Six bucks back.So what?

An online retailer spends $1,400 on advertising. The platform credits the campaign with $8,400 in sales. Six dollars back for every dollar. Most reports would colour that number green and move on.

Should the owner spend another $1,400?

Nobody in the meeting can answer that from the number on the screen. Most of them think they can.

Six bucks of revenue isn’t six bucks of profit

Do the arithmetic the dashboard skips. Say the average order is $120 and, after product, fulfilment and other variable costs, each order contributes $30. That $8,400 is 70 orders, worth $2,100 in contribution. On paper, the campaign made $700 after the media bill.

Now the harder part. To recover $1,400 from those first orders, the campaign needs to have caused at least 47 sales that wouldn’t have happened anyway. Forty-seven times $30 is $1,410. Production costs push that hurdle higher. Profitable repeat orders could lower it, if there’s evidence customers actually come back.

So the real question isn’t “was it six times?” It’s “did the ads cause at least 47 of those 70 orders?” The dashboard doesn’t know.

Attributed isn’t caused

The platform recorded that someone clicked an ad and then bought. That’s attribution: a rule for handing out credit. It isn’t evidence the ad changed anyone’s mind.

A loyal customer who was always going to reorder can still click an ad to get to the shop. The platform takes the credit. The business pays for a sale it already had.

This isn’t a fringe objection. Google’s own help page on conversion lift separates conversions allocated by attribution rules from incremental conversions, estimated by comparing people who saw the ads with a similar group who didn’t. If the company selling the ads draws that line, a CEO should too.

My view: return on ad spend is a fine reason to investigate and a terrible reason to double a budget.

Page one is for a decision

The channel specialist genuinely needs to know which ads ran, what they cost and where people dropped out. Keep all of it. The trouble starts when their working screen becomes the CEO’s monthly report because someone found the export button.

For whoever is allocating the money, name the choice. Renew at the current level? Expand into a second region? Fix the checkout before buying more traffic? “Review performance” lets a room of intelligent people admire the same numbers and leave with completely different instructions.

A report with no recommendation on page one passes the hard part upstairs to the person with the least information. The person closest to the numbers is best placed to say what they mean. Make that their job.

Keep four things apart

Most reporting confusion comes from mixing four different questions.

Delivery: what did we buy? Spend, placements and reach tell you whether the work got to market as intended. If half the audience never had a chance to see it, find out before anyone declares the idea a dud.

Response: what did people do? Clicks, enquiries, add-to-carts. Useful, but change the form and you change what you count without changing demand.

Commercial outcome: what did we sell, and what was it worth after costs? Compare groups old enough to have reached a decision. Dividing this month’s wins, many of which started last quarter, by this month’s new leads gives you a tidy percentage about two different groups of people.

Confidence: what did we observe, what did we estimate and what else could explain it? Seasonality, a competitor’s stock problem, the email that went out the same week.

Give uncertainty a job

Back to the retailer, an illustrative one. Finance can verify the order economics. The campaign manager can check that the platform counted sales the same way as the order system, cancellations included. Neither check tells you how many of the 70 would have bought anyway.

If the next decision involves real money and enough volume, run a proper holdout test. If you’ve got a handful of orders and no realistic way to run one, don’t publish a causal estimate with a decimal point. Look for evidence pointing the same way: where new customers came from, whether total contribution moved as spend changed, what else changed at the same time. Write down the limits. Then take a capped next step the business can afford to get wrong.

What to do on Monday

1. Convert your headline return into contribution. Revenue times margin, minus media and production. Put that number on page one.

2. Work out the hurdle: how many sales the campaign must have caused to pay for itself.

3. Ask your team or agency how many of the attributed sales came from existing customers.

4. Rewrite the report template so page one carries one decision, one recommendation, a spending cap and a review date.

5. At the next meeting, give whoever presents the first two minutes to say what they want the business to do and why. Then open the dashboard.

The question was never whether the number was green. It was whether you’d spend the money again.

Sources & notes

• Google Ads Help, About conversion lift (distinguishes attributed conversions from incremental conversions measured against a control group)

• The retailer and all figures are illustrative. Contribution is before fixed costs.

Continue with The best thing about your business never leaves the building. If you want a next step, see Hum’s fractional CMO page.