The B2B SaaS search auction has been the most reliably inflationary corner of the paid media landscape for the last five years. Between Q1 2021 and Q1 2026, the median non-brand SaaS CPC in our cohort rose by approximately 91%. Individual sub-categories — CRM, HR tech, cybersecurity — printed higher. The inflation was consistent enough that the joke, in the paid-search community, had become that a SaaS keyword's CPC was a leading indicator of its category's late-stage funding.
Q2 2026 is, on our data, the first quarter since Q3 2020 in which the median non-brand SaaS CPC did not rise. It rose 0.4% QoQ, which is well inside the noise band. The trailing four-quarter increase, at 8.7%, is meaningfully below the trailing eight-quarter figure of 21%. Something has changed. This piece is about what.
What we looked at
The cohort here is 63 B2B SaaS accounts, across CRM, developer tools, security, HR, and vertical SaaS. Combined quarterly paid-search spend is approximately £41m. We look at CPC, but we also look at impression share, click-through rate, and cost per pipeline dollar where the accounts have shared attribution data with us.
Three things showed up in the Q2 data that had not been present in previous quarters.
The first is a modest decline in impression share across the cohort. The median account saw its search impression share fall by approximately 3 percentage points quarter on quarter, without a corresponding change in bid strategy or budget. The interpretation, on inspection, is that a portion of the demand these accounts had been bidding on has migrated elsewhere.
The second is a small but detectable rise in average CTR — approximately 0.6 percentage points across the cohort. This is small but significant when combined with the impression share decline. If fewer impressions are producing more clicks, the queries the platform is now serving are, on average, higher intent. The low-intent queries have gone somewhere.
The third is a substantial rise in average time-to-close in the accounts where we have attribution data. The median time from first paid-search click to closed-won opportunity rose from 62 days in Q2 2025 to 78 days in Q2 2026. Buyers are taking longer.
Where the demand went
Our working thesis is that a share of high-intent B2B SaaS search demand has migrated to AI chat interfaces — ChatGPT, Claude, Gemini, and their peers — during the buyer's initial research phase. The migration is not total; the buyer still, in most journeys, ends up on a search engine before they book a demo. But the discovery and shortlist phases, which used to happen through generic query patterns like "best CRM for small business", are increasingly happening in a chat interface where those queries are answered directly.
The mechanism this produces at the auction level is worth naming. The queries that had been driving the highest CPCs were, disproportionately, the generic top-of-funnel queries — the ones where a buyer with no established preference typed a broad category term and clicked on the highest bidder. Those queries have thinned out. What remains, in the auction, is the queries that are still made from search: navigational queries ("[category leader] pricing"), comparison queries in mid-consideration ("[vendor A] vs [vendor B]"), and product-specific technical queries ("does [product] support [feature]"). These are higher-intent, but they are also a smaller volume than the queries that used to be there.
The result is an auction with fewer bidders per query in aggregate (because the sub-categories in which multiple new entrants competed on generic terms have partly gone quiet) but higher intent per query. CPCs have not risen because the aggregate competitive pressure per query is broadly unchanged; volumes have fallen but bid strategies have adjusted downward roughly in proportion.
"The plateau is not a reprieve. It is a symptom of an underlying migration. The queries that are gone from the auction are gone. The queries that remain are different in character. Anyone reading the CPC line as good news is misreading the situation."
Why the reprieve is temporary
Two forces will, on our reading, reintroduce inflationary pressure within two to four quarters.
The first is that the migration to chat interfaces has surfaced a subset of vendors whose brand awareness is not strong enough for them to be shortlisted by the chat interfaces' models. Those vendors are, currently, investing heavily in brand-building work to correct this — and a share of that brand investment will, downstream, translate into higher-intent search demand. When it does, it will re-enter the auction at higher CPCs.
The second is that the search platforms themselves are, quietly, adding new inventory to the paid-search auction. Google's expansion of search partners, its increased use of AI Overview citation placements as ad units, and the ongoing integration of shopping-style formats into non-shopping queries are all producing new competitive pressure that will land on CPCs over the next few quarters.
Our forecast is a return to 3-5% QoQ inflation by Q4 2026, and a resumption of the 15-20% trailing-four-quarter rate by Q2 2027. The plateau is a snapshot. The underlying dynamics have not reversed.
What to do in the interim
The single most useful thing a B2B SaaS paid lead can do with the current window, on our read, is to spend it on the categories of work that get displaced when the auction re-heats.
The first is landing-page investment. When CPCs are flat, the marginal value of a landing page conversion-rate improvement is fully captured by the account. When CPCs are inflating, the improvement is partly absorbed by the auction. Now is the moment for landing-page work.
The second is the operationalisation of the AI-referred traffic that is starting to appear in analytics. A share of the buyer journey now begins in a chat interface; a share of it ends on the brand's site through a direct visit that the attribution system will not correctly attribute. Setting up clean tracking, cohort analysis, and the beginnings of an internal understanding of the AI-referred journey is work that will compound in value as this channel grows.
The third is the honest conversation with the head of marketing about the incrementality of the paid-search line. In a period of CPC inflation, the argument for continuing to spend is straightforward: the auction is competitive, our competitors are in it, we need to be too. In a period of CPC plateau with declining volume, the argument gets harder. Now is a good time to have the incrementality conversation, because the answer will be less politically fraught than it will be when the auction re-heats.