How Geo Splits Show Where Brand Budgets Get Wasted

Geo-level brand budget waste happens when national reporting treats every market as if it faces the same competitive pressure. Break brand performance out by geography, then compare cost-per-click (CPC), impression share lost to rank, and competitor activity to identify where you're paying more than necessary for brand coverage and where competitors create a real need for stronger defense.

Why National Brand Reporting Hides Two Opposite Problems

A healthy blended brand ROAS can hide two opposite problems: overspending where there is little competition and under-defending markets where competitors are actively bidding on the brand. The fix is to break the national average into the auction conditions of each market.

Brand campaigns are particularly good at hiding misallocation. The click-through rate (CTR) is usually high, conversion rates look strong, and return on ad spend (ROAS) can remain healthy even when the account is paying too much in quiet markets and not defending contested ones aggressively enough.

Consider a hypothetical brand campaign spending $100,000 per month at an 8x ROAS. At the national level, there is no obvious problem. Split the same spend geographically, though, and the picture can change:

The trap is that nobody reviewing the national number sees an obvious fire. The CMO sees an 8x brand ROAS in the quarterly review. The acquisition team sees strong conversion rates and relatively cheap clicks. And if automated bidding is in place, Google is optimizing each auction toward the campaign's bidding objective, not asking whether every geography deserves the same level of brand defense. None of those views, by themselves, answer whether the company is paying the right price to defend the brand in Iowa versus Los Angeles.

I have run a version of this test. In a consulting engagement with a company selling across states on both coasts, we pulled the national campaign apart into a West Coast bucket and an East Coast bucket, each with its own budget, and let them run. Blended, the account looked healthy at roughly 3x. Isolated, one coast was running near breakeven while the other was several times more profitable, and automated bidding had been following the side with more available volume rather than the side that actually converted. The caveat I would add now is that a two-cell coastal split is too coarse to be a real test. It was enough to prove the national average was hiding something. It was not enough to tell us which states inside each bucket were the problem, which is why the tiering work in this article starts at the state level.

While there are tools that can account for auction conditions automatically, lowering bids in uncontested brand auctions and defending when competition appears, most brands are still managing brand search through standard Google Ads campaign structures. For those teams, the practical starting point is identifying where competitive pressure actually exists before changing bids or budgets.

Break brand performance out by geography and compare each market's competitive pressure. The national ROAS tells you whether brand looks efficient overall. It does not tell you where the brand budget is actually needed.

Building a Geo Competition Matrix From Auction Insights and Impression Share

To measure brand competition by geography, look for markets where three things move together: competitors appear more often, brand impression share becomes harder to hold, and CPC rises. The hard part is that Google Ads does not package those signals into a clean geo-level competitive report, so the matrix has to be built from several views, and only at the geographic level the data can actually support.

Start with the brand campaign's location report. In Google Ads, go to Campaigns → Insights & reports → When and where your ads showed → Matched locations and pull the longest recent period that still represents the current business, often 30 to 90 days rather than a single week.

Start at the state level. Export impressions, clicks, cost, average CPC, conversions, and available impression share metrics. What you're looking for isn't simply the most expensive state. You're looking for outliers.

If the national brand CPC is $0.45 and one meaningful market is consistently running at $0.90, put it on the investigation list. If another market is spending heavily but maintaining extremely high impression share at a much lower CPC, flag that too. Neither proves anything yet. You're identifying where the auction behaves differently.

Before assuming competition caused the difference, check Search terms. If one market has picked up more brand-plus-product searches, informational queries, or broader matched traffic, its CPC can rise even if the competitive auction hasn't changed. Google's low-volume search-term suppression means this check isn't perfect. Terms without enough query activity are omitted from the report entirely for privacy reasons, and thin geos are exactly where that omission bites hardest, so a market can show a CPC shift with no visible query cause. Treat a clean search terms report in a low-volume market as unproven rather than as evidence that competition caused the change.

Then open Insights & reports → Auction Insights using the same date range. This is where the workflow stops being clean.

The Matched Locations report can tell you that California's average brand CPC is materially higher than Iowa's. One caveat before trusting that number: matched locations are not user locations. Google defines a matched location as a user's physical location or a location they showed interest in, so under the default Presence or interest setting, California's row can include people who were never in California. Google removed the old user-location view from this report, so the physical-location cut now has to be rebuilt as a custom report in Report Editor. If the two views disagree materially in a market, resolve that before putting the market in a tier.

