seo

The AI search shift: a get-up-to-speed guide

John Dwyer John Dwyer | Digital Strategist 9 min read

Google’s Search box just had its biggest redesign in 25 years. AI Mode crossed a billion monthly users in twelve months. Search queries doubled last quarter alone. The blue links page is being replaced with custom mini-apps generated on the fly for every query. If your team is responsible for being found online, the rules underneath your website just shifted, and this guide is the fastest way to get your head around what changed and what to do about it.

The headline numbers

1 billion
monthly active users of AI Mode, one year after launch
2x
growth in AI Mode query volume last quarter alone
~50%
of search clicks now intercepted by AI answers

How fast this is moving

AI Mode query volume by quarter (indexed, illustrative)

What this means for you The product crossed a billion users inside a year. Whatever you think the long-term impact will be, it's compressing into a shorter window than the last few search shifts did.

Why this matters

Google states the growth as 'more than doubled every quarter since launch'. We've indexed the first quarter at 100 to show the shape. The absolute numbers aren't in the report, but the curve is the story.

Old playbook vs new playbook

The whole shift fits on one table. Print this. Pin it.

What you used to optimise forWhat you optimise for now
Keyword rankingsEntity citation frequency in AI answers
Organic sessions and clicksAI inclusion rate and impressions
Backlinks from any domainMentions on platforms AI trusts (Reddit, YouTube, Quora)
“Our Services” overview pagesHyper-specific service pages per industry or location
Bounce rateConversion depth from AI-referred visitors
Traditional CTRBranded search lift after an AI citation
Generic schema markupEntity-consistent JSON-LD with sameAs links

What it looks like in the data

What happens to a typical search result page in 2026

What this means for you Roughly half of searches now end without a click to any website. The visitor got the answer inside Google's interface. This is what 'traffic down, leads steady' actually looks like in the data.

Why this matters

The 43 to 60 percent range comes from several independent studies tracking AI Overview impact through 2025 and into 2026. We've used 52 percent as a working midpoint. The remainder splits between organic clicks and ads.

Conversions don’t necessarily follow traffic down. AI-referred visitors who do click through tend to arrive much closer to a decision. Early studies suggest they convert at multiples of standard organic traffic, though the data is single-source and worth treating with caution.

The change in the right metric is the takeaway. We’ve spent a decade telling business owners that traffic is the proxy for digital health. From 2026 onward, the better question is: when an AI is asked about my category, does my business get cited?

The 7-point readiness audit

This is the operational core of the piece. Tick the boxes that already apply to your business. Score updates live. Expand each row for the implementation detail your team can run with.

0 / 7
Tick the items that already apply to see where you sit.
  1. How to do this

    Use the schema type that fits your category (ProfessionalService, Restaurant, AutomotiveBusiness, and so on), not the generic LocalBusiness parent. Include the sameAs attribute linking to your official LinkedIn, Facebook, Google Business and YouTube profiles. Validate every page with the Schema.org Validator and Google’s Rich Results Test. This is the single highest-impact action you can take this quarter.

  2. How to do this

    One exact format. Same punctuation, same suite number, same phone format. Used identically on Google Business, LinkedIn, Yelp, Trustpilot, YouTube and your industry directories. AI agents cross-reference these to decide how confidently to recommend you. Inconsistency reads as low confidence. Audit by exporting your listings and diffing them in a spreadsheet.

  3. How to do this

    Mine actual customer questions from your email inbox, Google’s People Also Ask box, and the calls your team takes. Three to five questions per page, answers under thirty words each, leading with a direct factual statement. Apply FAQPage JSON-LD markup. This is the format conversational AI answers feed on.

  4. How to do this

    Remove tag pages, author archives, utility pages, 404s and 301s. Every URL in the sitemap should be canonical and live. Resubmit to Google Search Console. Boring work, real impact on crawl efficiency. Set a calendar reminder for the next audit in six months.

  5. How to do this

    Stop writing “Our Services” overview pages. Write a page per service per industry per location where it makes sense. The agent that gets briefed “find me a marketing agency for a Melbourne hospitality group” needs a page that says exactly that. Each page gets its own H1, its own meta description, its own real content. Generic catch-all pages are invisible to AI search.

  6. How to do this

    If your business name shares a word with something else, state your category in the first sentence of the About page. Use sameAs or knowsAbout in schema pointing to Wikipedia or Wikidata entries for your industry concepts. Link to chambers of commerce, licensing boards or industry associations where relevant. All of this anchors your professional identity in the AI’s trust graph.

  7. How to do this

    AI models cross-reference these platforms for human validation of your existence. A YouTube channel with a clear description matching your website entity is worth more to AI trust scoring than most people realise. Being mentioned in relevant subreddits counts. Pick two platforms, schedule one piece of content per month per platform, sustain it for a year.

What’s not on the list

A couple of things being talked about that we’d deprioritise.

Where this leaves your team

Search has been changing every year since we started doing this work. Each shift looks scary from inside it and ordinary from a few years out. This one is bigger than most, and the businesses that pay attention come out ahead because the rules favour clarity and specificity over volume.

The work, in one sentence: be radically specific about who you serve, make your data consistent across every platform you appear on, and build presence on the channels AI cross-checks for human validation.

If you’d like a hand walking your team through the audit, that’s something we do for Australian businesses. The contact form on the site goes straight to me.

Sources

  • Google, “Search at I/O 2026”. blog.google/products-and-platforms/products/search/search-io-2026
  • OtterlyAI GEO Study (February 2026). AI bot traffic and llms.txt request rates.
  • AEM CDN Log Audit (August 2025). 30-day enterprise crawl analysis.
  • Reboot llms.txt experiment (March 2026). Uncrawled URL visibility test.
  • Independent zero-click studies, 2025 to 2026. Methodologies vary; the 43 to 60 percent range is taken as the working envelope.

Talk to us

Book your discovery call

Pick a time that suits you and we will have a straight conversation about where your business is at and whether we can help. No obligation, no pitch.

  • No obligation A straight chat about where your business is at. No pitch deck, no pressure to sign anything.
  • Fifteen focused minutes We keep it tight and respect your time, so you get value even from a short call.
  • Clear next steps You leave with at least one practical idea you can act on, whether or not we work together.

Your consultation will be with

John Dwyer, Smashed Avo John Dwyer Fractional Digital Strategist