Why is your organic traffic dropping when nothing else changed?
Pull up your analytics and compare this half against last year. If organic traffic is sliding while your rankings hold steady, you're not imagining it and your marketing team didn't suddenly get lazy. Google changed the rules without telling anyone.
When a buyer searches "how to automate invoicing for my business," Google no longer serves ten blue links. It serves an AI Overview — a generated summary that answers the question right there on the results page. The user reads, learns, and leaves. No click. A 2026 randomized field experiment — researchers from the Indian School of Business and Carnegie Mellon, tracking real Chrome users rather than survey answers — puts a number on it: AI Overviews cut organic clicks by 38% on queries where they appear.
The half of the story almost nobody is telling: per Seer Interactive's analysis of millions of organic impressions, brands cited as sources inside those AI answers earn 35% higher organic CTR than brands that aren't cited. The click is losing value; the citation is gaining it. And the citation — ChatGPT, Claude, Perplexity, or Google itself saying "according to this company" — is the new currency.
What does competing for citations instead of clicks actually mean?
Think about it from your buyer's seat. A US operations director evaluating vendors doesn't compare ten search results anymore. She asks ChatGPT: "should we buy off-the-shelf software or build custom for our distribution business?" The answer she gets names two or three sources. If your brand is one of them, the conversation starts with you. If it isn't, you don't exist in it — no matter how good the article buried in your blog is.
This isn't Google-only — but watch out, because the assistants don't all work the same way. ChatGPT leans mostly on Bing's index to search the web; Perplexity rewards freshness and will cite sites that are only weeks old; Google's AI Overviews mostly cite pages already in the organic top 10. Different routes, converging on the same new question: no longer "where do we rank?" but "who does the machine cite when it answers our customers' questions?"
Is classic SEO dead, then?
No. Ahrefs analyzed a massive sample of AI Overviews and found that about 76% of cited URLs already ranked in Google's organic top 10: there, AI mostly cites what already ranks. Classic SEO didn't die — it became the price of admission. What changed is what you do once you're in.
How do you write content AI actually cites?
Language models cite what they can extract without effort: direct answers, structured data, questions answered head-on. Here's the structure we use — and yes, the article you're reading is built exactly this way, on purpose. Notice the summary up top, the question-shaped headers, and this table:
| Citable-format element | What it gives the AI | Supporting data |
|---|---|---|
| 40-80 word TL;DR under the title | A ready-to-extract, attributable summary | Standard extractability practice |
| H2 headers phrased as questions | Matches how real users actually ask | Standard extractability practice |
| One useful data table | Structured facts, easy to quote | Standard extractability practice |
| FAQ with FAQPage schema | Machine-readable Q&A pairs | ~2.7x more AI citations (Relixir, 2025) |
| Visible update within the last 30 days | Freshness signal | ~3.2x more AI citations (Writesonic) |
| Top-10 organic ranking | In AI Overviews, AI cites what already ranks | ~76% of citations come from the top 10 (Ahrefs) |
Every row is a concrete editorial decision. It's not magic and it's not a plugin: it's writing for two readers at once — the human who wants the fast answer, and the machine that will summarize you and, if you make it easy, cite you.
How much do schema and freshness really matter?
More than they look — and they're the two cheapest changes you can make.
Structured markup, to start: a 2025 Relixir study found pages with FAQPage schema — the code that tells search engines "this is a Q&A section" — earn nearly triple the citations from AI engines. It's an afternoon of work for a developer, and it's rarely implemented on SMB sites.
Freshness pulls similar weight. Per Writesonic, content updated within the last 30 days earns more than triple the AI citations. That's why this article carries a visible update date. "Publish and forget" — the blog model so many companies still run — looks, in that light, like one of the quietest ways to burn marketing money.
An honest aside: while preparing this piece we ran into more than one "GEO" statistic that didn't survive being traced back to its original source. The field is so new that inflated numbers circulate freely. Every number in this article links to its study for exactly that reason — and when someone sells you magical optimization, ask for the source.
Are we actually doing this ourselves?
Full transparency: yes — extractable TL;DRs, question-shaped headers, one data table per piece, schema-marked FAQs, and real (not decorative) update dates, on our own site. Results? We're not going to show off numbers we don't have yet: we're still measuring, and claiming wins without data is exactly what we criticize. When the numbers land, we'll publish them here.
What does this mean if you're a US company thinking nearshore?
Competing for citations rewards teams that can produce and refresh structured, well-sourced content every month — consistently, not in heroic quarterly sprints. That cadence is exactly where a nearshore partner makes sense: same-day collaboration on US Central time, bilingual output for English and Spanish-speaking markets, and senior attention at rates that don't require a big-agency retainer. It's the work we do at Kynoz from Puebla, Mexico: custom software (web, mobile, AI) plus content and SEO-GEO programs — in English and Spanish, for either side of the border.
The currency changed before habits did. By the time competing for citations is standard practice, the early-mover advantage will already have been cashed by someone — no reason it can't be you.
FAQ
What's the difference between SEO and GEO?
GEO (generative engine optimization) means optimizing so generative engines — ChatGPT, Claude, Perplexity, Google's AI Overviews — cite your brand in their answers. It doesn't compete with SEO: since about 76% of URLs cited in AI Overviews come from Google's top 10, SEO remains the prerequisite GEO depends on.
Should we keep investing in traditional SEO?
Yes. On Google, Ahrefs' data shows AI mostly cites what already ranks. Engines like Perplexity give newer content more of a chance, but in practice an authoritative, technically healthy site starts ahead on any of them. What needs to change is content format and update frequency — not the underlying SEO work.
How often should we refresh content?
The available data points to 30-day cycles: content updated within the last month earns more than triple the AI citations. Practically: pick a handful of your highest-value commercial pages and put them on a monthly review calendar.
Do we need special software for this?
Technical help speeds up the FAQPage schema and the measurement, but most of the work is editorial: structure, clarity, and consistent updating. If your current platform can't handle structured markup, that's a technical prerequisite worth fixing first.
How fast will we see results?
There's no guaranteed timeline — be skeptical of anyone who promises one. It depends on your starting point: if you already rank in the top 10 for your topics, you're competing for AI Overview citations as soon as Google recrawls your content — and engines like ChatGPT or Perplexity run on their own timelines and criteria. If you don't rank yet, there's base SEO work to do first.