Programmatic ‘cost of X’ and ‘price of X in [year]’ pages: do they still hold rankings now that AI answers quote prices directly?

1 reply

• 

  adminAdmin

  Original poster 2 hours ago

  Some still hold rankings. Fewer still earn the clicks those rankings used to bring. That gap is the real story: a “cost of X” page can sit at position 2 and lose most of its traffic, because an AI Overview (or an older price widget) now hands the searcher the ballpark figure the page was built to deliver. The pages that keep earning visits hold something a summary cannot compress into one sentence: the searcher’s own variables, local or dated primary data, an itemised breakdown, or a direct route to a quote or purchase.

  There is a second risk that often gets blamed on AI. Many programmatic price templates were exposed to Google’s quality systems long before AI answers quoted anything. A page that swaps a city name or a year into the same national average is close to the textbook case of scaled content with little added value. Those pages tend to lose rankings, not just clicks. So the first job is working out which problem you actually have, because the fixes are different.

  Two different failures that look the same in a traffic graph

  Most “our price pages died” threads mix up click loss on pages that still rank with ranking loss on pages Google has re-evaluated. Search Console usually tells you which one you are looking at, if you read the metrics separately.

  Pattern in Search Console	Most likely explanation	How to verify
  Position and impressions stable, clicks and CTR falling	SERP absorption: an AI Overview, AI Mode answer or price widget satisfies the ballpark question	Check the live SERP for your top queries and compare CTR within a fixed position band over time
  Position falling across most of the template at once	Quality reassessment of the template, often around a core update	Annotate the March 2026 and May 2026 core update windows; compare templated pages against hand-built pages on the same site
  Impressions down and average position “better” from September 2025, clicks roughly flat	Reporting artefact after Google stopped supporting the &num=100 parameter	Judge that period on clicks, not impressions or average position
  A minority of pages fine, the rest declining	Page-level value differences inside the same template	Compare the pages that held against those that fell: data depth, uniqueness, links, freshness

  Only the first pattern is really “AI answers quoting prices”. The second needs a content and template decision, the third needs no fix at all, and the fourth tells you what the fix should look like.

  Why a quoted price replaces some pages and not others

  An AI price answer works like the “from $X” line on a tradesperson’s flyer. It settles “roughly how much?” but not “how much for my job, in my suburb, this month?”. Pages that existed to restate the flyer are now redundant. Pages that do the equivalent of the itemised quote still get visited, because the searcher cannot finish the task without them.

  The mechanism behind that analogy is intent splitting. A price query usually hides one of four needs: a ballpark figure, a personal figure, verification of a figure they have seen elsewhere, or a next step (buy, book, request quotes). AI features serve the ballpark need almost completely. They serve the other three poorly, because those depend on inputs the searcher has not typed and data the summary does not hold.

  The click evidence supports the direction, though not a precise size for price queries. Pew Research Center’s browsing study of 900 US adults (March 2025) found users clicked a traditional result in 8% of visits when an AI summary appeared, against 15% when none did, and clicked a link inside the summary in just 1% of visits. Seer Interactive’s 2026 update (53 brands, January 2025 to February 2026) also found lower organic CTR on queries with an AI Overview, and found that being cited improved clicks per impression compared with not being cited, while still trailing queries with no AI Overview. Neither dataset isolates price queries, so treat them as direction rather than a forecast for your template.

  How exposure differs by type of price query

  Live market prices

  Gold, fuel, currency and crypto queries were answered by widgets well before AI features. A page repeating today’s figure has little reason to be clicked. Price history, charts, alerts and local retail comparisons are what still earn visits.

  Retail product prices

  Merchants with live price and stock compete here, and Google keeps moving its shopping surfaces towards merchant data, including checkout inside AI Mode with participating merchants (as of September 2026, select merchants with products eligible in the US, Canada and Australia, according to Google’s Merchant Center help). An informational “price of X in 2026” page without stock data sits awkwardly between the AI answer and the retailer.

  Service and project costs

  Roofing, dental implants and removalists have wide real variance by location, size, materials and access, so a national range rarely answers the actual question. These pages hold up best when built properly, and the lead value per visit is usually the highest of the four.

  City or year matrices

  “Cost of X in [city]” and “price of X in [year]” grids are the most exposed. If the number does not genuinely change between URLs, the pages are near duplicates at scale, and an AI answer can state the same range once.

  One core update dataset points the same way. Aleyda Solis’s analysis of the May 2026 core update (domain-level Sistrix visibility, US and UK) reported reference aggregators losing visibility, and framed the dividing line as whether a site is where the user completes the task or a layer summarising it. That is not a study of price pages, and domain-level visibility is not page-level traffic, but a templated price page with no unique data is exactly that kind of summarising layer.

  The [year] token is a promise, not a ranking trick

  “Price of X in 2026” expresses a real freshness need, so a current year in the title can help the click when the data behind it is current. The problem is the January find-and-replace. Google’s own helpful content self-assessment asks: “Are you changing the date of pages to make them seem fresh when the content has not substantially changed?” A template that rolls the year forward while the figures stay the same is doing exactly that, across thousands of URLs at once.

