Some pages should earn clicks. Others should earn citations.

Some pages should earn clicks. Others should earn citations.

Traffic can fall even when rankings look fine. On informational queries, Google’s AI Overviews can cut click-through rate by 20 to 40 percent while your position report barely moves, according to Search Console Tools.

That is why the SEO vs GEO debate goes nowhere. The useful question for a small SaaS team is simpler: which pages need the click, and which pages just need to be quotable where AI answers are eating the click anyway?

Which pages should rank in search vs get cited by AI?

Split pages by job, not by channel. Automaton says SEO and GEO reward different content shapes even when they target the same user, and Google’s own May 2026 AI search guidance still says classic SEO best practices matter for generative features.

That lines up with what operators are seeing in the field. Informational intent gets absorbed by AI summaries. Commercial and bottom-of-funnel intent still needs a site visit.

For a tiny team, that means:

• Search-click pages: pricing, comparisons, product pages, integration pages, docs, migration guides, use-case pages
• Citation pages: plain-English explainers, definition pages, benchmark summaries, how-to answers, short research-backed breakdowns
• Proof assets feeding both: case studies, original data, customer workflows, benchmarks, product experiments

Stop asking one page to do every job. A pricing page does not need to be the best thing for an AI model to quote. A benchmark page does not need to carry your whole conversion path.

Informational content now loses clicks even when it still ranks

Google’s AI Overviews sit above the normal results and answer the question on the results page. Search Console Tools says they now appear on the majority of informational queries in the US, and that stable rankings can still hide a 30 to 50 percent traffic loss on affected terms.

Search Engine Journal reports the same pattern from publishers: some have seen click-through rates drop by as much as 89 percent after AI Overviews rolled out broadly.

That matters because a lot of SaaS content calendars are still built like it is 2023. Teams publish top-of-funnel posts, watch them rank, and assume the job is done. It is not done if the answer gets consumed before the click.

Bottom-of-funnel pages behave differently. Search Console Tools explicitly says transactional and brand queries are largely unaffected. That is where you still want the click. The user needs pricing, screenshots, trust signals, implementation detail, and product truth that a summary box cannot close on its own.

Citation-friendly content needs a different shape than ranking content

AI systems quote pages they can parse fast and lift cleanly. Automaton argues that GEO rewards a different content shape from SEO, even when the topic is the same.

In practice, that usually means:

• direct answers near the top of the page
• short sections with one claim each
• plain wording that is hard to misread
• original numbers or firsthand evidence
• structure that makes attribution easy

That is not the same as the old SEO playbook of stretching a topic into a long post and hoping coverage depth carries the ranking.

If you want citations, write pages that can survive being excerpted. A model may pull one section, not your whole article. Make each section make sense on its own. Put the claim first. Put the proof right after it.

If you want clicks, build the page around the next action. Comparison pages should compare. Docs should resolve objections. Product pages should show the thing, not drift into essay mode.

One source of truth beats running two content programs

A 1-5 person team does not need separate SEO and GEO departments. Automaton says the foundations overlap even when the tactical layers split, which is the useful part for small operators.

Use one source pack for each topic:

1. the raw proof asset — benchmark, case study, workflow, experiment, customer result
2. the quotable explainer — a tight page that answers the informational query directly
3. the conversion page — pricing, comparison, feature, integration, or use-case page that turns demand into trials or demos

Same evidence. Three outputs. Different jobs.

That keeps the workload sane. You are not doubling content. You are packaging the same truth for different moments in the buyer journey.

Measurement has to follow that split too. Search Console Tools makes the blunt point that position can lie while CTR tells the truth. If clicks fall on informational pages, do not treat that as automatic failure if branded search, assisted conversions, or referral patterns improve.

The practical takeaway is simple. Build pages that deserve clicks where clicks still matter. Build pages that deserve citations where AI answers now intercept the visit. If you run a small SaaS, that is the cleanest way to stop publishing one-size-fits-none content.