Why is my campaign cannibalizing impressions?

What “impression cannibalization” really means in Google Ads

When advertisers say a campaign is “cannibalizing impressions,” they usually mean one campaign is absorbing traffic they expected another campaign to get. The most important thing to understand is that, inside a single Google Ads account targeting the same domain, your campaigns typically don’t “bid against each other” in the way two different advertisers do. Instead, the system chooses one eligible keyword (or one eligible campaign type) to enter the auction for that search, based on prioritization rules, relevance, eligibility, and Ad Rank.

So “cannibalization” is rarely a bug. It’s usually a signal that your account has overlapping targeting (keywords, search themes, Dynamic Search Ads, Performance Max, Shopping products, locations, audiences, etc.), and the system is resolving that overlap in a way you didn’t intend.

Common symptoms you’ll notice

You’ll often see this as impressions and clicks shifting week-to-week between campaigns, brand traffic showing up in a non-brand campaign, one ad group taking most impressions while “similar” ad groups fade, or Performance Max suddenly showing queries you assumed were “owned” by Search. In many cases, nothing is broken—your structure just isn’t telling the system clearly which campaign should match which intent.

Why your campaigns are cannibalizing impressions (the real causes)

1) Overlapping keywords: only one can trigger, and Google must pick a winner

If multiple keywords (across ad groups and campaigns) are eligible to match the same search term, only one of them can trigger an ad for that search. The system applies a prioritization process that heavily favors “identical” matching first (including spell-corrected terms), then uses relevance and Ad Rank to decide which ad group/campaign gets the impression.

This is why adding “every match type in every campaign” is a classic way to create the impression of cannibalization. You haven’t expanded coverage—you’ve created overlap, and overlap forces a selection.

2) “Identical to the search term” prioritization (and the nuance that trips people up)

When a user’s search is identical to an eligible keyword, that keyword gets preference in a predictable way. But advertisers get caught by two nuances.

First, “identical” includes spell-corrections, but it does not automatically include things like plurals or synonyms. Second, if you’re expecting a “more specific” ad group to always win, remember that when the search isn’t identical to a keyword, the system leans on relevance (meaning/intent) and then Ad Rank, which can cause a different ad group to consistently take the impression—even if both seem “close enough” to you.

3) AI-based prioritization + Ad Rank: the most relevant ad group may not be the one you expected

When the query isn’t identical to a keyword, the system can narrow down to the most relevant ad groups and then pick the winner by Ad Rank. In practical terms, your account structure (thematic grouping), landing pages, and creative quality are part of what shapes which ad group is considered “most relevant.”

This is where messy ad groups (too many mixed intents in one ad group, or multiple ad groups targeting the same intent with slightly different wording) create confusion. The system may consistently route impressions to the ad group it deems most relevant—even if you intended a different “bucket” for reporting or budgeting.

4) Budget and eligibility exceptions: the “right” campaign can lose even when it has priority

Even when prioritization says Campaign A should win, it can lose if it isn’t eligible at the moment of the auction. The most common reasons are budget restriction (including “Limited by budget”), low search volume keywords that are temporarily inactive, targeting mismatches (location, language, audience restrictions, exclusions), or creatives/landing pages being disapproved or otherwise not eligible.

This often looks like cannibalization, but it’s really “the preferred option wasn’t available,” so the system served the next-best eligible option.

5) Performance Max vs Search: Search wins only on truly “identical” eligible queries

If you run Performance Max alongside Search, overlap is expected. The key behavioral rule is that Search is prioritized when the user’s query is identical to an eligible Search keyword (any match type). When the query is not identical to an eligible Search keyword, the system can select whichever campaign/ad has the highest Ad Rank—meaning Performance Max can win impressions on a lot of “nearby” intent.

This is one of the biggest reasons advertisers feel like Performance Max is “stealing” impressions. In reality, it’s doing what it’s designed to do: expand into incremental queries and inventory, while still respecting eligible identical keyword coverage in Search.

6) Performance Max vs Shopping: product overlap can shift spend and impressions fast

If you have product feed-based campaigns and you introduce Performance Max for the same products in the same account, you can see spend and impressions decrease in your existing Shopping campaigns as Performance Max ramps up. This can happen even if Performance Max has a lower budget, because the system is prioritizing how those products are served across eligible surfaces.

If your reporting depends on separate Shopping campaigns, this can feel like cannibalization; strategically, it’s usually a signal you should simplify: decide which campaign type is meant to be the “primary owner” of those products and structure accordingly.

7) Negative keywords and brand controls can redirect impressions (sometimes unintentionally)

Negatives are supposed to sculpt traffic, but they’re also a common self-inflicted cause of cannibalization. If Campaign A blocks queries (or a portion of intent) through negatives, Campaign B becomes the only eligible place that traffic can go.

Two advanced gotchas matter here. First, account-level negative keywords apply broadly across Search and Shopping inventory in multiple campaign types, which can create account-wide shifts if you’re not careful. Second, negative keywords don’t automatically match close variants/expansions the same way positive keywords can—so you may think you “blocked it,” but you only blocked a specific variant, which leads to inconsistent routing.

For brand traffic specifically, brand exclusions are often the cleaner tool than trying to build massive negative keyword sets, especially in highly automated campaign types.

