Searches that Google matched to phrase-match keywords for the first time cost 28% more per lead than the searches those same keywords were already matching, according to a study New Path Digital published on October 27, 2026. The High Point, North Carolina, advertising agency examined nearly 50,000 keyword-and-search-term pairs from three Google Ads lead-generation accounts it manages, comparing what each keyword matched in 2022-23 with what it matched in 2025-26. Exact match moved the other way: conversions from newly matched exact-match queries cost 7% less than those from existing ones. The study, written by Digital Marketing Manager Mary Rippe, was last updated on November 5, 2026.

In Short

New Path Digital compared the searches three of its clients' Google Ads keywords matched in 2022-23 with the new ones the same keywords picked up in 2025-26, and found that new phrase-match searches cost $209 per lead against $163 for the old ones. It matters for lead-generation advertisers because the extra cost came from new searches converting less often, not from pricier clicks, while new exact-match searches cost $109 per lead against $116. The agency calls the work a field report rather than a benchmark: three accounts, correlations only, and every figure a cost per form fill rather than per qualified lead.

What the study measured

New Path Digital drew the data from three accounts it manages: a law firm, a higher-education advertiser and a local B2B service business. The figures were pooled, so every number in the study blends the three accounts and is split only by match type. Performance figures cover April 2025 through March 2026. The query comparison sets two periods against each other, labelled 2022-23 and 2025-26; the study does not give the start and end months of the earlier period.

The comparison rests on two labels. "We call queries 'existing' or 'new' throughout," the study states. An existing query appeared in the search terms report for the exact same keyword in both periods. A new query triggered that keyword only in 2025-26. A keyword qualified for the analysis only if it had impressions in both periods, with the account, campaign, match type and keyword all unchanged. The design holds the keyword constant and asks what Google started matching it to.

Three further conditions shape the numbers. The leads are form fills generated by non-brand keywords. Every campaign ran on Maximize Conversions, one of Google's Smart Bidding strategies. And each keyword-and-search-term pair was counted separately, so one search term can appear more than once if it matched several keywords.

The headline cost figures are not simple pooled averages. The three accounts have very different costs per lead, and new queries made up a different share of each. According to the study, pooling everything let that mix distort the answer: one pricier account supplied 24% of new exact-match conversions but only 6% of existing ones, which made new exact-match queries look more expensive in the pooled data even though they were cheaper in both accounts that ran exact match. To correct for this, the agency weighted each account's new-query cost per lead by its share of existing-query conversions. In the study's own words, the adjustment answers the question: "what would new queries have cost if they'd come from the same account mix as existing ones?" Click price and conversion rate were adjusted in the same way. The study does not publish the unadjusted pooled figures.

Cost and volume by match type

Metric Exact, existing queries Exact, new queries Phrase, existing queries Phrase, new queries
Cost per lead (account-mix adjusted) $116 $109 $163 $209
Change, new vs existing - -7% - +28%
Average click price $11 $10 $16 $16
Conversion rate 10% 9% 10% 8%
Keyword-and-search-term pairs, 2025-26 keywords 2,013 6,129 3,430 38,332
Pairs that spent money 588 1,055 649 3,234
Conversions 344 189 190 375
Share of pairs that spent money (PPC Land calculation) 29.2% 17.2% 18.9% 8.4%

Source: New Path Digital, "Google Ads Match Types: What Looser Semantic Matching Costs You [2026 Study]," October 27, 2026, updated November 5, 2026. The four pair counts sum to 49,904 and the conversions to 1,098 (PPC Land calculation).

Phrase match: same click price, fewer conversions

The phrase-match result is the larger of the two. Conversions from new phrase-match queries cost $209 each, against $163 for queries the same keywords already matched. The gap did not come from what Google charged per click. Clicks on new and existing phrase-match queries cost the same, about $16. The difference came from conversion rate, which fell from 10% on existing queries to 8% on new ones.

New queries also dominate phrase-match spending. According to the study, conversions from new phrase-match queries "ate up ~75% of phrase match cost." The pair counts point the same way. New phrase-match pairs number 38,332, or 92% of all phrase-match pairs and 77% of every pair in the study, by PPC Land's calculation. Most of them never spent anything: only 3,234 of those pairs, or 8.4%, recorded any cost. The 375 conversions they produced amount to 66% of phrase-match conversions in the raw counts, a smaller share than the roughly 75% of phrase-match cost the study assigns to them. Those raw counts are not adjusted for account mix, so the comparison is indicative rather than exact.

The study gives one example of the drift. The phrase keyword "veterans legal services" matched the new query "jag fort bragg nc." The two share no words. The agency's explanation is that JAG offices, the military's legal service, typically provide free legal help to service members, so the search comes from someone looking for a free option rather than a paying client. "This suggests semantic phrase matching may be a rougher fit for lead generation businesses," the study concludes.

