Start by confirming what “the right people” means in your account
When ads reach the “wrong” people, the root cause is usually one of two things: you’re showing to the wrong audience (reach problem), or you’re showing to the right audience at the wrong moment/intent (query and messaging problem). The fix is different depending on which bucket you’re in, so I always start with a quick reality check: are the clicks coming from the wrong geography/device/language/demographic, or are they coming from people who are simply researching, bargain-hunting, or looking for a different variant than you sell?
Fast diagnostic checklist (10 minutes)
- Search campaigns: Pull the Search terms report for the last 7–30 days and sort by spend and clicks. If you see lots of “adjacent intent” terms, this is mostly keyword/match/negatives, not audiences.
- Location sanity check: Compare “User location” versus “Locations matched” and look for out-of-area traffic that’s explained by “interest in” locations rather than physical presence.
- Networks: Confirm whether Search Partners is on, and whether any Search campaign is also opted into the Display Network.
- Display/Video/Demand Gen: Confirm each ad group has at least one real targeting method (audience/content/placement), and confirm whether optimized targeting is enabled.
- Performance Max: Confirm you’re treating audience signals as guidance, not hard targeting, and validate where conversions are actually coming from before tightening anything.
The most common reasons your ads reach the wrong people (and how to spot each one)
1) Location targeting is broader than you think (Presence vs Presence/Interest)
The single most common “wrong people” complaint I see is location-based. Many campaigns use a location option that includes people who have shown interest in your targeted locations, not just people physically in them. That can be great for tourism, moving services, or remote offerings—but it’s a disaster for local services that must be delivered in-person within a tight radius.
If you’re a local business, you generally want to bias toward reaching people in or regularly in your targeted locations, then use location exclusions to clean up edge cases. Keep in mind that location signals aren’t perfect, and you can still see some traffic that doesn’t look “local” at first glance. That’s why I recommend validating performance by location before making aggressive changes.
Also watch for layered location exclusions that accidentally wipe out areas you intended to target (for example, excluding a state while targeting a city inside that state, which results in the city being excluded too). When you see “my ads aren’t reaching the right people,” it’s often paired with “and I’m not getting enough volume,” and conflicting inclusions/exclusions are a frequent culprit.
2) Language targeting doesn’t work the way most advertisers assume
Language settings are not a “what language did the person type their search in” filter. They’re primarily based on the person’s language signals (such as browser/device language settings), which means you can accidentally exclude good users—or include users you didn’t expect—if you set language too narrowly.
For most advertisers in the United States who sell in English, targeting English is fine. Where it goes wrong is when advertisers try to over-control language and unintentionally block bilingual users, travelers, or users whose device language doesn’t match the language you assumed they use day-to-day.
3) Your campaign is showing on networks you didn’t intend (Search Partners and Display Network)
On Search, ads can be eligible to show beyond the core search engine via Search Partners. That isn’t inherently bad, but performance and user behavior can be different, and it can absolutely create a “these aren’t my people” feeling—especially if your offer is niche and you’re sensitive to lead quality.
Separately, a classic mistake (especially in older accounts) is enabling the Display Network on a Search campaign. That can dramatically widen where ads appear and change the type of user you attract. If your goal is high-intent demand capture, keep Search and Display separated so you can control targeting, creative, and expectations appropriately.
4) Keyword match behavior is doing exactly what it’s designed to do (but you didn’t build guardrails)
If you run Search, most “wrong audience” issues are actually match-type and intent issues. Today’s matching is more intent-based than many advertisers expect, and “close variants” are eligible across match types. You can’t opt out of close variants, so the way you regain control is through structure, negatives, and landing-page alignment.
Here’s what typically causes drift:
Broad match without the right conversion signals. Broad match can consider additional signals (like recent search behavior, other keywords in the ad group, and the content of your landing pages and assets) to determine intent. That can be powerful, but it also means broad match can expand into queries that are “related” rather than “perfect.” In practice, broad match is at its best when paired with Smart Bidding and clean conversion tracking, because you’re letting the system explore while your bidding strategy filters toward the users most likely to convert.
Phrase match expectations based on the old world. Phrase match behavior has evolved over time; it’s not the rigid “contains the exact phrase in this order” rule set many people learned years ago. It aims to match meaning, while still using word order as a signal. If you’re still building campaigns as if phrase match is a strict leash, you may be surprised by what it can match to.
