An ad report can give a campaign credit for a sale. That does not always prove the campaign caused the sale.
Attribution is the rule used to give credit. Incrementality asks how many extra results happened because the ads ran. A conversion is an action the ad account is set to count. These terms answer different questions.
Start with the two questions
- Attribution
- Which recorded ad or marketing step should get credit for a conversion?
- Incrementality
- How many more results happened because the marketing ran?
Attribution describes a path we saw. Incrementality compares that result with what likely would have happened without the ads.
Use attribution when deciding which ad, keyword, or page to review. If you need to know whether ads created new demand or took credit for people who would have bought anyway, attribution alone cannot answer that.
Attribution uses a rule to give credit
A last-click rule gives credit to the last allowed marketing step. Other rules split credit across several steps. The same customer path can get a different answer under a different rule.
This does not make attribution useless. It can show which campaigns appear before sales and where to look next. It becomes misleading only when assigned credit is called proof that the ad caused the result.
Incrementality estimates the extra result
You cannot show the same person an ad and not show that same person the ad at the same time. A fair test uses two similar groups instead. One group can see the ads. The other group cannot. The difference in results is an estimate of the extra effect.
Google describes this kind of test in its pages about Conversion Lift and modern measurement methods. Google treats incrementality, attribution, and marketing mix models as different tools for different questions.
A test result is still an estimate for a set group, time, campaign, and setup. It may not hold in another season, market, offer, or budget. Some small businesses do not have enough sales, clean records, or test options for a strong result.
Build the basic records before a complex test
- Name the result. Choose a form, good lead, booked job, paid sale, revenue, or profit.
- Test the action. Make sure it fires at the right time and does not count spam or repeats.
- Match the records. Connect ad actions to real people and later business results.
- Use attribution for path questions. Keep the credit rule and dates visible.
- Use a fair test for cause questions. Add a comparison group when the decision and data support it.
- Check the money. Extra sales are not always extra profit. Include cost, canceled work, and payment.
A complex test cannot fix a vague conversion or a customer record that loses the final result. Start with a clear and reliable path.
Match the tool to the decision
- Improve ads or pages
- Use tested conversion actions, ad credit, lead quality, and careful campaign changes.
- Understand the customer path
- Compare ad credit and customer records. Do not call either one proof of cause.
- Ask if ads created extra results
- Use a well-planned test when the platform, market, and number of results allow it.
- Review all marketing
- Use tests, business records, and broader models together. One dashboard cannot answer every question.
- Ask if the result was worth the cost
- Connect extra results to profit, capacity, finished work, and paid revenue.
Before asking for a new report, name the choice you will make if the answer changes.
Keep the limits clear
- A high return in an ad report does not prove the ads caused every credited sale.
- A low return does not prove the ads had no effect outside the chosen rule or time.
- A positive lift estimate does not tell you which message or page made the difference.
- An unclear test result does not prove there was no effect. The test may be too small or weak.
- Different ad and business records do not by themselves prove poor work or dishonesty.
- Better measurement does not fix poor lead follow-up, fit, price, capacity, or customer records.
When there are only a few sales, a simple review may be more honest. Test the conversion, follow customer paths, and make small choices while stating what remains unknown.
Say exactly what the evidence supports
Label each claim as recorded activity, assigned credit, matched business result, test estimate, or opinion. Clear labels show what is known and what still needs a test.
Ad planning and management belong in paid marketing. Missing customer names, status, or revenue between tools may belong in business systems consulting. These are different services that can share the same clear business question.