# How to find ad sets bidding against each other

> Tier: deep. Full contract: quantified decision rules and evidence requirements.

Where your own ad sets compete for the same people on Meta and TikTok, measured with the platforms' own overlap data, and what to merge or exclude.

**Intent.** Stop paying auction premiums against yourself, with the overlapping pairs named and sized.

## When to use

- CPMs creeping while reach stalls.
- After audience proliferation: many lookalikes and interest stacks built over months.
- Before consolidating an account structure.

## When not to use

- Accounts with one or two audiences: nothing to overlap.
- Broad or Advantage+ targeting everywhere: overlap is delegated to the platform there, and the finding would be decorative.

## Sources

Required: Meta Ads
Optional: TikTok Ads

## Preconditions

- Audiences and ad sets readable on the ticked accounts.
- Spend history to weight the findings.

## Inputs

- `scope`: Campaigns whose audiences to cross. Default: active prospecting.

## Procedure

1. List audiences and their definitions on Meta. — tools: `meta_get_audiences`, `meta_get_audience_details`
2. On TikTok, run the platform's own overlap measurement across the active audiences. — tools: `tiktok_get_audience_overlap`, `tiktok_get_audiences`
3. Map which ad sets use which audiences, with spend. — tools: `meta_get_adsets`, `meta_get_insights`, `tiktok_get_adgroups`
4. Rank overlapping pairs by combined spend and overlap share.

## Decision rules

- **When** Two ad sets with meaningful spend target audiences overlapping above 40% → **conclude** Consolidation candidate: merge, or exclude one from the other; show both options with their trade-offs.
- **When** Overlap is high but one ad set is remarketing and the other prospecting → **conclude** Fix is exclusion, not merging: remarketing pools must be excluded from prospecting, and the absence of that exclusion is the finding.
- **When** Overlap exists but combined spend is marginal → **conclude** Note and skip: auction self-competition only costs money where money is.

## Evidence requirements

- Every pair carries the overlap measurement, both spends, and the audience definitions.
- TikTok findings quote the platform's own overlap tool, not an inference.

## Human approval boundary

- Merging audiences and adding exclusions are edits humans apply; status changes proposed here go through Safe Writes.

## Failure modes

- Meta does not expose pairwise overlap percentages for every audience type via API; where measurement is unavailable, the skill says inferred-from-definitions and lowers confidence.
- Lookalikes of the same seed at different percentages nest by construction: 1% inside 5%. That is expected, and only a finding when both run prospecting simultaneously.
- Auction overlap is also time-dependent: dayparting can make overlapping audiences never actually meet.

## Output

- **Pairs, ranked**: Overlap, spends, definitions, proposed fix per pair.
- **Missing exclusions**: Prospecting ad sets without remarketing exclusions.
- **Structure proposal**: The consolidated target structure, if warranted.

## Related

- Skill: [meta-delivery-diagnosis](https://www.getmcpads.com/skills/meta-delivery-diagnosis.md)
- Skill: [account-audit](https://www.getmcpads.com/skills/account-audit.md)
- Guide: [campaign-audit](https://www.getmcpads.com/guides/campaign-audit.md)