Not using the Search terms report as a weekly control loop. The Search terms report is where you find the real “who” behind Search: the queries. If those queries aren’t your buyers, you don’t have an audience problem—you have a query-intent problem.
Negative keyword gaps (and negative keywords don’t match close variants). Many advertisers assume a negative keyword will block plural/singular and misspellings automatically. In practice, negatives don’t match close variants the way positives do, so you often need to add variations if you’re trying to surgically remove a theme. Also, negative match types behave differently than positive match types, so it’s worth being deliberate about whether you need negative broad, phrase, or exact.
5) Audience settings are misunderstood (Targeting vs Observation)
In Search campaigns, audiences are frequently added in Observation mode (which does not restrict who can see the ads). That’s not a mistake; it’s often the recommended setup because it lets you measure and adjust without cutting reach. But if you thought you “targeted an audience” and you’re still getting everyone, this is usually why.
On Display/Video, the opposite problem happens: advertisers accidentally create ad groups with no meaningful targeting method applied, so the ads can run very broadly within the campaign’s basic settings (like location, language, and content exclusions). If you’re buying awareness, that might be okay. If you’re trying to reach a specific segment, it’s a recipe for wasted spend.
6) Optimized targeting (and demographic expansion) is expanding beyond your inputs
For Display, Video, and Demand Gen, optimized targeting can use your selected signals as a starting point and then expand beyond them. This is often the hidden driver behind “why am I getting people outside my audience?” The platform is trying to find additional users likely to achieve your goal, not strictly stay inside your hand-picked segments.
In Demand Gen specifically, optimized targeting can also expand beyond demographic signals. If your product truly only fits a narrow demographic, you’ll want to be intentional here: expansion may improve volume, but it can dilute lead quality if your conversion tracking isn’t accurately rewarding the outcomes you care about.
One key nuance: exclusions still matter. For example, excluding users via certain first-party/customer data segment exclusions can prevent serving to those excluded groups even when expansion is on.
7) Performance Max audience signals are not strict targeting
Performance Max often surprises advertisers because audience signals are signals, not hard constraints. The system can show ads to relevant users outside your signals if it predicts they’re likely to convert and help you hit the campaign goal. If your expectation is “only show to this audience,” you’ll feel like you’re reaching the wrong people—even though the campaign is behaving as designed.
The correct way to evaluate Performance Max is to focus on the quality of conversions (and the down-funnel value, if you import it), then adjust inputs and exclusions based on what’s actually producing business results.
8) Demographic targeting and the “Unknown” bucket can skew what you think is happening
Demographic reporting includes “Unknown” categories when age/gender (and in some campaign types, parental status or household income) can’t be identified. Many advertisers exclude “Unknown” to try to get cleaner traffic, then wonder why reach collapses—or why the remaining traffic looks oddly unbalanced. In most accounts, “Unknown” is a meaningful slice of the real world, not a junk drawer.
My rule: only exclude demographic categories when you have a clear performance case, and only after you’ve confirmed you’re not blocking too much scale or biasing delivery into a narrow subset that isn’t representative of your true buyers.
9) You may be in a category with targeting restrictions (and that changes your options)
Some verticals have strict restrictions on personalized advertising and demographic targeting (especially for sensitive categories and certain regulated areas). Additionally, personalized advertising protections apply to users under 18, which limits how narrowly you can target in certain contexts. If you’re advertising in housing, employment, or consumer finance in the United States, your ability to use certain demographic targeting and some location granularity can be restricted, and “why can’t I reach exactly who I want?” becomes partly a compliance reality, not just an optimization issue.
A practical, systematic plan to get your ads in front of the right people
Step 1: Fix intent first (Search campaigns)
If your Search terms show mismatched intent, don’t start by stacking audiences. Start by tightening the query layer. Clean targeting at the keyword level typically improves both lead quality and bidding efficiency.
- Use the Search terms report weekly and add negatives for themes that are clearly not your customer (jobs, free, DIY, definitions, competitors if irrelevant, parts if you sell services, etc.).
- Choose match types based on how precise the intent must be. If you’re selling something expensive or lead quality is critical, lean more on exact/phrase for your core terms, and expand carefully.
- If you use broad match, pair it with strong conversion measurement and a bidding strategy built to optimize for outcomes (not just clicks). Broad match can use additional intent signals, but it needs the right “reward function” (your conversion tracking) to learn what “right people” means.
