of those in a row is a conversation worth having.

The monthly report should be short, comparable with last month, and finish with what happens next.

It should lead with the four numbers behind revenue: visitors, conversion rate, average order value and repeat purchases, each against the previous period and the same period last year. Then revenue and orders, then marketing cost as a share of revenue or contribution after cost of goods and marketing, so profitability is never invisible.

Under that, channel performance with spend, orders and cost per order, and a note on what changed. Then the work: what went live this month, what it was expected to move, and what actually happened.

Then the decisions. Three or four lines: what we are doing next month, what we are stopping, and what we need from you. A report with no decisions in it is a status update.

What does not help: dozens of platform screenshots, metrics nobody acts on, several platforms' revenue figures added together as though they were company revenue, and commentary that explains every fall as seasonality.

The report should also be honest about what did not work. Tests that failed are useful information, and a provider who only reports wins is either not testing or not telling you.

And it should be readable by someone who was not in the meetings, because at some point it will be.

One scheduled call a month is usually enough when the reporting is good, with quicker contact when something breaks. The call should be about decisions rather than about reading the report aloud, which means the report arrives beforehand.

What matters more than the meeting rhythm is the written trail. Every change to campaigns, the site, the feed or the flows recorded with a date and a reason. Six months later, that log is the only way to explain why a number moved, and it is also what makes a handover possible if anything changes on either side.

Alongside that, keep access clean. Accounts in your name, permissions granted rather than owned, billing on your own cards, and documentation in a place you control.

Then there are things that should not be happening every month. Restructuring the account is not a monthly activity, and repeated rebuilds usually mean nobody is measuring. Chasing a new channel every month spreads a budget too thin to learn anything. Changing bidding strategy or targets constantly prevents any of them from settling. Redesigning pages that are converting well is churn, not progress.

And a monthly discount, sent to the whole list, is the most expensive habit a store can pick up, because it trains customers to wait.

The simple test at the end of each month is whether you could name what changed, why it changed, and what it did. If you can, the month worked. If you cannot, ask for the change log before you ask for anything else.