Know what your headline metric is saying
Return on ad spend, or ROAS, compares revenue attributed to advertising with advertising spend. It is a useful reporting measure, but it does not automatically include product costs, fulfilment, creative production, refunds or the sales that might have happened anyway.
Imagine a campaign with €2,000 in media spend and €8,000 in attributed revenue. Its reported ROAS is 4x. This is an illustrative example, not a client result. If that revenue produces €3,200 in contribution before advertising, the remaining contribution after media is €1,200, before other campaign costs.
The example shows why revenue and profitability should be discussed separately. The business needs to know the assumptions behind a number before using it to increase a budget.
Distinguish attribution from incrementality
Attribution assigns credit for a recorded conversion. Incrementality asks what additional outcomes occurred because of the activity. Both can be useful, but they answer different questions.
Google’s Conversion Lift describes experiments comparing treatment and control groups to estimate additional conversions. Its documentation notes that the feature is not available to every account. See the measurement approach and availability.
For a business decision, the distinction matters when advertising reaches people already likely to buy. A strong attributed return may still include demand created by other activity. The appropriate response is to investigate the contribution, not automatically dismiss the channel.
Use a simple commercial measurement ladder
Start with delivery: did the campaign reach the intended audience and spend as planned? Then examine response: did people engage, visit or enquire? Next assess quality: were the enquiries relevant and did customers understand the offer?
Finally, review commercial outcomes over the appropriate sales cycle. For ecommerce, consider completed orders, refunds and contribution. For B2B services, consider qualified opportunities, progression and won business. Avoid judging a long sales cycle solely by purchases in the first week.
Keep a clear definition for each measure. An enquiry is not automatically a qualified lead, and a proposal is not revenue. Consistent definitions prevent reporting conversations from becoming arguments about what the dashboard labels mean.
Choose a test the business can support
Where volume and access allow, a controlled experiment can provide stronger evidence than a simple before-and-after comparison. Define the question, outcome, duration and comparison before the test begins. Consider whether groups are comparable and whether activity can spill between them.
A smaller business may not have enough conversions for a precise lift estimate. In that case, combine directional evidence with customer conversations and clear tracking. Record other changes, such as pricing, stock or seasonal demand, that could explain the outcome.
Do not manufacture confidence when the sample is small. It is better to say that a test is inconclusive than to use a weak result as the basis for a major spending decision.
Turn the reporting meeting into a decision
Every review should end with a specific action: continue, change the offer, improve the page, adjust the audience or gather more evidence. Write down why the team reached that decision and when it will be revisited.
Use a compact report with the commercial question, the relevant results, important caveats and the next action. A long dashboard can be available for investigation without becoming the centre of every discussion.
My approach is to connect marketing measurement to the decisions the business actually faces. ROAS can help, but its value increases when it sits alongside customer quality, contribution and a realistic assessment of causality.
What would this look like in your business?
Explore Analyse or tell me what you would like to improve.