- Align ad copy and landing pages to filter. Put qualifiers in ads (pricing, minimums, service area, “for businesses,” “for homeowners,” etc.) and mirror them on the landing page. This reduces curiosity clicks from the wrong segment.
Step 2: Fix reach next (Location, language, networks)
Once intent is under control, tighten reach so you’re not paying to educate people you can’t serve.
For local service businesses, I typically recommend testing a switch to a stricter location presence setting (people in or regularly in your locations) and then using exclusions to clean up any known problem areas. If you’re national or remote, keep the broader setting but monitor geographic performance and add exclusions where lead quality is consistently poor.
On language, avoid over-filtering unless you have a strong reason. On networks, separate Search from Display, and only keep Search Partners enabled if it’s proving incremental, profitable volume.
Step 3: Make Display/Video/Demand Gen targeting “real,” then decide how much automation you can tolerate
If you’re seeing irrelevant traffic on Display/Video/Demand Gen, the first fix is often structural: ensure every ad group has a clear targeting method and that you’re intentionally using Targeting vs Observation. Then decide whether optimized targeting belongs in that campaign based on your goal and the quality of your conversion signals.
If your conversion tracking is strong and your objective is efficient scale, optimized targeting can help find additional converters. If your conversion tracking is weak (or you’re optimizing to an easy but low-quality action), expansion will reliably find more of the wrong people—because you’re unintentionally telling the system those people are “right.”
Step 4: For Performance Max, treat “right people” as a measurement problem first
With Performance Max, your levers are more about inputs and feedback than strict targeting. If you want it to reach better users, prioritize: cleaner conversion definitions (quality over quantity), accurate values when possible, and thoughtful audience signals that represent your best customers. Expect it to go beyond signals, then judge it on business outcomes—not on whether every impression matches your mental picture of your audience.
Step 5: Build a “quality feedback loop” so the system can learn your true customer
The final step is what separates accounts that constantly fight audience mismatch from accounts that improve month after month: a quality feedback loop. If you generate leads, that means tracking qualified leads (not just form submits). If you sell e-commerce, it means optimizing to revenue and margin-aware values when feasible. The more clearly your measurement distinguishes “right people who buy” from “wrong people who click,” the less you’ll need to micromanage targeting—and the more the platform’s automation will work in your favor instead of against you.
Let AI handle
the Google Ads grunt work
| Area | Issue / Concept | How to Spot It | Recommended Fix | Key Google Ads Features / Docs |
|---|---|---|---|---|
| Initial diagnosis | Clarify “right people” = reach vs. intent |
Check if “wrong people” means: – Wrong geo / device / language / demographic – Or right people but wrong intent (researchers, bargain hunters, wrong variant) |
Separate problems into: – Reach (who can see ads: locations, languages, networks, audiences) – Intent (queries, match types, messaging, landing pages) Then fix intent first, reach second. |
Use the search terms report to see real queries behind clicks. |
| Fast check | 10‑minute account sanity check |
– Pull Search terms for last 7–30 days; sort by spend/clicks – Compare “User location” vs “Locations matched” – Check if Search campaigns include Search Partners or Display – On Display/Video/Demand Gen, verify there is at least one audience/content/placement and whether optimized targeting is on – For Performance Max, check where conversions actually come from |
Use this snapshot to decide if the main issue is: – Query/match/negatives (intent), or – Location/language/network/audience (reach). |
– Search terms report – Location targeting – Search Network & partners |
| Location | Presence vs. Presence or interest too broad |
– Traffic from outside your true service area in location reports – “User location” differs heavily from “Locations matched” |
– For local/in‑person services, prefer “people in or regularly in” your locations – Add exclusions for known bad areas – Watch for overlapping targets/exclusions (e.g. excluding a state while targeting a city inside it) |
– Target geographic locations – Exclude geographic locations – Location target types |
| Language | Language targeting misunderstood |
– Good potential customers are missing, or unexpected languages appear – You assumed “language = language of query” instead of device/browser signals |
– For US English advertisers, targeting English only is usually fine – Avoid over‑restricting languages; allow for bilingual and traveler scenarios unless you have a strong reason to narrow |
– Available languages & language targeting – View location and language targeting |
| Networks | Unintended networks (Search Partners / Display) |
– Search campaigns showing on partners or Display when you expected only core Search – Lead quality feels inconsistent or “not my people,” especially for niche offers |
