Service / 2026

03 scopes · 8 engines tested

AI search visibility,
measured honestly.

We test whether ChatGPT, Perplexity, Google AI Mode, Gemini, Copilot, Claude and Meta AI currently cite your business, fix what blocks them, and re-test against the baseline. Every scope below states its price. Every claim we cannot support is written down further down this page.

See the three scopes

Why we built this

South Africans are already searching for this service

We watch our own Search Console. Over the past month these queries reached our site, and every one of them arrived from someone looking for help with AI search visibility in South Africa.

IDJoy Search Console · queries received · July 2026
Query Impressions What the searcher wanted
how to improve ai search visibility for south african businesses 14 A method
what makes a business domain look trustworthy? 8 Trust signals
how to get found on perplexity in south africa 5 One engine, named
ai visibility south africa 5 Orientation
top 10 chatgpt search visibility specialists in south africa 3 A shortlist of providers
perplexity visibility optimisation provider cape town 2 Someone to hire in Cape Town
perplexity visibility optimisation providers in johannesburg 2 Someone to hire in Johannesburg
ai search visibility expert near me 1 Local help
perplexity visibility optimisation service in cape town 1 A priced service

Five of those nine queries name a provider, a specialist or a city. Those searchers wanted to hire someone and we had no page to send them to. This page answers that.

01 / 03

AI Visibility Audit

A measured baseline. You find out where you stand across eight engines before you spend anything on fixing it.

R 3,450.00

Once-off diagnostic · written report · 5 working days

What you get

  • Baseline citation test: we ask 20 buying-intent questions a real customer would ask, across ChatGPT, Perplexity, Google AI Mode, Gemini, Copilot, Claude and Meta AI, then record who gets cited today
  • Crawler access check on your domain for OAI-SearchBot, GPTBot, PerplexityBot, ClaudeBot, Claude-SearchBot, Googlebot, Google-Extended, bingbot and Meta-WebIndexer
  • Bing Webmaster Tools AI Performance report connected and a first reading taken, covering Citation Share and Grounding Queries
  • Query-language gap analysis comparing the words on your pages against the words your buyers type
  • Competitor citation check: which South African businesses the engines name instead of you, and on which questions
  • Written report with a prioritised fix list and a copy of the raw test results, yours to keep

Buy this on its own. If you go on to book the Foundation scope within 60 days, we credit the full R 3,450.00 against it.

Book an audit

03 / 03

Citation Tracking

Quarterly measurement after the Foundation work. Answer engines change their sources without notice, so a single test tells you about one day.

R 1,850.00

Per quarter · billed quarterly · cancel with 30 days notice

What you get

  • The same 20 questions re-run across all eight engines every quarter, so results stay comparable
  • Movement report against your original baseline: citations gained, citations lost, competitors who displaced you
  • Bing AI Performance figures pulled into the same report, plus Google's generative AI performance data once it reaches South Africa
  • One prioritised content recommendation per quarter, based on what the test found rather than a content calendar
  • A running record of every test, so you can see twelve months of movement rather than a snapshot

Requires the Foundation scope first. Tracking a domain we have not fixed produces a report you cannot act on.

Add tracking

Coverage

The eight engines we test, and how each one finds you

Crawler names below come from each vendor's own documentation. Any agency quoting you a crawler that does not exist is guessing.

ChatGPT Search

Live web + trained memory

Crawls with OAI-SearchBot and trains on GPTBot. Cites fewer sources than any other engine, often one or two, which raises the bar for getting picked.

Perplexity

Retrieval first

Uses PerplexityBot and Perplexity-User, and runs its own index of more than 200 billion URLs. Published source counts per answer range from four to twenty-two depending on who counted, so we report the range rather than pick one. Freshness is documented.

Google AI Mode

Google index

Reached through Googlebot. Google-Extended does not govern this surface, whatever you have been told. A Search Console setting added on 3 June 2026 does, and you must not be opted out of it.

