AI Visibility Engineering: SEO and AEO · Samurai Labs

AI Visibility Engineering: Found by buyers. Cited by AI.

Today

  • 3 of 48AI answers that name the brand
  • 0 of 10Google AI Overviews citing the site
  • 29.5 smobile page load
  • 1keyword ranked, position 54

Month 6

  • 8 of 15buyer prompts that name the brand
  • 3+AI Overviews citing the site
  • under 3 smobile page load
  • Top 10for 20 buyer terms

Sample programme, client anonymised, baseline November 2026. Month 6 figures are targets, not results. Drag the marker through the months.

Why now

  • 94%

    of B2B buyers use AI somewhere in the buying process, up from 89% a year earlier.

    Forrester, Jan 2026
  • 42%

    of B2B decision-makers use an LLM at the very first step of buying.

    Omniscient Digital, Jan 2026
  • 88%

    of tracked US commercial keywords showed a Google AI Overview in August 2026, up from 46% eleven months before.

    Semrush, 400-keyword panel
  • 60%

    of searches on traditional search engines end without a click through to another site.

    Bain & Company, Feb 2025
  • 35%

    higher organic click-through for brands cited inside an AI Overview than for brands left out of it. A correlation, not a proven cause.

    Seer Interactive, Q3 2025

Analyst and vendor studies: read them as direction, not as a forecast for your category. What has not changed is that Google still sends most of the traffic. In June 2025 it sent about 191 billion referrals to the top 1,000 sites, against 1.13 billion from AI platforms (Similarweb). Speed, crawlability and structure still decide whether a page is eligible to be cited at all.

What AI Visibility Engineering is.

AI Visibility Engineering is the work of getting a brand found in Google and named, with a citation, when a buyer asks ChatGPT, Gemini, Perplexity or Google AI Overviews who to shortlist.

SEO makes pages Google ranks. AEO, answer engine optimisation, makes pages AI engines quote. Both need the same three things: fast pages, one page per buyer question, and facts a machine can verify somewhere else.

  • Mentions beat backlinks. In an Ahrefs study of 75,000 brands, AI visibility tracked YouTube mentions (0.737) and branded web mentions (0.664) far more closely than backlinks (0.218).
  • What others say beats what you say. When people ask what customers think of a brand, 82% of the sources AI engines cite are earned media. Social sources grew from 4.9% to 7.2% of AI citations between January and August 2026, while brand-owned content sat at 2.5 to 2.8%.
  • Each engine has its own diet. The source type an engine cites most differs from one to the next, so the same page does not win everywhere.
Top source type by engine: ZipTie, March 2026, updated October 2026. Correlations: Ahrefs, as summarised by Topify, July 2026. Earned media: Omniscient Digital, 23,387 sources. Social share: Goodie, September 2026.
EngineCites mostShare of its citationsWhat that asks for
ChatGPTWikipedia7.8%A consistent company record and reference-grade facts
PerplexityReddit6.6%Real answers in the communities your buyers read
Google AI OverviewsYouTube18.2%Video that answers the query, with a transcript

Six services, one programme.

Start with the audit. Most clients then take the technical foundation and add content, earned presence and tracking as a monthly programme. Every part is run by the team that builds our products.

AI Visibility Audit ↗

01 · Entry product · fixed scope

Where your brand stands in Google and in AI answers today, with a screenshot, a number and a date behind every finding.

  • Prompts50 to 100 buyer questions, run across the AI engines
  • ScoreShare of voice against the competitors the engines name instead
  • EvidenceWho gets cited, from which pages, and why
  • OutputA graded scorecard and the fixes in order of impact
What the audit covers ↗

Technical AEO ↗

02 · Engineering

The part most marketing agencies hand off as a list. We fix it in the code, then test it again.

  • EntityOne fact set for the company, applied to schema and every profile
  • Crawlllms.txt, AI crawler policy, redirects and sitemaps
  • SpeedServer response, page weight and Core Web Vitals
  • TemplatesAnswer-first page types, generated with the right schema
How technical AEO works ↗
03

Answer‑first content

Comparison pages, best-for pages, FAQs, pricing pages and original data a model can lift and attribute. Eight to twelve assets a month, each with a named expert, a dated fact block and an answer-shaped FAQ.

04

Earned presence

Digital PR, listicle placements, review sites, Reddit, Quora, YouTube and LinkedIn, weighted toward the sources each engine prefers. Links are requested, not bought. Paid placements only with your approval, at cost.

05

Local AEO

Google Business Profile, reviews and local citations for businesses that are found on a map before a website. How local AEO works ↗

06

Monthly tracking

Prompt share of voice, citation count, AI-referred traffic and enquiries by source page. One report and one call, every month.

Five steps, measured at each one.

Foundation first, because nothing else works on a slow site with dead links. Then the pages. Then the compounding work that gets them cited.

01

Diagnose

Build the buyer prompt set, capture baseline visibility in Google and across the AI engines, and map the gap to the competitors being named.

02

Fix

The technical and entity foundation, in weeks one to four: speed, redirects, titles, schema, crawler policy and one consistent company record.

03

Publish

Eight to twelve answer-first assets a month. One page per buyer question, so each can rank and be quoted on its own term.

04

Seed

Earned mentions on the sources each engine prefers: trade press, directories, communities and video.

05

Measure

A monthly report and call. We re-run the prompts, re-check the rankings and put next month's effort where the numbers moved.