– Decide intentionally if Search Partners fit your goals; opt out if not – Keep Search and Display in separate campaigns so you can control targeting and creative differently |
– Google Search Network & partners |
| Keywords & intent | Match types and close variants causing drift |
– Search terms show “adjacent intent” (research, DIY, jobs, cheap, different product types) – Phrase and broad match reaching queries you didn’t expect |
– Make the Search terms report a weekly control loop; mine for themes to add or exclude – Use more exact/phrase for high‑value or sensitive terms – If you use broad, pair it with strong conversion tracking and outcome‑based bidding – Add qualifiers in ad copy/landing pages to self‑filter the wrong users |
– Search terms report – Negative exact match |
| Negatives | Negative keyword gaps & behavior |
– Irrelevant queries keep showing up even after adding some negatives – Plural/singular/misspellings of an unwanted term still appear |
– Remember negatives do not expand to all close variants; add important variations explicitly – Use appropriate negative match types (broad, phrase, exact) depending on how tightly you need to block a theme – Consider account‑level negative lists for global junk themes |
– Account‑level negative keywords – About negative keywords |
| Search audiences | Targeting vs. Observation confusion |
– You “added an audience” but still see everyone in Search – Display/Video ad groups running extremely broad with no clear targeting |
– In Search, use Observation to learn and bid‑adjust without cutting reach; use Targeting only for deliberate “RLSA‑style” segments – On Display/Video, always apply at least one real targeting method (audience/content/placement) and use Targeting vs Observation intentionally |
– Targeting and Observation settings – Your data (audience) segments |
| Display / Video / Demand Gen | Optimized targeting expanding beyond inputs |
– You see users well outside your defined segments or demographics – Volume is high but lead quality is weak |
– Treat optimized targeting as expansion, not strict targeting – Only enable it when your conversion tracking truly reflects quality outcomes – Rely on exclusions (including first‑party lists) where you must never serve |
– Optimized targeting – Video targeting tips |
| Performance Max | Audience signals ≠ hard targeting |
– Ads show to users outside your audience signals – Feels like “wrong people” even when conversions look okay |
– Understand that audience signals guide Google AI; they don’t restrict reach – Judge Performance Max on conversion quality and value, not on whether every impression matches your mental persona – Improve conversion definitions, values, and audience signals; use exclusions where needed |
– Performance Max overview – Audience signals in Performance Max |
| Demographics | Over‑excluding “Unknown” demographic bucket |
– After excluding “Unknown,” reach and conversions drop sharply – Remaining traffic looks skewed toward a narrow subset of users |
– Treat “Unknown” as a normal share of real users, not junk – Only exclude demographics (including Unknown) when there’s a clear performance or compliance reason and you’ve checked impact on scale |
– Demographic targeting & “Unknown” |
| Policy & restrictions | Restricted targeting categories |
– In housing, employment, credit and other sensitive areas, you can’t narrow targeting as aggressively as in other verticals – Some demographic/location options are disabled |
– Recognize that part of “I can’t reach exactly who I want” is policy‑driven – Design campaigns to comply first, then optimize within allowed levers |
– Personalized advertising (sensitive categories guidance) |
| Step‑by‑step fix | Step 1 – Fix intent (Search) |
– Search terms show misaligned intent vs. your true buyers – Lots of informational, job‑seeker, free/cheap, DIY, or wrong‑product queries |
– Use Search terms weekly to add negative themes (jobs, free, DIY, definitions, irrelevant competitors, parts, etc.) – Use more exact/phrase where precision is critical – Align ad copy and landing pages with qualifiers (price, minimums, service area, “for businesses,” “for homeowners,” etc.) to discourage bad clicks |
– Search terms report – Account‑level negatives |
| Step‑by‑step fix | Step 2 – Fix reach (location, language, networks) |
– You’re paying for clicks in areas you can’t serve – Over‑narrowed languages or networks limiting performance |
– For local services, test stricter presence settings and add specific exclusions – For national/remote, keep broader location but exclude chronic low‑quality regions – Avoid over‑filtering languages; separate Search and Display; only keep Search Partners if they add profitable volume |
– Location targeting – Search Network & partners – Language targeting |
| Step‑by‑step fix | Step 3 – Make Display/Video/Demand Gen targeting “real” |
– Ad groups have little or no explicit targeting – Optimized targeting + weak conversion tracking yields lots of low‑quality visits |
– Ensure each ad group has clear targeting (audience segments, topics, placements, or keywords) – Decide consciously how much automated expansion you’ll allow based on conversion signal quality |
