How Seasons Reshape AI Local Search

Does AI local search actually change by season, or just search volume?

Both change, and the second is the part nobody plans for. Everyone in the trades knows demand swings: furnaces in January, air conditioning in July, roofing once the snow clears. What Formative Digital's data shows is that the generative answer swings with it. The set of pages ChatGPT, Perplexity, Gemini and Google AI Overviews reach for is not fixed, so the business named in a March answer can vanish from the October one without losing a review.

Our nine-city scrape shows why. The heating answer was carried not by contractors but by furnace-quote aggregators and directories, all heating-framed sources. When the calendar flips to cooling season the engines reach for cooling-framed pages instead, and a contractor optimised for only one half of the year is invisible for the other.

When do furnace, AC and roofing prompts peak in Ontario?

Furnace intent peaks in deep winter, cooling intent in mid-summer, and roofing in the dry months between, giving a trades business two or three distinct citation seasons rather than one. Home-services search data syndicated by Stacker and reported by WebFX shows air-conditioning repair searches surging roughly 266% in July, while furnace repair searches peak about 137% in January. A 266% July spike means the cooling question arrives at nearly four times its winter baseline.

Ontario makes the heating side heavy. Statistics Canada reports forced-air furnaces were 51% of Canadian primary heating systems in 2021, concentrating demand into a narrow, predictable cold window across Brantford, Hamilton and Kitchener-Waterloo. Roofing climbs as the ground dries in late spring, with storm-damage spikes that are harder to forecast but just as real.

Read those windows backward and the publish date has to sit a season ahead of the peak, not weeks: by December the engines have already settled on the winter sources they trust.

Why do AI engines swap which sources they cite as the season turns?

They swap because retrieval is recency-aware and intent-aware, so a seasonal prompt pulls a seasonal slice of the index. Ask Perplexity "who can replace my furnace in Hamilton" in November and a live search surfaces pages that are recent and densely about heating; the same question in June surfaces cooling pages instead. The engine is not favouring one contractor over another, it reflects which sources the open web is currently feeding it for that season.

So there is no single moment to go live. Each engine lags differently between publishing and influencing the answer, so plan for the fastest-reacting one, Perplexity, which punishes lateness most sharply.

What does Formative Digital's 1,732-citation HVAC scrape reveal about seasonal sourcing?

It reveals that the sources carrying the Ontario trades answer are overwhelmingly seasonal-content publishers, not contractors, which is exactly why timing decides who gets named. The scrape, drawn from FD's matrix.db using DataForSEO in May 2026, asked ChatGPT, Claude, Gemini and Perplexity who the best HVAC companies were across Hamilton, Kitchener-Waterloo and seven more cities. Strip out Gemini's Vertex grounding wrapper (114 citations on its own) and the most-cited real publishers were threebestrated.ca at 37, furnaceprices.ca at 35 across six cities, google.com at 25, and HomeStars at 15 in every city tested.

None of those is a contractor; every one is a publisher whose content is fresh and seasonal by design, from furnaceprices.ca built for the heating season to the dated, updated shortlists threebestrated.ca and HomeStars keep current.

The reading is direct. A contractor whose furnace page was written once in 2022 competes for a winter citation against an aggregator that re-stamped its heating shortlist last month, and the engine takes the fresh source. That is the gap timing closes.

Want to know which season you are already losing? Formative Digital will run your business through the same four-engine scrape, in season and out, and show which months you are cited and which you disappear. Get your month-by-month citation readout, yours whether or not you work with us.

Why do furnace-quote aggregators win the AI answer over local contractors?

They win because they combine three things an engine rewards at once: a ranked, extractable list, a fresh date, and seasonal framing that matches the prompt. A furnace-quote page leads its first screen with a named, ranked shortlist, each entry carrying a city and a one-line reason. Kevin Indig's February 2026 Growth Memo analysis found about 44% of AI citations come from a page's first 30%; aggregators obey that rule by design while a contractor homepage opens with a hero image and a slogan.

Freshness compounds it. An aggregator republishes its heating shortlist before each winter and its cooling shortlist before each summer, while a typical contractor site has one static services page that never moves, so a recency-aware engine takes the dated aggregator regardless of who does better work. None of this is improper; aggregators simply publish the format and cadence the engines were trained to trust.

We documented this pattern in our read of how HomeStars holds the trades answer, and our read on how directories came to own AI local search shows it across verticals. Seasonality adds a clock: the aggregator has both a better-structured page and a more recently-dated one, and for a seasonal query the advantages stack.

How does Query Deserves Freshness affect seasonal trade queries?

Query Deserves Freshness pushes newer content to the top for exactly the recurring, time-sensitive queries seasonal trades generate, which makes a freshly-dated page worth more than an older, stronger one. QDF is Google's system for prioritising recent content for, in Search Engine Land's description, "regularly recurring events and information that frequently changes." A January furnace failure and a July AC failure are textbook recurring events, and because Google AI Overviews and Gemini both ground on Google's ranking, that preference flows straight into the generative answer.

What seasonal signals make AI re-surface you each year?

The signals that re-surface you reduce to three: a genuinely updated dateModified, a front-loaded season-framed answer, and local entity data that does not drift between seasons. An honest dateModified paired with real edits, refreshed pricing and current rebate references tells recency-aware retrieval the page is current for this season's recurring event. A date bump with no substantive edit is the wrong move; engines increasingly discount it.

Google Business Profile posts add a second signal: a timely "furnace tune-up season is here" post feeds the Maps and Knowledge Graph cards ChatGPT reads, helping the engine that swaps slowest. Skip the refresh or let the entity drift, and you lose to the aggregator that re-stamped last month.

How do you measure seasonal AI visibility before the spike hits?

You measure it by running your real customer prompts through each engine months before the season, while there is still time to fix what you find. Ask "best furnace repair in {your city}, Ontario" of ChatGPT, Claude, Gemini and Perplexity well ahead of the January peak, and record which engines name you, in what position, and which source they pulled. Track those as four separate scores, because the engines react at four different speeds.

The source layer is the early-warning system. A new, well-dated furnace page tends to show in Perplexity first and ChatGPT last, so if the aggregators have re-stamped their winter shortlists and you have not, the data tells you in September, while you can still act, rather than in January when the season is lost. We lay out the per-engine method behind that early warning in tracking how your AI answer shifts week to week.

That came from sustained, structured content work for a Brantford retail client whose full year-over-year numbers we published, growing from roughly 1,000 to more than 82,400 monthly organic visits (SEMrush, April 2026). Trades are a different market with sharper peaks, so outcomes depend on your city, competition and existing footprint, which is why we diagnose first and run the work through the Formative Forces, our orchestrated multi-agent system, rather than one seasonal template.