8) Shared budgets can make “who gets impressions” feel unpredictable

Shared budgets don’t directly change auction eligibility rules, but they absolutely change which campaigns have headroom at different times of day. If Campaign A is more aggressive (higher Ad Rank, broader reach, stronger conversion rate, or just more eligible volume), it can consume more of the shared budget earlier, leaving Campaign B constrained later—creating a daily pattern that feels like cannibalization.

9) Brand exclusions for Search changed starting May 27, 2025 (AI Max dependency)

If you use brand exclusions in Search, the workflow changed starting May 27, 2025: brand exclusions for Search began upgrading into AI Max. Practically, that means if you’re trying to manage brand vs non-brand splitting and your exclusions/settings aren’t where you expect them, you may be looking in the wrong place (or you may need AI Max enabled to create new brand lists in certain setups). This doesn’t change Performance Max behavior the same way, but it can absolutely affect how you isolate brand demand inside Search campaigns.

How to diagnose and fix impression cannibalization (a systematic playbook)

Step 1: Prove it’s overlap (not demand change)

Before changing structure, confirm you’re dealing with internal overlap rather than market demand shifting. Use your search terms reporting to compare which queries each campaign is actually serving, and whether the “missing” impressions moved to a different campaign or disappeared altogether.

Step 2: Run this fast diagnostic checklist

• Check campaign status and eligibility: Look for “Limited by budget,” disapproved ads/assets, ended schedules, location/language mismatches, or overly restrictive audience settings.
• Check for duplicate/overlapping keywords: Especially the same “core” keyword repeated across multiple campaigns and ad groups with different match types.
• Check Performance Max overlap: If Performance Max is live, assume it can win non-identical queries unless Search has an identical eligible keyword.
• Audit negatives (including account-level negatives): Confirm you didn’t block one campaign and inadvertently force traffic into another.
• Check shared budgets: If campaigns share a budget, confirm whether one campaign is draining headroom and causing the other to lose impression share later in the day.

Step 3: Decide what you want: “clean attribution” vs “max performance”

Many cannibalization problems are really reporting problems. If your top goal is clean channel segmentation (brand vs non-brand, product line A vs product line B, location vs location), you’ll need tighter structure and stronger traffic sculpting. If your top goal is maximum conversions at the account level, you’ll often accept some overlap and focus on making sure the best campaign wins consistently (rather than trying to force equal distribution).

Step 4: Fix overlap the right way (without over-engineering)

Start by simplifying your keyword architecture. You generally don’t need the same keyword repeated across multiple match types and multiple campaigns “for coverage.” Broader match types already capture narrower behavior, and duplicating similar keywords often just forces the system to pick one based on Ad Rank. Consolidate where possible, and keep ad groups tightly thematic so relevance signals are clear.

Use negatives surgically, not emotionally. If you must split intent, use negatives to create clean lanes (for example, excluding brand terms from non-brand). But avoid building huge negative lists to micromanage automation—especially in Performance Max—because overly restrictive negatives can reduce reach and starve learning. For brand shaping, consider brand exclusions/settings where appropriate rather than trying to block every misspelling and variant manually.

When Performance Max is involved, separate by goal, not by wishful thinking. If you’re running Performance Max for incremental reach plus Search for controlled keyword coverage, keep Search strong on the exact “must-win” queries (with eligibility intact: budget, approvals, targeting). Then let Performance Max do its job on the non-identical expansion. If you’re seeing Performance Max take traffic you truly want in Search, it’s often because your Search keyword isn’t identical to the query (or it’s not eligible when the auction happens).

Step 5: Stabilize budgets so the intended campaign can actually win

If a campaign is budget-restricted, it may not reliably claim impressions even when it has prioritization preference. Fixing “Limited by budget” (or simply giving the campaign more headroom) often “solves cannibalization” overnight because the campaign becomes eligible more consistently.

If you use shared budgets, be intentional: shared budgets can improve overall budget utilization, but they can also make intra-day impression distribution feel volatile. If stable separation matters, separate the budgets.

Step 6: For Shopping/feed advertisers: don’t let product overlap create accidental competition

If Performance Max and Shopping campaigns target the same products, expect impression and spend shifts. The cleanest fixes are usually structural: consolidate products into the campaign type you want to be primary, avoid duplicated product targeting across multiple asset groups, and ensure budgets are aligned during transitions so you don’t accidentally choke the campaign that’s meant to carry performance.

Step 7: Use the newer transparency features to validate changes (especially for Performance Max)

For Performance Max, use search impact and search term visibility to understand what queries are coming from keywordless matching versus your provided search themes, and whether your search themes are adding incremental traffic. This is the fastest way to confirm whether “cannibalization” is actually overlap on core demand, or incremental expansion you should keep.

If it feels like one Google Ads campaign is “cannibalizing” another’s impressions, it’s usually not because your campaigns are bidding against each other, but because Google is resolving overlapping eligibility: multiple keywords or campaign types can match the same query, and the system will pick a single winner based on prioritization (including “identical to the search term” matching), relevance signals, and Ad Rank—while budget limits, disapprovals, shared budgets, negatives, and brand controls can quietly make the campaign you expected ineligible so another one serves instead (and with Performance Max, near-identical intent can also shift depending on Ad Rank when there isn’t an exact eligible Search keyword). If you want a faster way to spot where overlap is coming from and what to adjust—keywords, negatives/brand settings, structure, or budget headroom—Blobr plugs into your Google Ads and runs specialized AI agents to surface concrete fixes, like a Negative Keywords Brainstormer to create clean lanes between campaigns and reduce unintended query overlap, so you can clarify “who owns which intent” without endless manual audits.