– Optimized targeting – Video campaign optimization |
| Step‑by‑step fix | Step 4 – Measure Performance Max by outcomes |
– Hard to see exactly who is being targeted – Temptation to judge by impression‑level “fit” instead of conversion value |
– Refine conversion goals to emphasize quality over quantity – Provide high‑quality audience signals and creative inputs – Use exclusions and negative controls where necessary, then judge by revenue/lead quality |
– About Performance Max campaigns – Audience signals |
| Step‑by‑step fix | Step 5 – Build a quality feedback loop |
– Platform optimizes toward cheap but low‑value actions – Automation keeps finding “wrong people who convert on the wrong thing” |
– Track and optimize to qualified leads, not just form submits – For e‑commerce, optimize to revenue and, when possible, margin‑aware or lifetime value signals – The better your measurement distinguishes “right people who buy” from “wrong people who click,” the less you need to micromanage targeting |
– Combine accurate conversion tracking with targeting and bidding guidance throughout the docs referenced above. |
If your ads feel like they’re reaching “the wrong people,” it usually helps to separate a reach problem (who can see your ads) from an intent problem (what people are actually looking for). In practice, the fastest way to diagnose it is to review your Search terms and location reports to see whether you’re paying for irrelevant queries, close-variant drift, or clicks coming from outside your true service area, then tighten match types, add explicit negatives, and confirm settings like presence vs. presence-or-interest, language signals, and unintended networks such as Search Partners or Display. If you want a more systematic way to keep that feedback loop running, Blobr connects to your Google Ads account and uses specialized AI agents to continuously surface wasted spend, suggest negative keywords, and check alignment between keywords, ads, and landing pages, so you can iterate toward higher-quality traffic without having to manually comb through reports every week.
Start by confirming what “the right people” means in your account
When ads reach the “wrong” people, the root cause is usually one of two things: you’re showing to the wrong audience (reach problem), or you’re showing to the right audience at the wrong moment/intent (query and messaging problem). The fix is different depending on which bucket you’re in, so I always start with a quick reality check: are the clicks coming from the wrong geography/device/language/demographic, or are they coming from people who are simply researching, bargain-hunting, or looking for a different variant than you sell?
Fast diagnostic checklist (10 minutes)
- Search campaigns: Pull the Search terms report for the last 7–30 days and sort by spend and clicks. If you see lots of “adjacent intent” terms, this is mostly keyword/match/negatives, not audiences.
- Location sanity check: Compare “User location” versus “Locations matched” and look for out-of-area traffic that’s explained by “interest in” locations rather than physical presence.
- Networks: Confirm whether Search Partners is on, and whether any Search campaign is also opted into the Display Network.
- Display/Video/Demand Gen: Confirm each ad group has at least one real targeting method (audience/content/placement), and confirm whether optimized targeting is enabled.
- Performance Max: Confirm you’re treating audience signals as guidance, not hard targeting, and validate where conversions are actually coming from before tightening anything.
The most common reasons your ads reach the wrong people (and how to spot each one)
1) Location targeting is broader than you think (Presence vs Presence/Interest)
The single most common “wrong people” complaint I see is location-based. Many campaigns use a location option that includes people who have shown interest in your targeted locations, not just people physically in them. That can be great for tourism, moving services, or remote offerings—but it’s a disaster for local services that must be delivered in-person within a tight radius.
If you’re a local business, you generally want to bias toward reaching people in or regularly in your targeted locations, then use location exclusions to clean up edge cases. Keep in mind that location signals aren’t perfect, and you can still see some traffic that doesn’t look “local” at first glance. That’s why I recommend validating performance by location before making aggressive changes.
Also watch for layered location exclusions that accidentally wipe out areas you intended to target (for example, excluding a state while targeting a city inside that state, which results in the city being excluded too). When you see “my ads aren’t reaching the right people,” it’s often paired with “and I’m not getting enough volume,” and conflicting inclusions/exclusions are a frequent culprit.
2) Language targeting doesn’t work the way most advertisers assume
Language settings are not a “what language did the person type their search in” filter. They’re primarily based on the person’s language signals (such as browser/device language settings), which means you can accidentally exclude good users—or include users you didn’t expect—if you set language too narrowly.
