Skip to content
Start a conversation
← All articles
SEOAI

How SEO changed and what is important for 2027

Buyers stopped finding vendors through links. What changed in SEO, the five things that decide whether AI puts you on the shortlist in 2027, and where to start.

Armando Valenzuela Williams7 min read
Abstract acrylic painting with red, yellow, blue and purple brushstrokes

Your buyers stopped finding vendors through a list of links. At Kirana Labs we see the effect from the operations side, through Digital Enterprise, our consulting line: companies with a good product and a decent website, and buyers who never arrive.

In the U.S., 68% of Google searches ended without a click between January and April 2026, according to SparkToro and Similarweb. The answer now appears before the links, written by Google's AI or by ChatGPT, and the buyer's shortlist gets built there.

This article covers what changed, the five things that matter for 2027, and why most of them are an operations problem, not a marketing one.

How SEO changed

For twenty years, SEO was a chain: rank, get the click, get the visit, get the form. In 2026 the chain broke at the click.

Ahrefs measured 300,000 keywords and found that when an AI Overview appears, the top-ranking page gets 58% fewer clicks. A year earlier, the same study measured 34.5%. Put plainly: for every 100 clicks the first result used to earn, Google now keeps 58.

Google's AI Mode goes further. In a field experiment with 1,100 U.S. users, researchers from Penn and Northeastern found that moving searches into AI Mode cut clicks to outside websites by 18.8 percentage points.

Buyers moved at the same time. Forrester's State of Business Buying, 2026 reports that generative AI is now the most cited source buyers use to research a purchase. And in 6sense's 2025 buyer data, 95% of the vendors that won a deal were already on the buyer's shortlist on day one.

The conclusion: the shortlist now gets built inside an answer, before anyone visits your site. Rankings and traffic still matter, but they measure less every month.

What is important for 2027

Five things decide whether an AI puts you on the shortlist. None of them is "write more blog posts."

1. Your company is an entity, and it has to say the same thing everywhere

AI models cross-check what your website, your Google Business Profile, LinkedIn, directories and Wikidata say about you. If one says "software development," another says "IT consulting" and a third has an old address, the model can't tell which is true. So it describes you wrong, or it leaves you out.

We apply this to ourselves. It's why our rule is always "Kirana Labs" and never "Kirana" alone: the short name is ambiguous for search engines and for LLMs.

2. Structured data is infrastructure, not a plugin

If a machine can't read it, it doesn't exist in the answer. That means schema markup for your organization, services and products. It means pages that load their content without depending on JavaScript, and prices, hours and catalogs written as text, not as images or PDFs. It also means a robots.txt that lets AI crawlers in.

3. Original data beats content volume

An AI can summarize "10 tips for choosing an ERP" without citing anyone, because a thousand pages say the same thing. It can't do that with a number only you have. A line like "in 18 of our last 25 projects, the problem was duplicated data, not missing software" gets cited, with your name next to it.

4. What others say about you outweighs what you say about yourself

Your site claims what you do. Reviews, directories, press, partner pages and clients who mention you by name confirm it. Models build a recommendation from that consensus, so a company nobody else talks about is a weak candidate, however good its website is.

5. Measure what's left

If clicks fall, traffic stops being a reliable scoreboard. Three signals hold up better:

  • Branded search. People who search your name already decided you're a candidate. Track that trend month over month.
  • "How did you hear about us?" One required field in every form. It's the only way to see the shortlists that formed inside ChatGPT.
  • AI referrals. Visits from chatgpt.com, perplexity.ai and Gemini show up in your analytics. Separate them from the rest.

The question is no longer "where do we rank?" It's "when someone asks, do they hear our name?"

Why this is an operations problem, not a marketing one

Look at the list again. Points 1, 2 and 5 aren't copywriting. They're about whether your company's data is consistent, readable and measured. That's operations work.

Inside most companies we work with through Digital Enterprise, sales, operations and finance see three different numbers for the same thing. Search in 2027 exposes the outward version of that problem. The website, the catalog, the Google listing and the distributor's directory each say something different, and the AI doesn't know which one to believe.

The fix is the same in both directions: decide which data is the master, connect what already exists, and keep it updated without anyone copying and pasting. A product catalog that lives in your ERP and publishes itself to your website, with structured data, makes the operation faster and makes the company readable to an AI. One project, two results. We cover the first step in Before you automate: what has to be in order in your data.

That's the work behind Digital Enterprise: better systems, well-designed processes and real-time visibility, not more software.

Where to start

You don't need a 2027 strategy deck. Four steps, in this order:

  1. Ask ChatGPT, Gemini and Perplexity what your company does and who they'd recommend in your category. Write down everything that's wrong or missing.
  2. Write one description of your company: name, what you do, for whom, where. Make every profile say exactly that.
  3. Add structured data to your website for your organization and your main services.
  4. Add the "How did you hear about us?" field to your forms this week.

Steps 2 and 4 take an afternoon. Step 3 takes a developer a few days. The hard part usually shows up behind them: the catalog, the prices and the systems that don't agree with each other. That part is what we fix in Digital Enterprise.

Frequently asked questions

Is SEO dead in 2027?

No. Rankings still feed the answers AI gives, and branded, local and transactional searches still get clicks. What changed is that the click is no longer the main result to measure.

What is the difference between SEO and being recommended by AI?

SEO gets your page ranked. Being recommended means an AI model names your company when a buyer asks about your category. The second depends on consistent company data, structured pages and what third parties say about you.

Does structured data help with ChatGPT and Google AI Overviews?

It helps machines read who you are and what you offer without guessing. It doesn't guarantee a mention, but missing or contradictory data makes one less likely.

How do I know if AI is sending me customers?

Track visits from chatgpt.com, perplexity.ai and Gemini in your analytics, watch branded search over time, and ask every lead how they heard about you.

About Kirana Labs

Kirana Labs is a nearshore software development and automation agency based in Monterrey that helps companies expand margin, founders get to revenue, and technical teams let go of production with confidence.

If the data behind your website, catalog and systems doesn't agree with itself, schedule a call.

Sources

Start a conversation.

Tell us the number you want to move: a launch date, a cost, a cycle time, a cash position. We'll come back with a baseline, a plan, and what we'd be accountable for.