The language of paid search is getting longer.
An updated Google Ads analysis from Jason Tabeling shows a sharp migration away from one- and two-word search terms and toward more specific, conversational queries. Between January 2025 and August 2026, the shortest query bucket fell from 42% to 24% of impression share in his dataset, while three- and four-word searches climbed from 33% to 48%.
The conversion data makes the shift more consequential. Five- and six-word queries increased their share of conversions from 3% to 9%, while searches containing seven or more words rose from 1% to 4%. In proportional terms, those long-query conversion shares tripled and quadrupled.
Tabeling argues that AI Mode and the broader adoption of conversational AI are teaching users to search differently. Google's own data independently shows that AI Mode queries are dramatically longer than traditional searches. But the Google Ads numbers in Tabeling's September 11 Search Engine Land analysis come from his advertiser dataset, not from an official global Google Ads study, and they do not by themselves prove that AI Mode caused the change.
Short queries lost nearly half their impression share
The clearest movement is at the short end of the query distribution.
One- and two-word searches accounted for 42% of impressions in January 2025. By August 2026, their share had fallen to 24%, an 18-percentage-point decline.
That is not a subtle movement around the edges of a stable search market. In the analyzed campaigns, the query mix itself changed materially.
Advertisers that historically treated short, generic search terms as the center of paid-search demand may therefore be looking at a user behavior model that is becoming less representative.
Three- and four-word searches became the new center of gravity
The impression share did not simply disappear.
Three- and four-word queries rose from 33% to 48% of impressions over the same period, gaining 15 percentage points.
Tabeling describes this bucket as the new center of gravity for search impressions.
The shift suggests that more users are including additional context in the initial query rather than beginning with a minimal keyword and refining through several subsequent searches.
The conversion shift extends further into the long tail
The most striking part of the updated analysis is not impression volume but commercial outcomes.
Five- and six-word queries increased from 3% to 9% of conversion share, while queries containing seven or more words moved from 1% to 4%.
Those are still smaller shares than the dominant query buckets, but their growth rates are substantial.
The data suggests that longer searches are not merely informational conversations. They are increasingly appearing in the part of the query mix that produces measurable advertiser conversions.
Three- and four-word queries also gained conversion share
The movement is not confined to extremely long prompts.
Tabeling's summary data says three- and four-word queries increased their share of conversions from 20% to 46% across the analysis period.
That is strategically important because a four-word query does not necessarily look like an AI prompt. It can simply be a more specific expression of need.
The transformation of search may therefore appear first as incremental specificity rather than every user suddenly typing paragraph-length questions.
Google says AI Mode searches are three times longer
Google's own behavioral data supports the broader direction of travel.
In a May 2026 analysis of AI Mode usage in the United States, Google said the average AI Mode search is three times the length of a traditional Search query.
The company also said AI Mode queries had more than doubled every quarter since launch and that Search's AI features were contributing to record query activity.
This is strong evidence that people interact differently with AI Mode. It is not, however, evidence that every increase in long queries inside a Google Ads account is caused by AI Mode.
Google is explicitly encouraging a more conversational search model
At Google Marketing Live 2026, Google again said AI Mode searches average three times the length of traditional searches.
Google's marketing presentation framed longer, conversational searches as richer expressions of user intent and argued that advertisers can use Google's AI systems to match against those more nuanced needs.
The company told marketers that they can no longer think only in terms of simple keywords.
That messaging aligns closely with the pattern Tabeling sees in advertiser search-term data.
AI Mode may be teaching a transferable search habit
One plausible explanation is behavioral transfer.
Users who become accustomed to describing a complete problem to ChatGPT, Gemini or AI Mode may carry that habit back into ordinary Google searches.
Instead of searching “running shoes” and refining later, a user may begin with a query describing the intended surface, injury concern, budget and preferred feature.
The search becomes less like a database keyword and more like a compressed conversation.
But the dataset does not isolate AI Mode as the cause
The timing is suggestive, not experimental.
Tabeling's analysis observes how query length changed during a period of rapid adoption of LLMs and Google's AI Search experiences.
It does not compare otherwise identical users randomly assigned to AI Mode and traditional Search, nor does it isolate the exact search surface that generated every ad query.
The correct conclusion is that the advertiser data is consistent with an AI-driven behavioral shift, not that it proves AI Mode independently caused the entire change.
ChatGPT and other LLMs may be part of the same effect
Tabeling's earlier analysis connected the trend to mass adoption of LLMs more broadly, not only to Google's product.
People are learning that modern information systems can understand full sentences, constraints and conversational context.
Once that mental model changes, the old habit of reducing a need to two carefully selected keywords becomes less necessary.
AI Mode may accelerate that transition inside Google, but it is operating within a much larger change in how users formulate digital questions.