Exact match: cheaper clicks, slightly lower conversion

Exact match moved in the opposite direction. Conversions from new exact-match queries cost $109, against $116 for existing queries, a 7% reduction according to the study. Here the click price did the work. New exact-match clicks averaged $10, compared with $11 for existing ones, and that saving more than covered a slightly lower conversion rate of 9% against 10%.

The agency's example is a legal keyword. "Military disability lawyer" matched the new, longer query "va lawyer for disability," and those clicks cost about $13 each, which the study puts at roughly 40% below the keyword's average. "Our read: when Google stretches an exact keyword to a query with the same intent," the study states, the result may be "a similar searcher at a lower price."

The dollar figures as published are rounded. Taken at face value, $109 against $116 is a 6% difference by PPC Land's calculation; the study's 7% is consistent with unrounded values within half a dollar of each published figure.

How far each match type strays

The second half of the study asks how far Google stretches each match type. The agency sorted every impression in both periods into five buckets by comparing each query with the closest keyword of the same match type in its campaign. That covered 75,698 exact-match and 119,808 phrase-match impressions in 2022-23, and 74,786 exact-match and 145,453 phrase-match impressions in 2025-26. Phrase-match impressions in the analysis rose 21% between the periods while exact-match impressions fell 1%, by PPC Land's calculation.

How the search query relates to its keyword, share of query volume

Relationship to keyword Exact 2022-23 Exact 2025-26 Phrase 2022-23 Phrase 2025-26
Identical to the keyword 29% 35% 7% 5%
Identical, with extra words 12% 9% 18% 10%
Same structure, with a one- or two-word synonym 10% 10% 6% 5%
Shares some words 49% 46% 68% 70%
No words in common <1% <1% 2% 9%
Impressions analysed 75,698 74,786 119,808 145,453

Source: New Path Digital. Shares are weighted by impressions and describe total query volume, not distinct queries. Columns may not sum to 100% because of rounding.

"Surprisingly, exact match didn't loosen across our client accounts," the study states. Queries identical to the keyword rose from 29% to 35% of exact-match volume. Almost half of exact-match volume, 46%, shares only some of the keyword's words, but 54% is the keyword itself, the keyword with extra words, or a one- or two-word synonym swap.

"Phrase match is where the drift shows," according to the study. Queries sharing some or no words with the keyword grew from 70% to 79% of phrase-match volume, while queries that were the keyword with extra words or a synonym fell from 24% to 15%. The share with no words in common rose from 2% to 9%. Phrase match, in its original design, showed an ad when the keyword appeared inside a longer search; by the study's own count, only 5% of phrase-match volume in 2025-26 was the keyword word for word, and a further 10% was the keyword with extra words.

The choice of the closest keyword matters for these shares. Google credits each query to a single keyword even when several could have matched, so the agency compared each query with the nearest keyword of the same match type in the campaign rather than only the credited one. "Using only the credited keyword, 31% of exact-match volume was identical," the study states, against 35% under the closest-keyword method. The 2022-23 shares use the same method.

Close to the retired broad match modifier

The study's most quotable comparison is with broad match modifier, which Google retired in 2021. Google announced on February 4, 2021 that phrase match would absorb broad match modifier traffic, starting on February 18, 2021, and from July 2021 new BMM keywords could no longer be created. The study defines the old rule plainly: "BMM required every keyword word (or a close variant) to appear in the query, in any order, with extra words allowed." Run against that rule, 75% of exact-match query volume passes, compared with 41% of phrase-match query volume.

The remaining 25% of exact-match volume did not contain every word of the keyword. The study describes that slice as "Google reading meaning where BMM would have needed the words." Its conclusion is that exact match now sits in a middle ground: "So think of exact match as looser than literal exact, tighter than phrase, and BMM-like with some semantic overflow."

The agency also restates the 2025-26 exact-match volume in terms of the jobs the match types were built to do. About 45% is the keyword or a near twin, meaning the same words or a swap such as "attorney" for "lawyer." About 9% is the keyword inside a longer search, which the study calls the phrase-match job. The remaining 46% shares some of the keyword's words without the same shape.

The result was not uniform

The pooled figures hide large differences between the three accounts. New phrase-match queries cost about 75% more per lead in one account, 29% more in another and about 3% more in the third. The study says the phrase result is concentrated in two of the three accounts. Only two of the three accounts ran exact match, and in both of them new exact-match queries were cheaper, by about 7% to 8%. The smallest account has only about 100 conversions behind its figures. The study does not say which account is the law firm, the higher-education advertiser or the B2B service business in this breakdown.

The agency is explicit about the limits. "These are correlations. Search behavior, negatives, budgets, bidding, and Google's own matching changes could all have shifted between periods, and three accounts make this a field report rather than a benchmark," the study states. Its own FAQ answers the question of whether the findings are a Google Ads benchmark with: "No."

