ROAS vs profit: how to judge an e-commerce campaign

1. Check what value is attributed to the ad.

The basic formula is ROAS = value attributed to advertising / advertising cost. With PLN 12,000 in revenue and PLN 3,000 in spend, you get 4, or 400%. Pay attention to the unit of presentation: 4 and 400% describe the same ratio, while 4% means something entirely different.

In an ad platform, the conversion value isn't always revenue. Google lets you assign events values that reflect business goals. Before comparing the metric with your store report, check the configuration: does the numerator contain the actual purchase value, a fixed amount per event or some other value? A mathematically correct result may be answering the wrong question.

2. Separate revenue from the amount that's left over.

A store gets identical revenue from two campaigns. The first sells a product with a high purchase cost, the second accessories with a different margin. With the same ad spend, ROAS will be equal, but the result after costs may differ. Comparing revenue alone won't capture that difference.

In your working calculation, add the cost of goods, payment fees, fulfillment costs and shipping subsidies the store actually pays. Account for returns without subtracting the same amount twice. Fixed costs and how taxes are handled need a separate, consistent approach. Flowpeek provides sales and marketing context; don't assume the report automatically includes your company's full cost accounting.

3. Calculate a working threshold on your own assumptions.

In a simplified example, out of PLN 100 of revenue, PLN 30 remains after variable costs other than advertising. Advertising can take up at most those PLN 30 for the result before fixed costs to be zero. That corresponds to a ROAS of about 3.33, because 100 / 30 = 3.33. This is the threshold for this particular model, not a universal campaign target.

With PLN 25 of spend, ROAS will be 4, but only PLN 5 will be left for fixed costs and the rest of the company's result. The example assumes the same value basis and no other costs. Before using it in a budget plan, agree on it with the person responsible for finance. Don't apply the margin of your whole catalog to a campaign selling a completely different range.

4. Check timing, currency and sales attribution rules.

Compare campaigns over periods long enough to observe the purchase. Today's cost may already be visible, while some interested people will only buy later. Purchases preceded by yesterday's clicks don't necessarily come from a new ad version launched this morning.

Check that sales and cost are in the same currency and that the chosen source actually supplies cost. Missing spend data doesn't mean free traffic or an infinitely good ROAS. When comparing with an ad platform, note the sales attribution model. In Flowpeek, also analyze unattributed orders, so you don't confuse worse source recognition with worse sales.

In the table shown, Google Ads gets PLN 40,257.05 in revenue under First Touch and PLN 34,626.90 under Last Touch. With the same campaign cost, you'll therefore get a different ROAS, even though neither spend nor the order base changed. This report doesn't include cost, so you can't calculate the return from it alone. First match the cost to the same period, currency and source, then note the model you used.

5. When results drop, find the stage where they changed.

First confirm that the difference isn't caused by an incomplete day, a cost change or a delayed import. Then check the number of orders and the average order value. A drop in revenue can mean fewer transactions, a cheaper mix of products or both at once. Each leads to a different action.

Next, look at traffic sources, products and the funnel. If the number of visits fell, focus on acquisition. If interest in a product holds up but fewer people move on to the cart, check the offer and the product page. When the problem appears at checkout, review shipping costs, payments and possible errors. The report marks out the area to check; it doesn't prove the cause on its own.

• Traffic: is the campaign bringing a similar number of visits?
• Offer: have the products sold, prices or availability changed?
• Purchase: at which stage of the journey does the difference show up?
• Measurement: are costs, sources and orders up to date?

6. Consider the scale and certainty of the observation.

A campaign with one expensive order can have an impressive ROAS and very little evidence of a repeatable result. Always write down the number of transactions and the spend next to the metric. Compare similar days of the week and note promotions and stock shortages. Don't create a hard budget rule based on a single low-traffic day.

A decision sheet should contain the observation, the hypothesis, the proposed change and a review date. For example: "Average order value dropped and the share of a cheaper product is growing; we'll check the product mix in the creatives." That way the conversation about ROAS ends with an action you can evaluate. You can't determine the causal impact of advertising from attribution alone; for that question, you need a properly designed comparison or experiment.