AdAce AdsDiagnose campaign overlap without treating it as the cause of every problem
Two campaigns can reach similar people or answer similar searches without explaining every cost increase. Overlap may be useful, accidental or simply visible in attribution. Before consolidating campaigns, define the kind of overlap you suspect and the business harm it would cause. An overlap diagnosis should distinguish an observed relationship from a causal claim.
Four different meanings of overlap
Use four separate definitions. Query overlap means similar search tasks can be eligible for more than one campaign. Audience overlap means the intended or observed populations intersect. Offer overlap means campaigns communicate similar propositions to the same decision. Reporting overlap means multiple sources credit a business outcome. These are different problems: duplicated reporting is not proof of duplicated spending, and intersecting intended audiences do not reveal which people actually received ads.
Check platform mechanics before using the phrase 'bidding against ourselves'. Google explains that eligible keywords targeting the same domain do not compete against one another in the auction; prioritisation selects which can trigger the ad. Eligibility exceptions matter. This is a specific Google Search-related mechanism, not a universal statement about every platform, campaign type or cross-account arrangement. For other contexts, inspect the applicable official documentation rather than extending the rule by analogy.
Educational example: after adding Campaign B, Campaign A's attributed sales fall from 20 to 12 while B reports ten. The business ledger moves from 20 unique sales to 22. The campaign totals could suggest displacement, while the business total shows a different question. Those figures still do not establish that B created two extra sales: seasonality, offer changes and customer demand may also have shifted. Use the example to define the comparison, not to claim incrementality.
Build an evidence table linking the overlap hypothesis to a plausible mechanism and an alternative explanation. Similar queries might expose inconsistent landing pages. Similar audiences might receive contradictory offers. Similar attributed sales might reflect different attribution models. If no business harm can be described and measured, removing overlap purely for a cleaner account diagram may sacrifice useful demand. Administrative simplicity is a valid objective, but it should be named separately from performance improvement.
Intersection, mechanism and comparison
- Observed intersection: what is actually visible, its source and its coverage. Keep intended settings separate from reports and distinguish partial evidence from a complete delivery picture.
- Business mechanism: the specific waste, confusion or processing burden suspected. It must be possible to explain how the overlap could affect a compatible outcome.
- Comparison plan: the portfolio result that would support or weaken the hypothesis. Include alternatives and a legitimate inconclusive result, not only a campaign-level winner.
Investigate the portfolio instead of moving credit
- Inventory campaign roles, offers and outcome definitions. A naming similarity is not sufficient evidence that two campaigns serve the same business task.
- Collect the available query, audience or attribution evidence for matching dates. Record coverage gaps and the scope in which the platform mechanics apply.
- Reconcile aggregate unique business outcomes where permitted. Do not add separately attributed sales into a total that counts the same transaction more than once.
- Write the suspected harm and alternative explanations, including demand shifts, measurement changes and fulfilment constraints. Reject hypotheses that cannot be assessed with the available evidence.
- Design the smallest useful comparison, keeping major offer and operational conditions stable where possible. Verify current experiment support instead of assuming ordinary campaign duplication is a controlled test.
- After any separately authorised change, evaluate the total portfolio and business records. Capture displaced attribution, useful volume and workload so a better-looking campaign does not conceal a worse overall result.
Overlap is a hypothesis, not a universal cause
- The sales counts are educational and do not describe a real client or prove additional sales. Before-and-after totals remain vulnerable to confounding.
- Do not import a Google keyword prioritisation rule into Meta or other platforms. Audience selection, observed delivery and causal interference require separate evidence.
- This is a read-only diagnostic method. Ask AI for an evidence table and written test plan without propose_change or other writes; consolidation and exclusions need separate concrete authorisation.
Sources and further reading
Try it on your own accounts
Create a workspace, connect Google or Meta in a couple of clicks and see your accounts clearly. Changes follow your approvals or the policy you configure.
Create your workspace