All posts
Deep Dive · 7 min read · 2026-11-22

The Four Reasons AI Search Won't Cite Your Service Business

Most AEO advice treats citation as a scoring problem: get enough points across schema, directories, reviews, and content, and you'll appear in AI answers. That framing misses something important. Some signals don't add points -- they eliminate you entirely, regardless of how well you score on everything else.

We've been calling these gatekeeper failures. And understanding them changes where you spend your optimization effort.

Where This Comes From

In our 2026-11-09 review of the AEO research literature, Scout filed findings from arXiv:2605.25517 -- "What Gets Cited: Competitive GEO in AI Answer Engines," accepted at SIGIR 2026. The researchers at Sprinklr ran 252,000 controlled citation trials across six LLMs using a head-to-head method: for each trial, two candidate documents competed for the same citation slot, with one variable changed between them.

The result was an 18-factor taxonomy of citation signals. At the top were factors that behave differently from the rest. Most factors are additive -- stronger schema helps, more reviews help, more directory presence helps. But four factors are disqualifying. Fail any one of them and your document loses the citation comparison regardless of how strong its other signals are.

Those four: topic mismatch, price not mentioned, old timestamp, and lower list position.

Gatekeeper One: Topic Mismatch

This one is intuitive but often missed in practice. AI platforms are not retrieving your homepage and finding the most relevant section. They retrieve pages in response to specific query variants -- "how much does it cost to fix a burst pipe in [city]" is a different retrieval target than "emergency plumber near me."

A page about "our plumbing services" that never mentions burst pipes, water line repair, or emergency response will not match the first query well enough to enter the citation candidate pool for it. Topic mismatch is not about keyword density -- it's about whether the page's actual content is a close match to what the AI engine asked for.

The fix is not rewriting your one service page to cover everything. It is building pages around specific service and query contexts, not broad category pages that try to cover the whole business in 300 words.

Gatekeeper Two: Price Not Mentioned

This is the finding that most surprised us, and the one most service businesses have not addressed.

For queries where price is a primary buyer concern -- "how much does AC installation cost," "what does a root canal cost without insurance," "affordable roofing contractor near me" -- the absence of pricing information on a page is disqualifying. Not a disadvantage. Not a signal the page has to overcome. A disqualifier.

In our 2026-11-10 methodology rec, Scout applied this finding to our audit categories. The specific verticals affected: HVAC, plumbing, roofing, electrical, dental and healthcare, legal services, moving services, pest control, and home cleaning.

If you are a plumber and your service pages contain no pricing information -- no ranges, no "starting at" figures, no cost factors ("what affects the price of a drain clearing"), no estimate request CTA -- then you are failing the price-absent gatekeeper on every cost query your potential customers are asking. Phase 1 work (directory presence, schema, NAP cleanup) will improve your position in the retrieval candidate pool. It cannot compensate for a gatekeeper failure at the content level.

The practical fix is not publishing exact prices you're uncomfortable committing to. It is including cost-factor content -- the variables that affect price, typical ranges in your market, what makes one job more complex than another. That content answers the buyer's intent without overcommitting you on specifics.

This content is exactly what AI platforms retrieve when a buyer asks a cost query. The page that answers it gets cited. The page that doesn't, doesn't -- even if everything else about it is solid.

Gatekeeper Three: Old Timestamp

Content freshness is documented across multiple data points in our research vault. The SIGIR 2026 study adds a sharper framing: in a head-to-head citation comparison, the more recent document wins, and this effect is not marginal. Old timestamp is a gatekeeper, not a tiebreaker.

What counts as old depends on the platform and query type. Perplexity applies a tight freshness window -- pages not updated within the last 30 days show meaningfully lower citation rates in our observations. Google AI Mode weights recency heavily for queries where the latest information matters.

For service businesses, the practical implication is that service pages last updated two or three years ago are competing at a structural disadvantage on every freshness-sensitive query. This is not about adding a new paragraph monthly. It is about whether the page contains a date signal the AI platform can verify -- a "last updated" indicator, a year referenced in the content, a review or case study timestamp.

A page that says "our HVAC team has helped homeowners since 2008" with no other date signal presents as potentially outdated. A page that references specific current pricing factors, links to recent reviews, or notes a service area update signals recency in a way AI platforms can retrieve and verify.

Gatekeeper Four: Low List Position

This is the one that Phase 1 work directly addresses -- and the one that explains why content optimization alone underperforms.

In October 2026, Scout filed findings from arXiv:2610.15033, a 158,000-answer, 468,000-citation study. The result: retrieval factors (whether the page enters the citation candidate pool at all) explain 32% of citation variation. Page-level content properties -- schema, FAQ formatting, heading structure, passage length -- explain 2.8%.

The list position gatekeeper is a retrieval pool problem. If your page does not rank well enough to be retrieved as a candidate source for a given query, no amount of content optimization changes your citation probability. You are not in the pool.

Directory presence, NAP consistency, and accumulated referring domains are the signals that determine list position. This is why the Phase 1 infrastructure work we document in every Sourcepull audit comes before content recommendations. Schema and FAQ optimization are amplifiers for pages already in the pool. They are not substitutes for getting into the pool.

The 32% vs. 2.8% split is the most important number in our research vault right now. It tells you where the return on optimization effort actually lives.

Why These Four Travel Together

A service business with thin, undated service pages that never mention pricing, covering a city where they have no directory presence, is failing all four gatekeepers simultaneously. That is more common than it sounds.

The standard AEO advice -- "add FAQ schema," "use question headings," "make your content more conversational" -- is entirely about the generation-stage improvement that sits inside the 2.8%. It is not useless. But it is addressing the wrong stage of the problem.

The diagnostic question for any service business is: which gatekeeper is the binding constraint? If you have no directory presence, the list position gatekeeper dominates. If you have good local rankings but no cost content, the price-absent gatekeeper is what's holding you back.

A Signal Check audit at sourcepull.ca runs your business against the platforms to show where you're appearing and where you're not -- which is the starting point for figuring out which of these is your actual problem. A gatekeeper failure looks different from a Phase 2 optimization gap, and the fix is different too.

Identifying which gate you're failing at is the first move.

See how your business scores on AI platforms.

Check your score — free