Auction Insights can tell you that

Geo-level brand budget waste happens when national reporting treats every market as if it faces the same competitive pressure. Break brand performance out by geography, then compare cost-per-click (CPC), impression share lost to rank, and competitor activity to identify where you're paying more than necessary for brand coverage and where competitors create a real need for stronger defense.

Why National Brand Reporting Hides Two Opposite Problems

A healthy blended brand ROAS can hide two opposite problems: overspending where there is little competition and under-defending markets where competitors are actively bidding on the brand. The fix is to break the national average into the auction conditions of each market.

Brand campaigns are particularly good at hiding misallocation. The click-through rate (CTR) is usually high, conversion rates look strong, and return on ad spend (ROAS) can remain healthy even when the account is paying too much in quiet markets and not defending contested ones aggressively enough.

Consider a hypothetical brand campaign spending $100,000 per month at an 8x ROAS. At the national level, there is no obvious problem. Split the same spend geographically, though, and the picture can change:

The trap is that nobody reviewing the national number sees an obvious fire. The CMO sees an 8x brand ROAS in the quarterly review. The acquisition team sees strong conversion rates and relatively cheap clicks. And if automated bidding is in place, Google is optimizing each auction toward the campaign's bidding objective, not asking whether every geography deserves the same level of brand defense. None of those views, by themselves, answer whether the company is paying the right price to defend the brand in Iowa versus Los Angeles.

I have run a version of this test. In a consulting engagement with a company selling across states on both coasts, we pulled the national campaign apart into a West Coast bucket and an East Coast bucket, each with its own budget, and let them run. Blended, the account looked healthy at roughly 3x. Isolated, one coast was running near breakeven while the other was several times more profitable, and automated bidding had been following the side with more available volume rather than the side that actually converted. The caveat I would add now is that a two-cell coastal split is too coarse to be a real test. It was enough to prove the national average was hiding something. It was not enough to tell us which states inside each bucket were the problem, which is why the tiering work in this article starts at the state level.

While there are tools that can account for auction conditions automatically, lowering bids in uncontested brand auctions and defending when competition appears, most brands are still managing brand search through standard Google Ads campaign structures. For those teams, the practical starting point is identifying where competitive pressure actually exists before changing bids or budgets.

Break brand performance out by geography and compare each market's competitive pressure. The national ROAS tells you whether brand looks efficient overall. It does not tell you where the brand budget is actually needed.

Building a Geo Competition Matrix From Auction Insights and Impression Share

To measure brand competition by geography, look for markets where three things move together: competitors appear more often, brand impression share becomes harder to hold, and CPC rises. The hard part is that Google Ads does not package those signals into a clean geo-level competitive report, so the matrix has to be built from several views, and only at the geographic level the data can actually support.

Start with the brand campaign's location report. In Google Ads, go to Campaigns → Insights & reports → When and where your ads showed → Matched locations and pull the longest recent period that still represents the current business, often 30 to 90 days rather than a single week.

Start at the state level. Export impressions, clicks, cost, average CPC, conversions, and available impression share metrics. What you're looking for isn't simply the most expensive state. You're looking for outliers.

If the national brand CPC is $0.45 and one meaningful market is consistently running at $0.90, put it on the investigation list. If another market is spending heavily but maintaining extremely high impression share at a much lower CPC, flag that too. Neither proves anything yet. You're identifying where the auction behaves differently.

Before assuming competition caused the difference, check Search terms. If one market has picked up more brand-plus-product searches, informational queries, or broader matched traffic, its CPC can rise even if the competitive auction hasn't changed. Google's low-volume search-term suppression means this check isn't perfect. Terms without enough query activity are omitted from the report entirely for privacy reasons, and thin geos are exactly where that omission bites hardest, so a market can show a CPC shift with no visible query cause. Treat a clean search terms report in a low-volume market as unproven rather than as evidence that competition caused the change.

Then open Insights & reports → Auction Insights using the same date range. This is where the workflow stops being clean.

The Matched Locations report can tell you that California's average brand CPC is materially higher than Iowa's. One caveat before trusting that number: matched locations are not user locations. Google defines a matched location as a user's physical location or a location they showed interest in, so under the default Presence or interest setting, California's row can include people who were never in California. Google removed the old user-location view from this report, so the physical-location cut now has to be rebuilt as a custom report in Report Editor. If the two views disagree materially in a market, resolve that before putting the market in a tier.