For most advertisers in the United States who sell in English, targeting English is fine. Where it goes wrong is when advertisers try to over-control language and unintentionally block bilingual users, travelers, or users whose device language doesn’t match the language you assumed they use day-to-day.
3) Your campaign is showing on networks you didn’t intend (Search Partners and Display Network)
On Search, ads can be eligible to show beyond the core search engine via Search Partners. That isn’t inherently bad, but performance and user behavior can be different, and it can absolutely create a “these aren’t my people” feeling—especially if your offer is niche and you’re sensitive to lead quality.
Separately, a classic mistake (especially in older accounts) is enabling the Display Network on a Search campaign. That can dramatically widen where ads appear and change the type of user you attract. If your goal is high-intent demand capture, keep Search and Display separated so you can control targeting, creative, and expectations appropriately.
4) Keyword match behavior is doing exactly what it’s designed to do (but you didn’t build guardrails)
If you run Search, most “wrong audience” issues are actually match-type and intent issues. Today’s matching is more intent-based than many advertisers expect, and “close variants” are eligible across match types. You can’t opt out of close variants, so the way you regain control is through structure, negatives, and landing-page alignment.
Here’s what typically causes drift:
Broad match without the right conversion signals. Broad match can consider additional signals (like recent search behavior, other keywords in the ad group, and the content of your landing pages and assets) to determine intent. That can be powerful, but it also means broad match can expand into queries that are “related” rather than “perfect.” In practice, broad match is at its best when paired with Smart Bidding and clean conversion tracking, because you’re letting the system explore while your bidding strategy filters toward the users most likely to convert.
Phrase match expectations based on the old world. Phrase match behavior has evolved over time; it’s not the rigid “contains the exact phrase in this order” rule set many people learned years ago. It aims to match meaning, while still using word order as a signal. If you’re still building campaigns as if phrase match is a strict leash, you may be surprised by what it can match to.
Not using the Search terms report as a weekly control loop. The Search terms report is where you find the real “who” behind Search: the queries. If those queries aren’t your buyers, you don’t have an audience problem—you have a query-intent problem.
Negative keyword gaps (and negative keywords don’t match close variants). Many advertisers assume a negative keyword will block plural/singular and misspellings automatically. In practice, negatives don’t match close variants the way positives do, so you often need to add variations if you’re trying to surgically remove a theme. Also, negative match types behave differently than positive match types, so it’s worth being deliberate about whether you need negative broad, phrase, or exact.
5) Audience settings are misunderstood (Targeting vs Observation)
In Search campaigns, audiences are frequently added in Observation mode (which does not restrict who can see the ads). That’s not a mistake; it’s often the recommended setup because it lets you measure and adjust without cutting reach. But if you thought you “targeted an audience” and you’re still getting everyone, this is usually why.
On Display/Video, the opposite problem happens: advertisers accidentally create ad groups with no meaningful targeting method applied, so the ads can run very broadly within the campaign’s basic settings (like location, language, and content exclusions). If you’re buying awareness, that might be okay. If you’re trying to reach a specific segment, it’s a recipe for wasted spend.
6) Optimized targeting (and demographic expansion) is expanding beyond your inputs
For Display, Video, and Demand Gen, optimized targeting can use your selected signals as a starting point and then expand beyond them. This is often the hidden driver behind “why am I getting people outside my audience?” The platform is trying to find additional users likely to achieve your goal, not strictly stay inside your hand-picked segments.
In Demand Gen specifically, optimized targeting can also expand beyond demographic signals. If your product truly only fits a narrow demographic, you’ll want to be intentional here: expansion may improve volume, but it can dilute lead quality if your conversion tracking isn’t accurately rewarding the outcomes you care about.
One key nuance: exclusions still matter. For example, excluding users via certain first-party/customer data segment exclusions can prevent serving to those excluded groups even when expansion is on.
7) Performance Max audience signals are not strict targeting
Performance Max often surprises advertisers because audience signals are signals, not hard constraints. The system can show ads to relevant users outside your signals if it predicts they’re likely to convert and help you hit the campaign goal. If your expectation is “only show to this audience,” you’ll feel like you’re reaching the wrong people—even though the campaign is behaving as designed.
