Generative Engine Optimization · South Africa

When buyers ask AI who to buy from,
your competitors get named.

What GeoRank does

GeoRank is a Generative Engine Optimization (GEO) practice that makes South African B2B and e-commerce brands citable by ChatGPT Search, Perplexity, Google AI Overviews, Gemini and Claude. Engagements start with a 4-pillar technical audit scored 0 to 10, a tracked set of 25 to 50 buyer prompts measured against named competitors, and a 30-day fix roadmap.

Book a Visibility Call See the 4 pillars

30 minutes. We run live prompts on your category before the call.

What Generative Engine Optimization actually is

GEO is the practice of making a site retrievable and quotable by AI answer engines. Not ranked by them. Quoted. When someone asks ChatGPT, Perplexity, Google AI Overviews, Gemini or Claude which supplier to use, the engine composes one answer and names a few companies in it. GEO is the work that decides whether you are one of them.

It is not a rebrand of SEO, and it does not replace it. The two share a foundation, a site that loads, that crawlers can read, that says true things. Above that foundation they diverge, because a ranking algorithm and a language model want different things from a page. One is choosing a link to list. The other is choosing a sentence to repeat.

Traditional SEO GEO
What you compete for A position in a list of links A sentence inside the answer itself
What the buyer sees Ten results, and they choose One answer naming three or four companies
The unit of success A click A citation, which may never produce a click
What you optimise Keywords, links, page experience Crawler access, fact density, answer capsules, entity schema
The main metric Keyword rankings and organic sessions Citation rate across a fixed prompt set, against named competitors
How you find out you are losing Rankings fall, and the tool tells you Nothing tells you. There is no referrer on an answer you were left out of

That last row is the whole problem. A lost ranking shows up in a dashboard. A lost citation leaves no trace at all, which is why the only way to know is to ask the engines directly and write down what they say.

Three numbers that frame the work

What gets checked before anything else, where the citation is actually won or lost, and the one assumption behind any revenue estimate we put in front of you.

6

AI crawlers must be explicitly allowed before anything else matters: GPTBot, ChatGPT‑User, OAI‑SearchBot, PerplexityBot, ClaudeBot and Google‑Extended. A browser request will not reveal a block: the test has to send the real user-agent strings.

5

Engines where the citation is won or lost: ChatGPT Search, Perplexity, Google AI Overviews, Gemini and Claude. Each retrieves differently, so being cited by one tells you nothing about the other four.

~30%

The share of category search volume we model as migrating to AI answers when we size your opportunity. Your own volume and average order value replace our defaults in the real estimate.

Sources, in order: the crawler list is the set our Pillar 1 audit verifies on every engagement; the engine list is who we track citations across; the migration factor is the planning assumption in our opportunity model, not a measured figure. We recompute it with your own volume and order value. You will not find a South African adoption statistic on this page yet, because we have not measured one. When we have, we will publish the method with it.

Sounds familiar?

  • You ask ChatGPT who the best supplier in your category is, and it names everyone but you.
  • Your rankings are fine, but organic traffic keeps sliding and nobody can explain where it went.
  • An AI engine describes your product with specs, prices or policies that are wrong.
  • Your agency reports on keywords, and has no answer when you ask about AI search.
  • Your specs, stock and prices only appear after JavaScript loads, so crawlers see an empty page.
  • You suspect you are losing deals in a channel you cannot currently measure.

We do not guess at AI visibility.
We audit it across four pillars

Each pillar is scored, aggregated into a 0 to 10 rating, and mapped to a maturity tier that decides the scope of work. This is the same framework we run on every engagement.

Technical Crawlability

If an AI crawler cannot reach or parse your raw HTML, you are invisible to AI search and nothing else on this page can help. This pillar gates the other three.

  • All six AI user agents allowed in robots.txt
  • No CDN or firewall challenge served to AI user agents
  • Prices, specs and tables present in server-rendered HTML
  • No parameter crawl traps, canonicals in place

Fact Density and Information Gain

Language models extract concrete facts and ignore marketing adjectives. "Highest quality" is unciteable. "100% extra-fine merino, 280 GSM, OEKO-TEX certified" is three facts an engine can quote.

  • Exact SKU and MPN visible on every product page
  • Every adjective replaced with a measurable value
  • Compatibility and fitment as structured HTML tables
  • Certifications and standards named explicitly

Answer Capsules and Extraction

AI engines look for a concentrated 40 to 60 word summary directly under the H1 to quote as a source. No capsule means nothing clean to cite, so the engine cites a competitor that has one. The grey block under this page's headline is exactly that.