Auction Insights can tell you that [competitor brand] is appearing more often in the same auctions as your brand. What Google does not give you in one national campaign is a standard Auction Insights location dimension that says [competitor brand] → California → 35% overlap.

That creates a reporting gap that Excel cannot fix. You can put the Matched Locations export in one tab and Auction Insights in another, but there is no state or designated market area (DMA) key to VLOOKUP if Auction Insights never contained that geography in the first place. A spreadsheet can join dimensions Google gave you; it cannot manufacture one Google withheld. The constraint is structural. Auction Insights can be segmented by device and by time, and that is the entire list. There is no location segment, and the report is still not exposed through the Google Ads API, so this is not something a script or a Looker Studio connector solves for you either. Whatever geographic resolution you get comes from how the account is structured, not from the report.

If the account is already separated geographically at the campaign or ad-group level, you have a common structure to work with. Pull Auction Insights for the geographically isolated campaign and compare it with that market's performance.

If everything sits inside one national brand campaign, be more careful. A rising California CPC and rising national competitor overlap are two useful clues, but they do not prove that competitor caused California's increase.

There is another reporting wall: Auction Insights only appears when traffic meets Google's eligibility thresholds. Google currently says Auction Insights is not shown when impression share is below 10%.

Thin markets can therefore be exactly where you want more competitive detail and exactly where Google gives you less of it. When that happens, roll the analysis up to a state or regional cluster rather than assigning a city-level tier from a handful of auctions.

That is why geo granularity should follow the data, not the ambition of the analysis. Start with states. Move to DMA only where volume supports it. Go to city only when there are enough auctions for the numbers to survive a sanity check against a longer date range. If a city's "competitive spike" disappears when you move from seven days to 30 or 60, it probably never deserved its own tier in the first place.

For each market you can defend analytically, keep three signals next to each other:

• Competitor overlap: How consistently are other advertisers showing when the brand shows?
• Lost impression share (IS) to rank: Is Ad Rank actually costing the brand eligible impressions?
• CPC delta: How much more or less are brand clicks costing compared with the account's normal baseline?

One caveat on the second signal: impression share is calculated against the impressions your campaign was eligible to receive, and Google's own guidance for reading it market by market is to run a separate campaign per location. In a single national brand campaign, treat lost impression share to rank as a signal you have when the geography is already isolated at campaign or ad-group level, and lean on competitor overlap and CPC delta when it isn't.

The combination matters more than any one number. A working matrix might look like this:

At this stage, don't worry about turning the matrix into a single competition score. Read across each row and ask whether the signals tell the same story.

Market A is the cleanest uncontested case. Competitor overlap is only 4%, almost no impression share is being lost to rank, and CPC is 12% below the brand baseline. None of those metrics proves the market is uncontested on its own, but together they make a strong case that the brand is paying to defend a relatively quiet auction.

Market C shows the opposite pattern. Competitors overlap in 37% of the auctions captured by Auction Insights, lost impression share to rank is 14%, and CPC is running 31% above baseline. All three signals point in the same direction. That's the kind of market that deserves a closer look at whether the current brand setup is providing enough defense.

Market B is where operator judgment matters. Competitor overlap is higher than Market A, but CPC is only 8% above baseline and lost impression share to rank is 6%. There is some competitive pressure, but the evidence isn't strong enough to make an aggressive bid or structural change from one reporting period. Keep it in the middle, watch whether those signals persist or move together, and resist turning normal auction movement into a strategy change.

When several signals agree over enough volume and time, the classification becomes more defensible. When they don't, the right decision may be to keep watching rather than force the market into a tier.

Once those patterns are visible, the next step is to define repeatable rules for turning them into uncontested, moderate, and contested tiers, and deciding whether any of those tiers actually deserve a different bid strategy.

Setting Geo Tiers Without Overreacting to Noise

Put a market into an uncontested, moderate, or contested tier only when the underlying signals support the classification. Competitor overlap, lost impression share to rank, CPC movement, and sufficient auction volume should determine the tier, not one metric crossing an arbitrary threshold.

Once the matrix is built, the next job is deciding what evidence is strong enough to change how much you're willing to pay for brand traffic.

Start with the obvious ends of the spectrum.

An uncontested market should show little competitor activity, little or no impression share lost to rank, and CPC at or below the normal brand baseline.