Foundation

Days 1 to 30

Fix what blocks everything.

  • Server response under 0.5 s
  • Every page titled and described
  • 0 dead URLs
  • Baselines locked

Build

Days 31 to 90

One page per product, problem and service.

  • 30+ indexed pages
  • First top-20 rankings
  • First new AI mention on a tracked buyer prompt

Compound

Days 91 to 180

Rank, get cited, get asked.

  • Top 10 on the agreed terms
  • Named on the tracked prompts
  • Cited in Google AI Overviews
  • Enquiries attributed by page

Targets, not forecasts. Rankings and AI citations are things we control through the work. Enquiries are reported, not promised.

We build the tools that run it.

Because we are an engineering team, the repetitive parts of search run as software. The people on your programme spend their time on judgment: what to build, what to say, whom to ask for a link.

ToolWhat it does
AI Visibility TrackerSends your buyer prompts to ChatGPT, Gemini, Perplexity and Claude every month, logs who is named and which pages are cited, and diffs against last month.
SERP and AI Overview MonitorCaptures live Google results by market with screenshots, records AI Overview presence and the domains it cites, and flags when one of your pages enters the top 10.
Site Health CrawlerCrawls every URL weekly for titles, descriptions, headings, alt text, schema, dead links, redirects and response time, and opens a ticket when something regresses.
Fact Set and Schema GeneratorOne structured fact set for the company, its products, people and proof, rendered as Organization, Service, Person, FAQ and Event schema, llms.txt and profile text.
Page FactoryTemplated page generation from a structured sheet, so every page of a type ships with the same structure and the same schema.
Two-scoreboard DashboardSearch Console, analytics, rank tracking and the AI tracker in one view: rankings, citations and enquiries by source page.

Marketing agencies recommend. We build, ship and measure.

Ranked work, with the numbers.

Two technology companies with precise buyers and modest search volumes. Figures are from a public ranking database, pulled on 3 November 2026.

PushOwl

Product + search programme

From 32 to 81 top-10 rankings in six months. We build product and web surfaces for PushOwl and run its search programme: one page per use case, comparison pages against the alternatives buyers name, answer-shaped help content, and a technical sprint that halved page weight. The site also appears inside Google's AI Overview for its own category query.

PushOwl results, May 2026 against October 2026
MeasureMay 2026Oct 2026
US keywords in the top 103281
Est. organic visits a month80435

153% more top-10 rankings in six months. 408 US keywords ranked in all.

pushowl.com ↗

MagicFit

Launch + search

66 ranked keywords from a new domain, in under a year. An AI fit and size assistant for fashion stores, launched in November 2025 on a new domain with no brand search to start from. We launched with a page per question buyers ask, dated fact blocks, FAQ schema and listings on the directories its buyers browse.

MagicFit results, launch in November 2025 against November 2026
MeasureAt launchNov 2026
US keywords ranked066
In the top 10013
At position one02
Est. organic visits a month0314
magicfit.pushowl.com ↗
  • Bold CommerceProduct + webCheckout, subscriptions and pricing platform for ecommerce brands.
  • SellEzzyProduct + webOur own WhatsApp commerce product, launched in 2025.

Results, not targets. Rankings and keyword counts are from a public ranking database, pulled on 3 November 2026. Organic visits are that database's estimates, not analytics figures.

If the retainer ended in month seven, it would keep working.

That is the standard we build to. Everything we make stays with you, and keeps producing enquiries without us.

  • Pages and templates. Every page, article and template, in your CMS.
  • Profiles and listings. Every one of them, under your logins.
  • The measurement. The tracker, the keyword set, the prompt set and every monthly report.
  • The fact set. One record of the company, so the next person keeps it consistent.

Questions, answered.

AI visibility is how often, and how accurately, a brand is named or cited when someone asks an AI engine a buying question. We measure it by running a fixed set of buyer prompts through ChatGPT, Gemini, Perplexity and Claude, and by capturing Google AI Overviews on the terms your buyers search.

SEO makes pages that Google ranks. AEO (answer engine optimisation) and GEO (generative engine optimisation) make pages that AI engines quote. They need the same foundation: fast pages, one page per buyer question, and facts a machine can verify on other sites. We run them as one programme with two scoreboards.

With a prompt set agreed at the start, usually 50 to 100 buyer questions in an audit and a tracked subset every month. Each prompt is sent to each engine with web search on. We log whether your brand is named, whether your site is cited as a source, and which competitors are named instead, then compare against the previous month.

The technical foundation lands in the first 30 days and is checkable the day it ships. In the sample programme on this page, the first new AI mention on a tracked buyer prompt is a day-90 target and citations in Google AI Overviews are a month-five and month-six target. These are targets we report against, not forecasts.

No. Nobody controls what a search or AI engine returns. We commit to the work, to exit measures for each phase, and to reporting the same numbers with the same tools every month. Enquiries are reported, not promised.

The audit is a fixed scope at a fixed fee. A programme is priced after the audit, once we both know what needs fixing and building. Send an enquiry and we will scope it with you.

Yes. Local AEO covers Google Business Profile, reviews and local citations for businesses that are found on a map before a website, such as restaurants, schools, clinics and shops.

Not in the answer yet?

(PING US)

Get an AI visibility audit ↗

Or reach us at contact@buildwithsamurai.com

Get an audit ↗