What a $1M Shopify Brand Should Track

A store doing around a million dollars a year is past the point where instinct is sufficient and well before the point where complexity is justified. The main risk at this size is not measuring too little — it is buying reporting built for a business three times larger and never using it.

The four numbers

Gross margin, monthly. Net sales minus landed cost of goods, as a percentage. It should be stable. When it moves, one of three things happened: supplier prices changed, discounting deepened, or the product mix shifted. Each has a different fix and all three are worth catching early.

Contribution margin rate, weekly. After shipping, payment fees and packaging. This is the fastest-moving number in the business and the earliest warning of anything going wrong. A promotion running longer than planned or a free-shipping threshold set too low shows up here within days.

Contribution margin per order against acquisition cost. The comparison that decides whether growth funds itself. Both sides in currency, no attribution model required.

Net profit, quarterly, with your salary in it. Monthly is too noisy at this size — one bulk stock purchase distorts it entirely.

That is the whole set. Everything else is elaboration.

What to skip for now

Multi-touch attribution. It answers which channel deserves credit. At this size you probably have two paid channels and can reason about them from blended figures and the occasional holdout test. Attribution modelling is expensive and its output is a percentage split you are unlikely to act on differently.

Cohort analysis. Genuinely valuable, and mainly when you are deciding whether to bid above first-order break-even. Most stores at a million a year should be profitable on the first order, in which case the cohort curve is interesting rather than decisive.

Custom dashboards. A store this size has a fixed set of questions. Fixed reports answer them. A report builder mostly offers new ways to build the same report.

Forecasting. Useful when planning inventory across long lead times. Otherwise it converts uncertainty into a chart without reducing it.

None of these are bad tools. They solve problems that arrive later, and buying them early costs money and — more expensively — attention.

The two errors that actually cost money here

Uncosted products. Every variant without a cost is counted as pure margin, which inflates gross margin, contribution margin, break-even ROAS and your acquisition ceiling in one move. Check the count of orders shipping with no COGS attached before trusting any number above.

Founder salary at zero. In the example, $11,200 of fixed costs includes a market-rate salary. Take it out and net profit looks like $16,560 rather than $5,360 — a business that appears three times healthier than it is, and that cannot survive its founder taking a month off.

Both errors run in the same direction. Both are invisible in any report that does not deliberately look for them.

Cadence that works at this size

Weekly, ten minutes. Contribution margin rate, acquisition cost, orders. Three numbers, one line each. You are looking for direction, not precision.

Monthly, an hour. Full P&L with the prior month and the same month last year alongside. Percentages of net sales beside every value.

Quarterly, half a day. Net profit trend, product-level margin ranking, and one deliberate investigation — the worst-performing region, the discount code behaving oddly, whatever the monthly reviews flagged and nobody chased.

That rhythm catches almost everything a store this size needs to catch, and it costs about four hours a month.

The one cut worth making early

If you add anything to the four numbers, make it margin by product — and specifically, the contribution margin rate of your top five products by volume.

Those five usually carry most of the store, so drift there reaches the bottom line faster than anything further down the catalogue. It is also the cut most likely to produce a surprise, because gross margin ranks products by how well they were bought and contribution margin ranks them by how well they can be sold, and the two orderings rarely match.

The finding at this size is almost always the same: one high-volume product is heavier or bulkier than the rest, carries more of its parcel, and runs well below the store’s blended rate. It has usually been promoted hardest precisely because it sells well.

Fixing it rarely means discontinuing anything. A smaller carton, a small price rise, or excluding it from the free-shipping threshold typically recovers most of the gap — and each of those applies to every future order rather than needing to be repeated.

When this set stops being enough

Four signals, any of which means it is time to add something.

A second sales channel or market, which makes blended figures misleading because you are averaging two different businesses.

Enough SKUs that per-product margin cannot be held in your head. Usually somewhere past fifty active variants.

Retention becoming central to the model — subscriptions, consumables, anything where the first order is deliberately unprofitable. That is when cohorts stop being optional.

More than one person making spending decisions, at which point everyone needs to be reading the same number, and “the number” has to exist somewhere other than a file on one laptop.

Until one of those is true, the four numbers above will support essentially every decision you have to make, and the effort saved is better spent on product and acquisition than on reporting.