  • A capsule under the H1 of every category and service page
  • Identity, specs and practical facts in the capsule
  • Answers front-loaded into the first sentence
  • Key facts in bullets and tables, not prose walls

Entity Topology and Schema

Visible HTML tells a crawler what a page says. JSON-LD tells it what the page is and how it connects to entities the model already trusts. That connection is what turns a page into a recommendation.

  • Specific schema subtypes, never a bare LocalBusiness
  • sameAs array anchored to Wikidata, GBP and LinkedIn
  • Offer, priceCurrency and availability on every product
  • Return policy markup for AI shopping surfaces

How an engagement runs

Measure first, fix second, defend third. You see the baseline before you commit to any of the work.

Week 1

Baseline and audit

We run your category's real buyer prompts across the major engines and record who gets cited and who is missing, with screenshots. In parallel we score all four pillars and assign a maturity tier.

Week 1

Opportunity and roadmap

You get a one-page report: findings per pillar, the visibility evidence, a revenue-at-stake estimate with its assumptions written out, and a prioritised 30-day fix list your developers could execute without us.

Weeks 2 to 5

Implementation

Answer capsules written and shipped, JSON-LD deployed and validated, fitment and spec tables built, crawler access unblocked, and off-site entity seeding across Wikidata, directories and authority profiles.

Ongoing

Tracking and defence

The same prompt set re-runs every month so Share of Voice is comparable over time. You get citation rate against named competitors, coverage per engine, a work log, and next month's priorities.

What the first audit puts on your desk

Concrete artefacts, not a slide deck about the future of search.

4 pillars

Each one scored and aggregated into a single 0 to 10 rating, plus the maturity tier that determines whether you need a full overhaul, a targeted fix package, or defensive maintenance.

25–50 prompts

A tracked set built from the questions your buyers actually ask, recorded with screenshots so month two is comparable to month one.

30 days

A prioritised remediation roadmap written as developer-ready specs, with the code templates for capsules, schema and spec tables included.

Frequently Asked Questions

No. SEO competes for a position in a list of links. GEO competes to be the source a model quotes when it writes a single answer. The work is different in kind: crawler access for AI user agents, extractable fact density, answer capsules under the H1, and deep entity schema. Strong SEO helps, but plenty of sites that rank on page one are never cited by an AI engine, and that is exactly the gap we audit.
Brands whose buyers research before they buy: B2B suppliers and distributors, e-commerce catalogues with real technical specs, and considered-purchase service businesses. The more your category involves compatibility, certification, specification or comparison, the more an AI answer decides the shortlist, and the more there is for us to work with.
With a fixed prompt set. We build the natural-language questions your buyers actually ask, run them across the engines under controlled conditions, and record which brands are cited and which are omitted, with screenshots. Re-running the identical set each month makes Share of Voice trendable. It is the same discipline as rank tracking, applied to answers instead of links.
Crawler access problems are fixed in days and are often the single biggest unlock, since a blocked user agent means zero visibility regardless of content quality. Capsules and schema take effect as engines recrawl, typically within weeks. Entity authority compounds over months. We report on the same prompt set monthly so you can see the curve rather than take our word for it.
No, and be careful with anyone who does. Generative engines are probabilistic and their retrieval changes without notice. What we guarantee is the work: every pillar remediated to a documented standard, every schema validated, every capsule shipped, and honest monthly reporting whether the number moved up or down. Specific commitments are agreed on the call once we have seen your baseline.
Yes, and some clients do exactly that. The audit is written as developer-ready specs with the code templates included, so an in-house team can execute it without us. We take the implementation on when a team has no capacity, or when the off-site entity work (Wikidata, directories, authority profiles) is outside what they can do internally.
Usually, and it is one of the more urgent reasons to start. Misinformation generally traces back to a stale third-party page, a thin entity record, or a fact that appears nowhere in machine-readable form on your own site. We locate the source, correct the record where it can be corrected, and make the true fact easier to extract from you than the false one is from anywhere else.
No. Keyword tools read clickstream panels, and those panels are thin outside the United States, which makes South African long-tail readings especially unreliable. A zero measures the panel, not the market. We treat prompt volume as something to observe directly in the engines rather than something to look up in a tool built for a different kind of search, and we run the prompts in the language and phrasing your buyers actually use.
Before we speak we run a handful of your category's buyer prompts and check your robots.txt, rendering and schema. On the call we show you what came back, which pillars are costing you citations, and what the first 30 days would cover. If GEO is not the right priority for you right now, we will say so.

Find out what AI says about you

Book a 30 minute visibility call. We will run your category's buyer prompts beforehand and walk you through what the engines actually returned, whether or not you work with us.

Book a Visibility Call