Gemini

Grounded search

No crawler of its own. Grounded citations return a Google redirect rather than your domain, so your analytics undercount it. Published source counts per answer disagree, so we report what we measure instead of quoting a figure.

Microsoft Copilot

Bing index

Runs on bingbot. No separate Copilot crawler exists. Bing reports your Citation Share for free, which makes this the one engine you can measure today.

Claude

Three-bot split

Uses ClaudeBot, Claude-User and Claude-SearchBot, restructured in February 2026. Anthropic has not published how it selects sources, so we test rather than assume.

Meta AI

WhatsApp and Instagram

Runs five crawlers. Meta's documentation names Meta-WebIndexer as the one that helps it cite and link your content. Reaches South Africans inside WhatsApp, and sends no referrer your analytics can identify.

Grok

Testing only

xAI has not published verified crawler names, and independent log analysis suggests inconsistent robots.txt behaviour. We measure whether it cites you and make no access claims.

Read this before you buy

What we cannot guarantee

01

No engine sells placement

OpenAI, Google, Perplexity, Anthropic, Microsoft and Meta run no paid route into their cited sources. Nobody can buy you a citation, and nobody can promise one. An agency offering guaranteed citations is describing a product that does not exist.

02

Citations move on their own

Ask an engine the same question twice and you can get two different source lists. We test 20 questions across eight engines to average out that noise, and we report the pattern rather than a single result. Treat any month-to-month figure as a trend, not a score.

03

Established domains hold an advantage we cannot delete

The largest published study of AI citations, covering roughly two million citations across four engines, found that domain-level authority outweighs the strongest page-level fix by six times. A domain registered last month competes against domains the models have known for years.

We can raise how clearly and specifically your site answers a question. We cannot hand you ten years of reputation in six weeks.

04

Measurement in South Africa remains partial

Bing's AI Performance report works here now. Google announced its generative AI performance report on 3 June 2026 and has rolled it out to the United Kingdom, India, Switzerland and the United States. South Africa has not received it. Until it arrives, part of your Google AI visibility stays invisible to everyone, including us.

Third-party citation trackers run roughly 70 to 90 percent accurate. Independent testing of one popular tool found a 97.5 percent discrepancy against actual ChatGPT mentions. We run our own tests and show you the raw results instead.

05

Results arrive in quarters

Because domain authority dominates, this work compounds rather than switches on. Expect the first measurable movement at the 60-day re-test and a clearer picture after two quarters. Anyone promising results in a fortnight is selling you the audit and hoping you forget the re-test.

Work we will not sell you

  • An llms.txt file as a visibility lever. Four independent studies covering more than 400,000 domains found over 97 percent of these files are never fetched, and Google has confirmed it ignores them.
  • An FAQ-schema-and-Core-Web-Vitals package sold as AI optimisation. Those correlations disappear once you control for domain authority. We will still fix your page speed, as web performance work, priced as such.
  • Optimisation for a CopilotBot. No such crawler exists. Copilot reads the Bing index.
  • A guaranteed number of citations, mentions or AI referrals in a contract.
  • Case studies we cannot show you. Ask us for client names and we will give you names.

Fit

Who this work suits, and who should wait

Book it if

You sell a considered service. Buyers research before they call, and they now research inside an assistant. Accountants, attorneys, engineers, medical practices, industrial suppliers and B2B service firms see the clearest benefit.

Book it if

You already rank on Google. Google AI Mode and Gemini pull from the same index as classic search, so existing ranking gives you a running start.

Wait if

Your website cannot convert a visitor yet. Sending AI referrals to a site that does not explain your offer wastes the visibility. Fix the website first, and we will say so rather than take the booking.

Wait if

You have no proof to publish. Engines cite specifics. With no client results, no pricing and no data to put on the page, the Foundation scope has nothing to work with. Run the R 3,450.00 audit and decide from there.

Start with the baseline

Find out who the engines name instead of you.

Book the R 3,450.00 audit