A contested market should show sustained competitor presence alongside evidence that the auction is getting more expensive or harder to win. Everything that doesn't clearly fit either description stays moderate.

A starting framework might look like this:

The important word is sustained.

Suppose a small market normally gets 200 brand impressions a month. A competitor appears in 20 auctions one week and CPC jumps. Technically, several metrics moved. Operationally, that's nowhere near enough evidence to rebuild the bidding strategy around.

A larger market showing the same pattern for four consecutive weeks is different. If competitor overlap remains elevated, CPC stays above its normal baseline, and impression share is being lost to rank, there is a much stronger case for treating that market as contested.

This is why a universal minimum such as "500 impressions" is hard to defend. Volume requirements should reflect the advertiser's normal brand traffic and the volatility of the market. Smaller geos need longer observation windows before the pattern means anything.

If the evidence isn't strong enough, leave the market in the moderate tier and keep collecting data.

Restructuring Campaigns or Using Bid Adjustments Without Fragmenting Data

Once a market clearly needs different treatment, the bidding strategy determines how much control you actually have. Manual CPC allows direct location bid adjustments; Smart Bidding does not give you the same lever, which can force a choice between keeping the market in the shared campaign or separating it.

This is where the geo analysis runs into Google Ads account structure.

With Manual CPC, the connection is fairly direct. If a market is uncontested, you can test a lower location bid adjustment and watch whether impression share holds. If a market is contested, you can bid more aggressively in that location without changing bids everywhere else.

Smart Bidding changes the equation. Google already uses location as an auction-time signal, so a traditional location bid adjustment does not give you the same direct control. It is more absolute than that. Under Target CPA, Target ROAS, Maximize conversions and Maximize conversion value, manual bid adjustments are not applied at all, with the device -100% exclusion as the only meaningful exception. Google does not remove them from the interface, so the field keeps showing a location adjustment that is doing nothing, which is why audits keep finding +30% location modifiers that were set two years ago on a Target ROAS campaign. Putting Texas in the contested tier does not mean adding a +25% location adjustment will make Target ROAS bid 25% harder there.

That leaves a structural question: can the market stay inside the shared campaign, or does it need its own campaign economics?

Before answering that, check the Search terms report one more time. A market with higher CPC may simply be picking up more brand-plus-product, informational, or broadly matched searches than other markets. If the query mix is different, fix or account for that first. Otherwise you're restructuring campaigns to solve what is really a search-term problem.

If the competitive difference is real, campaign separation gives you more control, but don't treat it like moving a row in a spreadsheet. Pull a major DMA out of a national campaign and you've changed the traffic and conversion mix feeding the original campaign while creating a new, thinner bidding environment for the DMA.

Google can put a bid strategy into Learning when campaigns, ad groups, or keywords are added to or removed from the strategy, and performance can fluctuate while bidding adjusts. Google calls this a composition change, and the part teams forget is that it applies to the campaign left behind as well as the new one. Pull a DMA out of a national brand campaign and both sides of the split can show Learning at the same time, which is the worst possible moment to be reading a week of performance data and drawing conclusions from it. Don't judge the new structure after an arbitrary seven or 14 days. Evaluate it over conversion cycles, especially when offline conversions arrive several days after the click.

Don't use a universal monthly conversion threshold to decide whether a geo can stand alone. The practical question is whether enough conversion data survives the split to evaluate the new campaign over a reasonable number of conversion cycles. A DMA producing conversions every day is a very different candidate from one producing three conversions a month, even if both look equally contested.

So don't split a market simply because its CPC is higher. Use two tests: How different is the auction, and how much data survives the split?

The last option is often the practical middle ground. If five markets consistently behave like contested auctions, they don't necessarily need five campaigns. A single contested-market campaign can create a separate budget and bidding environment while preserving more conversion density.

Before launching the new structure, open the campaign's Location options. Google's default Search targeting can include people who are in the location or have shown interest in it. That may be broader than you want when the entire purpose of the campaign is to isolate a contested market.

If the campaign is supposed to isolate people actually in or regularly in that market, evaluate whether Presence targeting better matches the strategy. Otherwise a campaign built specifically to defend New York can still pick up users outside New York based on location interest. Even Presence is not a perfect geographic fence, so check the Matched Locations report after launch rather than assuming the setting worked exactly as intended. Expect that switch to cost impressions. Google's own guidance is that most campaigns see a decrease when they move off Presence or interest, so the honest tradeoff is narrower, more accurate geographic data against less brand coverage in that market. On a contested-market campaign built specifically to measure and defend one geography, that trade is usually worth making. On a thin market already short of auctions, it can push the campaign below the volume you needed for the analysis in the first place.