Advertiser settings can also change the observed query mix
Google Ads itself has become more automated during the analysis period.
Broad matching, Performance Max and AI-powered query expansion can expose advertisers to searches that are more varied than the keyword lists they manually selected.
Google's AI Max for Search campaigns is explicitly designed to find relevant queries beyond existing keywords and match ads to more specific user intent.
Any longitudinal Google Ads dataset therefore reflects both changes in human search behavior and changes in how advertising systems decide which queries an advertiser can enter.
AI Max is designed for the same more nuanced query environment
Google introduced AI Max as a way to help Search campaigns reach new and more specific queries while using AI to adapt targeting and creative.
Google's AI Max announcement says modern Search increasingly involves complex and articulated questions that provide richer intent signals.
In 2026, Google moved AI Max out of beta and continued shifting legacy Search automation toward it.
The product strategy and the query-length trend point in the same direction: Google expects keyword-to-query matching to become less literal and more intent-driven.
Longer queries can contain more commercial information
Word count itself does not create intent.
But additional words give users room to express details that can reveal where they are in a buying decision.
A query can specify a product type, use case, price range, location, compatibility requirement and urgency in a single search.
That richer context can make a long query more useful for ad matching and landing-page selection than a generic head term whose intent is ambiguous.
A seven-word query can be closer to purchase than a two-word query
Traditional keyword strategy often associates long-tail searches with lower volume but greater specificity.
The new data suggests that this familiar principle may be gaining importance as conversational interfaces normalize detailed requests.
“CRM software” can represent research, navigation or purchase intent. A longer query specifying company size, integration requirement and pricing need tells the advertiser considerably more.
The commercial value comes from the constraints expressed in the query, not simply from crossing a seven-word threshold.
Advertisers should analyze their own search-term length distribution
Tabeling's numbers should be treated as a prompt for account-level analysis rather than a universal benchmark.
An advertiser can export search terms over a meaningful historical period, count the words in each query and group them into comparable buckets.
Impression share, click share, conversion share, conversion rate, CPA and ROAS can then be compared by query length.
The key question is whether the same migration toward longer, more specific searches is happening in the advertiser's own market.
Do not optimize for word count as if it were a ranking factor
Once a pattern becomes visible, it is tempting to turn it into a mechanical rule.
An advertiser should not conclude that seven-word queries are inherently better and attempt to force campaigns toward arbitrary length thresholds.
A long query can still be irrelevant, informational or low value. A two-word brand query can be extremely profitable.
Query length is useful as an analytical dimension because it can reveal changing behavior, not because the number of words is itself a business objective.
Search intent taxonomy needs to become more sophisticated
Keyword lists traditionally organize demand around product categories and modifiers.
Conversational queries can combine several intent dimensions at once.
A single search may include a problem, audience, desired outcome, product constraint and comparison criterion.
Advertisers may therefore need taxonomies that classify intent semantically rather than relying primarily on exact recurring phrases.
Negative keyword strategies may need review
Longer queries contain more words, which creates more opportunities for a traditional negative keyword to block a search that is valuable in context.
A term that historically indicated low intent may appear inside a much more specific commercial request.
Advertisers should review negatives against actual modern search terms rather than assuming old query patterns remain stable.
This does not mean removing controls indiscriminately. It means evaluating them in the context of richer natural-language searches.
Ad creative should answer more specific needs
If users increasingly describe detailed problems, generic advertising messages can become less competitive.
Creative assets should communicate concrete capabilities, differentiators and use cases that Google's systems can match to nuanced intent.
Landing pages likewise need enough substantive information to satisfy a user who arrived with a detailed set of requirements.
A conversational query raises expectations: the user has explained more, so the resulting experience should respond with comparable specificity.
Landing-page architecture may matter more than exact-match keyword pages
A world of infinitely varied natural-language queries makes it impractical to build a separate page for every phrasing.
Advertisers need strong destination pages that comprehensively address a coherent need and allow Google's matching systems to connect many query variations to the same useful experience.
This is another reason the strategic unit is moving from literal keyword repetition toward intent coverage.
The page should be specific enough to convert without becoming a thin template for every possible long-tail phrase.
Conversion reporting should separate share from absolute volume
The reported tripling and quadrupling refer to conversion share for the longer query buckets.
A rise from 1% to 4% is a fourfold increase in share, but seven-plus-word queries still represent a minority of total conversions in the analyzed data.
That distinction matters when translating a striking growth rate into budget decisions.
Advertisers should inspect both the proportional growth and the absolute number and value of conversions generated by each query group.
The three- to four-word bucket may be the more immediate commercial shift
The longest queries produce the most dramatic relative growth rates, but the middle of the distribution carries much more volume.
