AI for CMOs

How do I defend brand spending when last-click gets the credit?

ShareLinkedIn

SimplSolutions editorial team · Measurement · 5 min read

Published · Updated

AI-assisted editorial guidance. Jordan is a fictional campaign character, not a verified human author. Worked examples are illustrative unless explicitly identified otherwise.

Campaign photographs across four industries arranged on a creative desk
Illustrative editorial image, not a customer result.

Why the dashboard and the budget conversation disagree

Last-click reporting answers where a recorded conversion journey finished under a particular tracking setup. A CMO defending brand spending needs to answer a different question: did the campaign create demand that would not otherwise have existed? Neither a strong last-click number nor a weak one resolves that question by itself.

The way forward is not a new acronym on a slide. Build an evidence table that distinguishes recorded activity, directional demand signals and a comparison intended to test incrementality. This guide shows how to prepare that table, avoid common comparison errors and give finance a decision it can assess.

Explain the model before explaining the result

Document the reporting model, conversion definition, date range, lookback assumptions where applicable and which interactions the system can observe. A phone inquiry, a forwarded email or a conversation at a trade show may not appear in the same dataset as a website form. Missing visibility is not proof that those interactions had no effect.

Do not say that last click is useless. It can help describe recorded paths and monitor closing channels. Its limitation is the interpretation. Crediting the last recorded interaction is not an experiment showing that all previous activity was unnecessary. Google's attribution overview explains that attribution assigns conversion credit using rules or models.

For the board, use plain language: "This report allocates credit within the interactions we observed. It does not estimate what would have happened without the campaign." That sentence prevents a reporting convention from becoming a causal claim.

Build three separate evidence columns

The first column is observed conversion activity: inquiries, qualified opportunities or purchases under agreed definitions. The second is directional demand: branded query patterns, direct visits or relevant awareness research. The third is evidence from a designed comparison, if you have one.

EvidenceUseful questionWhat it cannot prove alone
Recorded conversionsWhere did tracked conversions finish?Incremental demand caused by brand activity
Branded queriesIs recorded search interest changing?Which campaign caused the change
Direct visitsIs this observed traffic category changing?That all visits reflect remembered advertising
Buyer interviewsWhat do interviewed buyers recall?Market-wide prevalence or causal lift
Planned comparisonDid outcomes differ against a credible counterfactual?Certainty if groups differ or the test is contaminated

Avoid adding these columns together. A person can appear in more than one. They are different views of evidence, not independent revenue sources to sum.

An illustrative regional comparison

Imagine two regions with comparable prior demand, sales coverage and offer availability. In a four-week baseline, region A records 80 qualified inquiries and region B records 100. During the planned campaign period, A records 104 and B records 110. A rose by 24; B rose by 10.

A simple difference-in-changes calculation is 24 minus 10, or 14 additional inquiries relative to the comparison's absolute change. That arithmetic is not automatically a valid causal estimate. Baseline scale differs, parallel trends may not hold, sales follow-up may have changed and four weeks may be inadequate. An analyst should decide whether absolute changes, normalized outcomes or another design is appropriate before the test starts.

The value of showing the example is its assumptions. If region A also received a new sales representative while B did not, you cannot confidently assign the difference to the brand campaign. Write down the competing explanation instead of hiding it in a footnote.

Never call the 14 inquiries revenue. They are inquiries under the chosen definition. Qualification, opportunity creation and revenue realization require their own follow-through.

Jordan and a strategist reviewing campaign work

Illustrative editorial image, not a customer result.

Decide whether a comparison is credible

Before spending, ask the analyst to review prior patterns, seasonality, regional offer differences, audience overlap and how much movement the design could plausibly detect. A low-volume test may be too uncertain to support a precise answer. Agree the observation period and analysis before seeing results.

Record contamination: national advertising reaching the comparison group, shared sales teams, changing promotions, missing data or buyers moving between regions. If contamination is substantial, report the result as inconclusive or limited. Do not improve confidence by relabeling the same data.

If a credible experiment is not feasible, that does not force an immediate cancellation. It means the funding decision rests on a more limited evidence package and an explicit uncertainty allowance. Finance can decide whether the strategic case warrants another test, a smaller commitment or no expansion.

Replace the attribution argument with a decision brief

For each campaign, write the intended business effect, the time horizon, the evidence available and the cost of getting a stronger answer. Ask finance which uncertainty actually changes the budget decision. You may not need a precise lift estimate to decide against doubling a campaign with a broken landing page.

Separate measurement budget from media budget. A better experiment can require implementation work, analysis and delayed expansion. Make that tradeoff visible instead of promising complete measurement at no cost.

Use this meeting structure:

  1. What is observed, with its reporting definition?
  2. What is a plausible interpretation, with competing explanations?
  3. What remains unknown?
  4. Which spending decision can the current evidence support?
  5. What would the next test resolve?

Questions finance is right to ask

Why did direct traffic rise? You may not know. Verify tagging, reporting changes, promotions and other activity before treating the rise as brand recall.

Why not use a different attribution model? Another allocation can be informative, but a different allocation rule is not necessarily a causal test. Compare the models as sensitivity analysis, not as competing truths.

Can we cut brand and watch revenue? Possibly, but account for the delay between activity and buying, other changes and the risk of an uncontrolled interruption. Have an analyst scope the test; do not improvise after a disappointing dashboard screenshot.

Your next useful deliverable

Create one page with observed activity, directional signals, comparison evidence, limitations and the recommended next commitment. Mark every number as measured, estimated or illustrative. Remove unsupported influence percentages. Use the board-update structure to communicate the decision. If you want help organizing approved evidence across existing tools, talk to SimplSolutions.

Jordan, your AI guide

Your role. Your questions.

Need CMO guidance?
Ask Jordan.

Talk through an idea, ask about the tools you already use, or find out what a first project could look like.

Jordan is a fictional campaign character and AI guide. Our team handles demo requests.