After restructuring, watch search impression share, lost impression share to rank, CPC, conversion volume, and Auction Insights, not just ROAS. If the contested campaign gains visibility but CPC rises materially without a meaningful improvement in defended demand, more control has simply turned into more expensive brand traffic.

Re-Tiering Monthly Without Creating Whiplash

Review geo tiers monthly, but don't move a market because of one bad week. Re-tier only when competitive pressure changes for several weeks and more than one signal confirms it; otherwise normal auction noise quickly turns into unnecessary bid and campaign changes.

Brand auctions can change quickly. A competitor launches a promotion, pushes harder for a few days, and disappears. CPC jumps. Auction Insights suddenly looks ugly. Two weeks later, everything is back to normal.

If the team re-tiers on that first spike, it ends up chasing the auction instead of managing it.

Review the tiers monthly using a rolling 28-day window against the prior 28 days. The comparison won't remove seasonality or auction noise, but it makes a one-week spike less likely to drive a structural account change.

Look at the same signals used to classify the market in the first place: competitor overlap, CPC versus the brand baseline, and impression share lost to rank.

Don't move a market because one metric changed. Look for at least two signals moving in the same direction for multiple weeks.

For example, if competitor overlap rises for three consecutive weeks, CPC remains above its normal range, and lost impression share to rank starts climbing, there is a reasonable case that the auction has changed. If CPC spikes for five days while overlap and impression share remain normal, leave the tier alone.

The same rule applies on the way down. A contested market shouldn't become "uncontested" because a competitor disappeared for a week. Require the quieter auction conditions to persist before reducing defensive pressure.

Seasonality needs its own sanity check. Black Friday, a major product launch, or a competitor promotion can temporarily change brand auctions without representing a permanent shift. Annotate those periods and compare them with the same business cycle where possible. Otherwise the team can end up baking a temporary event into the next month's bidding strategy.

CPC baselines also need maintenance. Search CPCs move over time, so a market shouldn't be labeled more competitive simply because paid search became more expensive everywhere. Recalculate the account's brand CPC baseline quarterly and measure each geo against the updated baseline rather than a number established the previous year.

A monthly review should be boring:

• Refresh the geo performance report using the same reporting window.
• Compare the latest four weeks with the previous four.
• Check CPC, competitor overlap, and lost impression share to rank together.
• Flag one-week spikes rather than immediately re-tiering them.
• Check promotions, seasonality, and major account changes before blaming competition.
• Move a market only when the change persists across multiple weeks and sufficient volume.
• Recalibrate the CPC baseline quarterly.
• Document every tier change and the evidence behind it.

That last step saves more pain than it sounds. Three months later, when someone asks why Texas was moved into the contested tier, there should be an answer beyond "the dashboard looked bad."

Monthly review does not mean monthly changes. The purpose of the cadence is to catch real changes in competitive pressure without rebuilding the account every time the auction gets noisy.

Break the national average apart, look for markets where competitor activity, CPC, and impression share consistently tell the same story, and only change bids or structure when the evidence is strong enough to justify it. That gives acquisition teams a much clearer answer to the question that blended brand ROAS can't answer: where are we paying to defend the brand, and where does that defense actually matter?

Frequently Asked Questions

How do I measure competitor pressure by geo in brand?

Use geographic performance data alongside Auction Insights to look for markets where competitor overlap, CPC, and impression share lost to rank move together. Start at the state level and move to DMA or city only when volume supports it. Multiple sustained signals are more reliable than treating any single metric as proof of competition.

What rules decide brand bids by market tier?

Classify markets as uncontested, moderate, or contested based on sustained competitor activity, CPC movement, and impression share lost to rank. Low-volume markets should remain in the default tier until enough data accumulates. The tier should guide bidding decisions, but the available controls depend on whether the campaign uses manual or automated bidding.

How do I avoid under-defending contested markets?