A second caveat concerns what counts as a lead. "Every figure here is cost per form fill, not cost per qualified lead, enrollment, sale, or signed deal," the study states. A form fill from the JAG search and a form fill from a paying client carry the same weight in every number above.

A third concerns visibility. "Nearly 43% of eligible search terms were hidden by Google, mostly on phrase match," according to the study. The agency could analyse only the queries visible in the search terms report, so it notes that the "shares some words" and "no words in common" shares for phrase match could be larger than reported.

The audit the study describes

New Path Digital closes with a description of how it runs what it calls a new-query audit. According to the study, the process has six steps:

  1. Export search terms with keyword, match type, cost and conversions.
  2. Label each keyword-and-search-term pair as existing or new.
  3. Compare cost per lead within each match type before looking at totals.
  4. Add CRM outcomes.
  5. Review the highest-spend new queries for intent.
  6. Repeat the exercise every six months.

The agency pairs that audit with four positions of its own. It argues that a keyword marks the center of an intent and the match type sets how far Google can wander from it. It argues that negatives and ad copy steer the system, and that headlines in responsive search ads should be written, and the important ones pinned, so that every combination states what the business does. "Creative is targeting in 2026, and paid search advertisers can borrow the mindset Meta advertisers have had for years," the study states. It argues for rewarding qualified leads rather than raw form fills, using conversion adjustments to retract conversions that missed the mark and offline conversion import to feed back CRM outcomes. And it argues that smaller accounts, or larger ones just starting out, launch on exact match before adding phrase match. Those are the agency's recommendations, drawn from three of its own client accounts.

Twelve years of looser matching

The drift the study measures has a long record. Google stopped offering exact match without close variants in 2014, an option Judge Amit Mehta declined to restore in his September 2, 2025 remedies ruling in the US search case. On September 6, 2018, Google extended exact-match close variants to queries with the same meaning, citing about 3% more exact-match clicks and conversions in its own tests. Adobe advised advertisers on the change the following month, recommending closer monitoring of search query reports and new negative keywords where needed. The New Path Digital study summarises Google's current position the same way: an exact-match keyword can trigger ads on searches with the same meaning or intent, even when the wording differs, and close variants cannot be switched off.

Visibility has narrowed over the same period. In September 2020, Google limited the search terms report to terms searched by a significant number of users, without defining the threshold. Seer Interactive measured the share of search spend with visible terms falling from 98.7% on August 31, 2020 to 71.0% the next day, and Google partly reversed the change in September 2021. Later estimates put hidden traffic at about 40% of clicks and spend (Mark Meyerson of One Egg Digital, 2024) and at a median 44.0% of clicks in the separate category report across 88 accounts (Doctor Ads, August 2026). New Path Digital's figure of nearly 43% sits in the same range, although it counts search terms rather than clicks or spend.

Practitioner evidence has pointed in more than one direction. In November 2025, Brian Lasonde of PPC Boost reported that the exact keyword "best hypoallergenic food for dogs" matched 13 queries, none containing the word "hypoallergenic," all labelled as exact-match close variants. A four-month test by Xavier Mantica, published the same month, put cost per conversion at $43.97 for phrase match against $97.67 for phrase close variants, and $52.69 for exact match against $61.65 for exact close variants. Mantica compared match types and close variants rather than existing and newly matched queries, so the two sets of figures are not directly comparable. On phrase match they point the same way. On exact match they diverge: Mantica's close variants cost more, while New Path Digital's new exact-match queries cost less.

Since then the pressure has moved to AI Max for Search. Across 383 million EMEA ecommerce impressions in AI Max-activated campaigns, smec's Mike Ryan found that the share served by genuine exact match fell to about 71% by July 2026, with AI Max-broadened matching taking about 29%. On August 5, 2026, Google told advertisers that campaigns using the campaign-level broad match setting would upgrade to AI Max from September 1. In October the search terms report gained four source columns, including one that labels where each search term match came from. Google's documentation also says that, for searches in AI Mode, AI Overviews, Lens and auto-complete, the reported search term is the best approximation of the user's intent rather than the words typed, and in September Google began testing text ads in AI Mode for exact and phrase keywords. The New Path Digital study does not mention AI Max, and its performance window closed in March 2026, before the September migration.

Microsoft Advertising has moved in two directions at once. In December 2025 it said an exact-match keyword takes priority over Performance Max for the same query, regardless of Ad Rank. In July 2026 it said Predictive Matching is moving to search term matching within AI Max.