The correct way to evaluate Performance Max is to focus on the quality of conversions (and the down-funnel value, if you import it), then adjust inputs and exclusions based on what’s actually producing business results.
8) Demographic targeting and the “Unknown” bucket can skew what you think is happening
Demographic reporting includes “Unknown” categories when age/gender (and in some campaign types, parental status or household income) can’t be identified. Many advertisers exclude “Unknown” to try to get cleaner traffic, then wonder why reach collapses—or why the remaining traffic looks oddly unbalanced. In most accounts, “Unknown” is a meaningful slice of the real world, not a junk drawer.
My rule: only exclude demographic categories when you have a clear performance case, and only after you’ve confirmed you’re not blocking too much scale or biasing delivery into a narrow subset that isn’t representative of your true buyers.
9) You may be in a category with targeting restrictions (and that changes your options)
Some verticals have strict restrictions on personalized advertising and demographic targeting (especially for sensitive categories and certain regulated areas). Additionally, personalized advertising protections apply to users under 18, which limits how narrowly you can target in certain contexts. If you’re advertising in housing, employment, or consumer finance in the United States, your ability to use certain demographic targeting and some location granularity can be restricted, and “why can’t I reach exactly who I want?” becomes partly a compliance reality, not just an optimization issue.
A practical, systematic plan to get your ads in front of the right people
Step 1: Fix intent first (Search campaigns)
If your Search terms show mismatched intent, don’t start by stacking audiences. Start by tightening the query layer. Clean targeting at the keyword level typically improves both lead quality and bidding efficiency.
- Use the Search terms report weekly and add negatives for themes that are clearly not your customer (jobs, free, DIY, definitions, competitors if irrelevant, parts if you sell services, etc.).
- Choose match types based on how precise the intent must be. If you’re selling something expensive or lead quality is critical, lean more on exact/phrase for your core terms, and expand carefully.
- If you use broad match, pair it with strong conversion measurement and a bidding strategy built to optimize for outcomes (not just clicks). Broad match can use additional intent signals, but it needs the right “reward function” (your conversion tracking) to learn what “right people” means.
- Align ad copy and landing pages to filter. Put qualifiers in ads (pricing, minimums, service area, “for businesses,” “for homeowners,” etc.) and mirror them on the landing page. This reduces curiosity clicks from the wrong segment.
Step 2: Fix reach next (Location, language, networks)
Once intent is under control, tighten reach so you’re not paying to educate people you can’t serve.
For local service businesses, I typically recommend testing a switch to a stricter location presence setting (people in or regularly in your locations) and then using exclusions to clean up any known problem areas. If you’re national or remote, keep the broader setting but monitor geographic performance and add exclusions where lead quality is consistently poor.
On language, avoid over-filtering unless you have a strong reason. On networks, separate Search from Display, and only keep Search Partners enabled if it’s proving incremental, profitable volume.
Step 3: Make Display/Video/Demand Gen targeting “real,” then decide how much automation you can tolerate
If you’re seeing irrelevant traffic on Display/Video/Demand Gen, the first fix is often structural: ensure every ad group has a clear targeting method and that you’re intentionally using Targeting vs Observation. Then decide whether optimized targeting belongs in that campaign based on your goal and the quality of your conversion signals.
If your conversion tracking is strong and your objective is efficient scale, optimized targeting can help find additional converters. If your conversion tracking is weak (or you’re optimizing to an easy but low-quality action), expansion will reliably find more of the wrong people—because you’re unintentionally telling the system those people are “right.”
Step 4: For Performance Max, treat “right people” as a measurement problem first
With Performance Max, your levers are more about inputs and feedback than strict targeting. If you want it to reach better users, prioritize: cleaner conversion definitions (quality over quantity), accurate values when possible, and thoughtful audience signals that represent your best customers. Expect it to go beyond signals, then judge it on business outcomes—not on whether every impression matches your mental picture of your audience.
Step 5: Build a “quality feedback loop” so the system can learn your true customer
The final step is what separates accounts that constantly fight audience mismatch from accounts that improve month after month: a quality feedback loop. If you generate leads, that means tracking qualified leads (not just form submits). If you sell e-commerce, it means optimizing to revenue and margin-aware values when feasible. The more clearly your measurement distinguishes “right people who buy” from “wrong people who click,” the less you’ll need to micromanage targeting—and the more the platform’s automation will work in your favor instead of against you.