Keep contested markets in the shared campaign when separating them would leave too little conversion data to evaluate performance reliably. When competitive pressure is sustained and the market has enough conversion activity to stand on its own, a separate campaign can provide greater control over budgets and bidding targets without unnecessarily fragmenting the rest of the brand campaign.

is appearing more often in the same auctions as your brand. What Google does not give you in one national campaign is a standard Auction Insights location dimension that says competitor.com → California → 35% overlap.

That creates a reporting gap that Excel cannot fix. You can put the Matched Locations export in one tab and Auction Insights in another, but there is no state or designated market area (DMA) key to VLOOKUP if Auction Insights never contained that geography in the first place. A spreadsheet can join dimensions Google gave you; it cannot manufacture one Google withheld. The constraint is structural. Auction Insights can be segmented by device and by time, and that is the entire list. There is no location segment, and the report is still not exposed through the Google Ads API, so this is not something a script or a Looker Studio connector solves for you either. Whatever geographic resolution you get comes from how the account is structured, not from the report.

If the account is already separated geographically at the campaign or ad-group level, you have a common structure to work with. Pull Auction Insights for the geographically isolated campaign and compare it with that market's performance.

If everything sits inside one national brand campaign, be more careful. A rising California CPC and rising national competitor overlap are two useful clues, but they do not prove that competitor caused California's increase.

There is another reporting wall: Auction Insights only appears when traffic meets Google's eligibility thresholds. Google currently says Auction Insights is not shown when impression share is below 10%.

Thin markets can therefore be exactly where you want more competitive detail and exactly where Google gives you less of it. When that happens, roll the analysis up to a state or regional cluster rather than assigning a city-level tier from a handful of auctions.

That is why geo granularity should follow the data, not the ambition of the analysis. Start with states. Move to DMA only where volume supports it. Go to city only when there are enough auctions for the numbers to survive a sanity check against a longer date range. If a city's "competitive spike" disappears when you move from seven days to 30 or 60, it probably never deserved its own tier in the first place.

For each market you can defend analytically, keep three signals next to each other:

• Competitor overlap: How consistently are other advertisers showing when the brand shows?
• Lost impression share (IS) to rank: Is Ad Rank actually costing the brand eligible impressions?
• CPC delta: How much more or less are brand clicks costing compared with the account's normal baseline?

One caveat on the second signal: impression share is calculated against the impressions your campaign was eligible to receive, and Google's own guidance for reading it market by market is to run a separate campaign per location. In a single national brand campaign, treat lost impression share to rank as a signal you have when the geography is already isolated at campaign or ad-group level, and lean on competitor overlap and CPC delta when it isn't.

The combination matters more than any one number. A working matrix might look like this:

At this stage, don't worry about turning the matrix into a single competition score. Read across each row and ask whether the signals tell the same story.

Market A is the cleanest uncontested case. Competitor overlap is only 4%, almost no impression share is being lost to rank, and CPC is 12% below the brand baseline. None of those metrics proves the market is uncontested on its own, but together they make a strong case that the brand is paying to defend a relatively quiet auction.

Market C shows the opposite pattern. Competitors overlap in 37% of the auctions captured by Auction Insights, lost impression share to rank is 14%, and CPC is running 31% above baseline. All three signals point in the same direction. That's the kind of market that deserves a closer look at whether the current brand setup is providing enough defense.

Market B is where operator judgment matters. Competitor overlap is higher than Market A, but CPC is only 8% above baseline and lost impression share to rank is 6%. There is some competitive pressure, but the evidence isn't strong enough to make an aggressive bid or structural change from one reporting period. Keep it in the middle, watch whether those signals persist or move together, and resist turning normal auction movement into a strategy change.

When several signals agree over enough volume and time, the classification becomes more defensible. When they don't, the right decision may be to keep watching rather than force the market into a tier.

Once those patterns are visible, the next step is to define repeatable rules for turning them into uncontested, moderate, and contested tiers, and deciding whether any of those tiers actually deserve a different bid strategy.

Setting Geo Tiers Without Overreacting to Noise

Put a market into an uncontested, moderate, or contested tier only when the underlying signals support the classification. Competitor overlap, lost impression share to rank, CPC movement, and sufficient auction volume should determine the tier, not one metric crossing an arbitrary threshold.

Once the matrix is built, the next job is deciding what evidence is strong enough to change how much you're willing to pay for brand traffic.

Start with the obvious ends of the spectrum.

An uncontested market should show little competitor activity, little or no impression share lost to rank, and CPC at or below the normal brand baseline.