Three- and four-word searches reached 48% of impression share and, according to Tabeling's summary, 46% of conversion share.
That suggests the near-term transformation may be less about paragraph-length prompting and more about users routinely adding one or two extra pieces of context.
For most advertisers, that middle bucket may have a larger immediate financial impact than the headline-grabbing seven-plus-word segment.
CTR and CPC need to be read alongside the query migration
Tabeling's analysis also examines how engagement and cost metrics evolve across query-length buckets.
A change in query mix can alter account-level averages even if individual campaign mechanics remain similar.
If more impressions move toward highly specific searches, CTR, conversion rate and CPC can shift because the underlying population of auctions has changed.
Historical benchmarks should therefore be interpreted with awareness that the search terms feeding them may no longer look the same.
Year-over-year keyword benchmarks can hide a structural change
An advertiser comparing 2026 with 2025 may attribute a performance change entirely to bids, creative or competition.
But if the account is now receiving a fundamentally different mix of short and conversational queries, part of the difference may come from user behavior itself.
Adding query-length and intent analysis to year-over-year reporting can expose that hidden variable.
This is particularly important for mature accounts where campaign structures have remained relatively stable but search behavior has not.
The keyword is becoming less useful as the sole planning unit
Paid search was built around a powerful abstraction: the advertiser chooses a keyword, Google matches it to a user's search term and performance is optimized around that relationship.
That abstraction has been weakening for years as match types broadened and automation expanded.
Conversational Search pushes the change further because the user's query can express combinations of intent no keyword planner could enumerate efficiently.
Keywords remain useful controls and signals, but they increasingly sit inside a larger semantic matching system.
First-party conversion data becomes more valuable
As query language becomes more varied, advertisers cannot rely only on manually predicting every phrase that high-value customers will use.
Accurate conversion measurement gives automated systems a stronger signal about which kinds of nuanced searches actually produce business outcomes.
That makes clean conversion definitions, value data and offline conversion imports more strategically important.
The system needs to understand not merely which queries receive clicks, but which underlying intentions produce profitable customers.
The shift creates an opportunity for richer audience understanding
Long queries are not only media-buying inputs. They are qualitative customer research.
They reveal vocabulary, concerns, constraints and comparison criteria in users' own words.
Aggregating those themes can inform landing pages, product positioning, FAQs, sales enablement and content strategy beyond Google Ads.
As users become more conversational, search-term reports can become a richer source of voice-of-customer information.
Privacy and reporting thresholds can limit visibility
Advertisers should remember that Google Ads search-term reporting does not expose every query.
Privacy thresholds and reporting rules can leave some search activity unreported at the individual query level.
That means an account-level query-length analysis is based on the terms available in the report rather than a perfect census of every search that triggered an ad.
This limitation does not make the exercise useless, but it should be acknowledged when interpreting small changes or comparing accounts.
Industry differences may be substantial
A B2B software buyer can naturally express a long, constraint-heavy query. A consumer looking for a nearby petrol station may not need many words.
Travel, healthcare, finance, ecommerce, local services and SaaS can therefore develop different query-length distributions even under the same broader conversational trend.
Tabeling's dataset provides evidence of a meaningful aggregate shift within the analyzed advertiser data, not a universal formula for every vertical.
Account-specific baselines remain more useful than copying the published percentages into forecasts.
The causal story should remain more cautious than the behavioral story
There are two claims here with different levels of evidence.
The first is observational: in Tabeling's Google Ads data, longer queries gained substantial impression and conversion share from January 2025 through August 2026. The second is causal: AI Mode is responsible for that migration.
Google's own statistics make the causal interpretation plausible because AI Mode searches are much longer than traditional queries, and the product explicitly encourages conversational interaction.
But other LLMs, Google Ads automation, account composition and broader cultural changes in search behavior can contribute. The evidence supports a strong trend and a credible hypothesis, not a clean causal experiment.
Advertisers should prepare for intent, not simply longer strings
The most useful lesson from the data is not that marketers should start counting words in every keyword.
It is that people increasingly expect search systems to understand detailed intent.
Tabeling's dataset shows that one- and two-word queries fell from 42% to 24% of impression share while three- and four-word searches rose from 33% to 48%. At the same time, five- and six-word queries tripled their share of conversions and seven-plus-word searches quadrupled theirs.
Google's first-party data independently confirms that AI Mode encourages much longer questions. Whether AI Mode is the sole cause of the advertiser trend is unproven, but the direction is difficult to ignore.
The competitive advantage in paid search is therefore shifting from predicting the exact words a user will type toward understanding the richer need those words express. As Search becomes more conversational, the advertisers best prepared for the change will be the ones optimizing for intent rather than nostalgia for the two-word keyword.