The Ecomm Analyst

Most brands under a few million in revenue do not have anyone whose job is measurement. The founder is doing it at night, or a generalist marketer is squeezing it in between launching campaigns and answering support tickets. That constraint should shape the entire framework. If your measurement setup cannot be run in about an hour a week by someone who has four other jobs, it will quietly stop being run, and then you are flying blind with a very expensive dashboard nobody opens.

The first mistake I see is copying enterprise. People read how a brand with a five-person data team operates and try to reproduce it with no team. You do not need forty metrics, a warehouse, and a BI tool to make good decisions at this size. You need a small number of figures you check on a fixed cadence and a clear idea of what each one makes you do.

Three layers, not forty metrics

I think about measurement in three layers, and most weeks I only touch the top one. The planning layer is the blended view of the whole business, revenue against total spend, checked weekly. The channel layer is the platform numbers, used directionally to decide what to scale or cut inside each ad account. The diagnostic layer is everything else, the deeper cuts you only pull when something in the top two layers looks wrong. Keeping these separate stops you from drowning. Most weeks the business is fine, you glance at the planning layer, and you move on.

Pick one source of truth for dollars

Before any of this works you have to decide what counts as revenue and stick to it. For almost everyone that is Shopify, net of refunds. Your ad platforms will each claim a different revenue number and none of them will match Shopify, because they are counting attributed conversions on their own windows, not money in the bank. Pick Shopify as the dollar truth, pull spend from each platform, and never let a platform self-reported revenue become the number you plan against.

The weekly hour

My weekly ritual is short. Total spend for the week. Blended efficiency, which is real revenue divided by total spend. The split between new and returning customer revenue, because a healthy blended number can hide weak acquisition when repeat purchases are carrying it. And the two or three biggest movers, a campaign that doubled, a channel that fell off a cliff. That is it. Fifteen minutes if the data is in one place, an hour if you are still copying between tabs.

Zoom out once a month

The weekly view catches fires. The monthly view is where you check whether the trend is going the right way, whether blended efficiency is improving as spend grows or you are buying more revenue at worse and worse rates. A brand spending 30,000 a month at a 3.0 blended that climbs to 60,000 a month at a 2.0 is growing the top line and quietly destroying its unit economics. You only see that by zooming out past the week.

Attach a decision to every number

A metric with no decision attached to it is decoration. Before you add anything to your weekly view, write down the action a bad reading triggers. If blended efficiency drops below your floor, you pull back spend on the worst channel. If new customer revenue share falls, you investigate acquisition before celebrating the blended number. If you cannot name the decision, do not track the metric yet. This single rule kills most dashboard bloat.

Do not over-instrument early

There is a strong temptation to wire up event tracking, server-side tagging, and cohort dashboards before you have the volume to read them. At a few hundred orders a month, most of what you will see is noise, and you will make confident decisions off random variation. Instrument when you have enough volume that the patterns are stable, not before. Until then, the blended view and a clear head will beat a noisy dashboard every time.

The framework is supposed to grow with you. Start with the boring version, the weekly hour and three numbers you actually act on. When you hire your first analyst or cross into real volume, you will have clean habits and a clear sense of which questions matter, which is worth far more than a dashboard you built too early and never trusted.