A contested market should show sustained competitor presence alongside evidence that the auction is getting more expensive or harder to win. Everything that doesn't clearly fit either description stays moderate.

A starting framework might look like this:

The important word is sustained.

Suppose a small market normally gets 200 brand impressions a month. A competitor appears in 20 auctions one week and CPC jumps. Technically, several metrics moved. Operationally, that's nowhere near enough evidence to rebuild the bidding strategy around.

A larger market showing the same pattern for four consecutive weeks is different. If competitor overlap remains elevated, CPC stays above its normal baseline, and impression share is being lost to rank, there is a much stronger case for treating that market as contested.

This is why a universal minimum such as "500 impressions" is hard to defend. Volume requirements should reflect the advertiser's normal brand traffic and the volatility of the market. Smaller geos need longer observation windows before the pattern means anything.

If the evidence isn't strong enough, leave the market in the moderate tier and keep collecting data.

Restructuring Campaigns or Using Bid Adjustments Without Fragmenting Data

Once a market clearly needs different treatment, the bidding strategy determines how much control you actually have. Manual CPC allows direct location bid adjustments; Smart Bidding does not give you the same lever, which can force a choice between keeping the market in the shared campaign or separating it.

This is where the geo analysis runs into Google Ads account structure.

With Manual CPC, the connection is fairly direct. If a market is uncontested, you can test a lower location bid adjustment and watch whether impression share holds. If a market is contested, you can bid more aggressively in that location without changing bids everywhere else.

Smart Bidding changes the equation. Google already uses location as an auction-time signal, so a traditional location bid adjustment does not give you the same direct control. It is more absolute than that. Under Target CPA, Target ROAS, Maximize conversions and Maximize conversion value, manual bid adjustments are not applied at all, with the device -100% exclusion as the only meaningful exception. Google does not remove them from the interface, so the field keeps showing a location adjustment that is doing nothing, which is why audits keep finding +30% location modifiers that were set two years ago on a Target ROAS campaign. Putting Texas in the contested tier does not mean adding a +25% location adjustment will make Target ROAS bid 25% harder there.

That leaves a structural question: can the market stay inside the shared campaign, or does it need its own campaign economics?

Before answering that, check the Search terms report one more time. A market with higher CPC may simply be picking up more brand-plus-product, informational, or broadly matched searches than other markets. If the query mix is different, fix or account for that first. Otherwise you're restructuring campaigns to solve what is really a search-term problem.

If the competitive difference is real, campaign separation gives you more control, but don't treat it like moving a row in a spreadsheet. Pull a major DMA out of a national campaign and you've changed the traffic and conversion mix feeding the original campaign while creating a new, thinner bidding environment for the DMA.

Google can put a bid strategy into Learning when campaigns, ad groups, or keywords are added to or removed from the strategy, and performance can fluctuate while bidding adjusts. Google calls this a composition change, and the part teams forget is that it applies to the campaign left behind as well as the new one. Pull a DMA out of a national brand campaign and both sides of the split can show Learning at the same time, which is the worst possible moment to be reading a week of performance data and drawing conclusions from it. Don't judge the new structure after an arbitrary seven or 14 days. Evaluate it over conversion cycles, especially when offline conversions arrive several days after the click.

Don't use a universal monthly conversion threshold to decide whether a geo can stand alone. The practical question is whether enough conversion data survives the split to evaluate the new campaign over a reasonable number of conversion cycles. A DMA producing conversions every day is a very different candidate from one producing three conversions a month, even if both look equally contested.

So don't split a market simply because its CPC is higher. Use two tests: How different is the auction, and how much data survives the split?

The last option is often the practical middle ground. If five markets consistently behave like contested auctions, they don't necessarily need five campaigns. A single contested-market campaign can create a separate budget and bidding environment while preserving more conversion density.

Before launching the new structure, open the campaign's Location options. Google's default Search targeting can include people who are in the location or have shown interest in it. That may be broader than you want when the entire purpose of the campaign is to isolate a contested market.

If the campaign is supposed to isolate people actually in or regularly in that market, evaluate whether Presence targeting better matches the strategy. Otherwise a campaign built specifically to defend New York can still pick up users outside New York based on location interest. Even Presence is not a perfect geographic fence, so check the Matched Locations report after launch rather than assuming the setting worked exactly as intended. Expect that switch to cost impressions. Google's own guidance is that most campaigns see a decrease when they move off Presence or interest, so the honest tradeoff is narrower, more accurate geographic data against less brand coverage in that market. On a contested-market campaign built specifically to measure and defend one geography, that trade is usually worth making. On a thin market already short of auctions, it can push the campaign below the volume you needed for the analysis in the first place.