Lead quality is the other open question. Google's help page on invalid leads states that its filters do not prevent invalid leads on their own, and lists reCAPTCHA, double opt-in email confirmation and server-side validation as measures advertisers add themselves. In May, Chris Chambers of Understory argued that lead generation, with its lower conversion volumes, is more exposed than ecommerce to changes in search term reporting.

Why this matters for advertisers

The first point is the vertical. The AI Max figures smec published came from EMEA ecommerce accounts, whereas New Path Digital's study covers lead generation: a law firm, a higher-education advertiser and a B2B service business, with conversions counted as form fills. In that setting a loosely matched search does not only risk a wasted click; it can produce an enquiry from someone unlikely to become a paying client, and the JAG example shows the mechanism. The query shared no words with the keyword and came from a searcher looking for free help.

The source of the 28% gap is the second point. Every campaign in the study ran on Maximize Conversions, a strategy in which Google sets the bid for each auction. New phrase-match clicks nonetheless cost the same on average as existing ones, $16, while converting at 8% instead of 10%. The study does not examine why the click price did not fall with the conversion rate.

Scale is the third. In these three accounts, about three-quarters of phrase-match spending went to searches the same keywords had not matched in 2022-23, and new phrase-match pairs make up 77% of all pairs in the data. Any change in how those searches convert therefore moves the whole phrase-match line.

The exact-match result sits apart from the examples above. In the three accounts, queries with no words in common with the keyword stayed below 1% of exact-match volume in both periods, and the share identical to the keyword rose. That is a narrower finding than the 13 unrelated queries Lasonde documented, and it comes from a far smaller pool than smec's 383 million impressions, but its design holds each keyword constant across two periods three years apart.

What the study does not establish

The study does not give the months covered by its 2022-23 period, spend per account, the unadjusted pooled figures, confidence intervals or tests of statistical significance, or which account produced the 75%, 29% and 3% phrase-match results. It does not say whether broad match keywords, AI Max or Performance Max ran alongside the analysed campaigns, and it does not follow any lead past the form fill. Its headline figure that 35% of exact-match volume was identical to the keyword uses the closest-keyword method; under Google's own keyword crediting the figure is 31%.

Key facts

  • Study: "Google Ads Match Types: What Looser Semantic Matching Costs You [2026 Study]" (page title: "Google Ads Match Type: What Looser Matching Costs You [2026]")
  • Publisher: New Path Digital, advertising agency, High Point, North Carolina (founded 2015, according to the agency)
  • Author: Mary Rippe, Digital Marketing Manager
  • Published: October 27, 2026; last updated November 5, 2026
  • URL: https://newpathdigital.com/blog/google-ads-match-types-new-query-study/
  • Sample: nearly 50,000 exact and phrase keyword-and-search-term pairs (49,904 in the published counts, PPC Land calculation) from three lead-generation accounts: a law firm, a higher-education advertiser and a local B2B service business
  • Comparison periods: 2022-23 and 2025-26; performance figures April 2025 through March 2026
  • Conversions: form fills from non-brand keywords; every campaign on Maximize Conversions
  • Definitions: existing query = matched the same keyword in both periods; new query = matched that keyword only in 2025-26; keywords qualified only with impressions in both periods
  • Adjustment: new-query metrics weighted by each account's share of existing-query conversions
  • Impressions classified: 75,698 exact and 119,808 phrase (2022-23); 74,786 exact and 145,453 phrase (2025-26)
  • Hidden search terms: nearly 43% of eligible search terms, mostly phrase match
  • Headline results: new phrase-match queries $209 per lead vs $163 (+28%); new exact-match queries $109 vs $116 (-7%); phrase queries with no words in common 2% to 9% of volume
  • Google definition cited by the study: https://support.google.com/google-ads/answer/7478529

Timeline

Summary

Who: New Path Digital, an advertising agency in High Point, North Carolina, and its Digital Marketing Manager Mary Rippe, using data from three lead-generation clients: a law firm, a higher-education advertiser and a local B2B service business.

What: A study of nearly 50,000 exact and phrase keyword-and-search-term pairs found that conversions from searches Google newly matched to phrase-match keywords cost 28% more ($209 against $163) because they converted less often, while conversions from newly matched exact-match searches cost 7% less ($109 against $116) because clicks were cheaper. Phrase-match queries sharing no words with the keyword rose from 2% to 9% of volume; 75% of exact-match volume contained every word of the keyword.

When: Published October 27, 2026 and last updated November 5, 2026. The study compares 2022-23 with 2025-26, with performance figures from April 2025 through March 2026.

Where: Google Ads Search campaigns in three lead-generation accounts managed by the North Carolina agency.

Why: The agency set out to test whether Google's move toward meaning-based matching helps or hurts advertisers. Its answer is that it depends on the match type, with the caveats that the data covers three accounts, shows correlations rather than causes, counts form fills rather than qualified leads, and could not see nearly 43% of eligible search terms, which Google hid.