  I have not seen credible evidence that a year in the title is a ranking factor in either direction. Its main effect is on CTR and trust, and a stale number under a current year damages both once a searcher compares it with the AI answer or a quote they received. My rule: keep one evergreen URL per topic (redirect old year-specific URLs into it), change the year only when the data was actually refreshed, and show a true “prices checked” date with the source.

  Watch for this:

  Google’s scaled content abuse policy lists “stitching or combining content from different web pages without adding value” and generating many pages “without adding value for users”. A city by service by year matrix fed from one national average fits that description whether a person or a script wrote it. AI use itself is not the violation. Low-value pages at scale are.

  What a price page needs now to be worth the click

  • Inputs the summary lacks: a calculator or selector for size, location or specification, so the page answers “my number”.
  • Primary data with a method: where the figures come from (quotes, invoices, supplier lists), the sample size, the region and the date range.
  • An itemised breakdown: labour versus materials, fixed versus variable costs, and the add-ons that blow out budgets.
  • What moves the price, and by roughly how much, in plain terms.
  • A next step that fits the intent: request quotes, check stock, book an assessment.
  • A true last-checked date, not a template date.

  On AI citations, keep expectations honest. Google’s May 2026 guide to generative AI features says you do not need special schema, llms.txt, artificial chunking or AI-specific rewording, and describes the work as still SEO. Citation is not guaranteed by any tactic, and it is drifting away from ranking: Ahrefs data reported by Search Engine Journal showed the share of AI Overview cited pages ranking in the top 10 falling from 76% (July 2025) to 38% (March 2026), although the detection method changed between the two datasets, and BrightEdge reported 17% overlap using a different method. Track citations as their own metric rather than assuming your rank protects you.

  Useful reframe:

  A price page that an AI answer can fully replace was usually a thin page already. The AI answer made that visible in your click data. It did not create the weakness.

  How I would audit a programmatic price section

  1. Segment by template and query class. Use page regex filters to split cost-of, price-in-year, city matrix and service pages. Flag pages whose headline figure is identical across URLs.
  2. Pull 16 months per segment. Clicks, impressions, CTR and position, annotated for September 2025 (num=100), the 2026 core updates and your own template releases.
  3. Record the SERP manually for the top queries in each segment. Note whether an AI Overview appears, whether it quotes a figure, whether you are cited, and whether your figure matches theirs. Search Console will not do this for you: AI Mode data has been folded into Performance totals since June 2025 without a separate click breakdown, and the generative AI performance reports show impressions only, with no queries or clicks. They began with a subset of properties (UK first) and reached all sites worldwide on 31 August 2026, so older windows may have no AI feature data at all.
  4. Measure business outcomes per segment. Leads, quote requests and revenue per session. Fewer clicks from a better-qualified audience can be an acceptable result, but verify it rather than assume it.
  5. Score each page on data uniqueness, freshness, links earned and conversions, then apply the decision table below.

  Page profile	Action	Main trade-off
  Unique data, converts, position holding while CTR falls	Keep; add inputs and breakdowns; test titles that promise the specific answer	Ongoing cost of collecting and refreshing data
  Template duplicates with no unique data, links or conversions	Consolidate into one stronger page per service or product and 301 the rest	Losing some exact-match long-tail rankings, usually low value
  Duplicates that still bring some leads	Merge into a parent page with a location selector; keep standalone pages only where local data genuinely differs	Engineering effort and a temporary dip during consolidation
  Separate URLs per year (2023, 2024, 2025)	Redirect into one evergreen URL	Minimal, and it concentrates links and history

  Hypothetical example:

  A home services lead generation site has 3,000 “cost of [service] in [city]” pages. About 2,800 show the same national range with the city name swapped. The other 200 use figures from real jobs in those cities. Across the site, positions are mostly steady but clicks are sliding. The audit would likely show the 200 data-backed pages losing far less CTR and producing most of the leads, while the 2,800 duplicates slowly lose position as well. The sensible move is to fold the duplicates into service-level pages with a location-aware calculator, keep and deepen the 200, and only create new city pages when real local data exists.

  A test worth running before rebuilding the whole template

  Hypothesis: adding local primary data, a calculator and a dated method note to city cost pages increases clicks at stable position and leads per page.

  • Test and control: two randomised groups of comparable pages (same service, similar traffic, similar AI Overview presence). Do not hand-pick the test group.
  • Primary metric: clicks, and CTR within a fixed position band. Secondary: leads per page, citation rate in the AI answer, impressions.
  • Duration: 8 to 12 weeks. If a core update lands mid-test, compare both groups inside that window rather than discarding it.
  • Confounders: seasonal demand, competitor changes, AI feature rollouts, and deployment timing that differs between groups.
  • Limitations: page-level SEO tests are noisy, and small groups may produce no clear result. An inconclusive test is still information.

  If the test pages hold CTR and lift leads, roll the approach out and prune the rest. If clicks stay flat but leads per click rise, the investment may still pay. If nothing moves, that query class is probably absorbed by the SERP, and the budget belongs in channels where the searcher still needs you.

  If this were my site, I would not ask whether price pages still rank. I would ask which of my price pages hold information the searcher cannot get from the answer box, and whether those pages drive revenue. Build and refresh those. Consolidate the rest before a core update makes the decision for you.

  Need help with this on your own site?

  Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

  See SEO services