After restructuring, watch search impression share, lost impression share to rank, CPC, conversion volume, and Auction Insights, not just ROAS. If the contested campaign gains visibility but CPC rises materially without a meaningful improvement in defended demand, more control has simply turned into more expensive brand traffic.

Re-Tiering Monthly Without Creating Whiplash

Review geo tiers monthly, but don't move a market because of one bad week. Re-tier only when competitive pressure changes for several weeks and more than one signal confirms it; otherwise normal auction noise quickly turns into unnecessary bid and campaign changes.

Brand auctions can change quickly. A competitor launches a promotion, pushes harder for a few days, and disappears. CPC jumps. Auction Insights suddenly looks ugly. Two weeks later, everything is back to normal.

If the team re-tiers on that first spike, it ends up chasing the auction instead of managing it.

Review the tiers monthly using a rolling 28-day window against the prior 28 days. The comparison won't remove seasonality or auction noise, but it makes a one-week spike less likely to drive a structural account change.

Look at the same signals used to classify the market in the first place: competitor overlap, CPC versus the brand baseline, and impression share lost to rank.

Don't move a market because one metric changed. Look for at least two signals moving in the same direction for multiple weeks.

For example, if competitor overlap rises for three consecutive weeks, CPC remains above its normal range, and lost impression share to rank starts climbing, there is a reasonable case that the auction has changed. If CPC spikes for five days while overlap and impression share remain normal, leave the tier alone.

The same rule applies on the way down. A contested market shouldn't become "uncontested" because a competitor disappeared for a week. Require the quieter auction conditions to persist before reducing defensive pressure.

Seasonality needs its own sanity check. Black Friday, a major product launch, or a competitor promotion can temporarily change brand auctions without representing a permanent shift. Annotate those periods and compare them with the same business cycle where possible. Otherwise the team can end up baking a temporary event into the next month's bidding strategy.

CPC baselines also need maintenance. Search CPCs move over time, so a market shouldn't be labeled more competitive simply because paid search became more expensive everywhere. Recalculate the account's brand CPC baseline quarterly and measure each geo against the updated baseline rather than a number established the previous year.

A monthly review should be boring:

• Refresh the geo performance report using the same reporting window.
• Compare the latest four weeks with the previous four.
• Check CPC, competitor overlap, and lost impression share to rank together.
• Flag one-week spikes rather than immediately re-tiering them.
• Check promotions, seasonality, and major account changes before blaming competition.
• Move a market only when the change persists across multiple weeks and sufficient volume.
• Recalibrate the CPC baseline quarterly.
• Document every tier change and the evidence behind it.

That last step saves more pain than it sounds. Three months later, when someone asks why Texas was moved into the contested tier, there should be an answer beyond "the dashboard looked bad."

Monthly review does not mean monthly changes. The purpose of the cadence is to catch real changes in competitive pressure without rebuilding the account every time the auction gets noisy.

Break the national average apart, look for markets where competitor activity, CPC, and impression share consistently tell the same story, and only change bids or structure when the evidence is strong enough to justify it. That gives acquisition teams a much clearer answer to the question that blended brand ROAS can't answer: where are we paying to defend the brand, and where does that defense actually matter?

Frequently Asked Questions

How do I measure competitor pressure by geo in brand?

Use geographic performance data alongside Auction Insights to look for markets where competitor overlap, CPC, and impression share lost to rank move together. Start at the state level and move to DMA or city only when volume supports it. Multiple sustained signals are more reliable than treating any single metric as proof of competition.

What rules decide brand bids by market tier?

Classify markets as uncontested, moderate, or contested based on sustained competitor activity, CPC movement, and impression share lost to rank. Low-volume markets should remain in the default tier until enough data accumulates. The tier should guide bidding decisions, but the available controls depend on whether the campaign uses manual or automated bidding.

How do I avoid under-defending contested markets?

Keep contested markets in the shared campaign when separating them would leave too little conversion data to evaluate performance reliably. When competitive pressure is sustained and the market has enough conversion activity to stand on its own, a separate campaign can provide greater control over budgets and bidding targets without unnecessarily fragmenting the rest of the brand campaign.