To make a South African business easier to discover and assess in Perplexity, check that PerplexityBot can reach key pages, explain services and local details clearly, and support claims with evidence. Then test a fixed set of prompts and record the answers and cited URLs. These steps do not guarantee a mention or citation.
Perplexity’s public material reviewed for this guide describes its search product, citations, crawlers and publisher partnerships. I found no self-serve South African business directory for companies to submit themselves to. That is a finding from the public documentation checked on 5 November 2026, not a claim about private partner arrangements.
Perplexity Search returns answers with links to sources. You can make key business information accessible and useful to people who follow those links. You cannot guarantee that Perplexity will mention or cite your business.
Updated 5 November 2026: clarified what Perplexity documents about its crawlers and citations, and separated documented product behavior from content recommendations to test.
How Perplexity discovers pages
Perplexity documents two agents.
PerplexityBot
Perplexity describes PerplexityBot as a crawler used to surface and link websites in Perplexity search results. Allowing it to reach a page makes the page available to that crawler; it does not establish that the page will appear in an answer.
Perplexity recommends allowing the bot in robots.txt and permitting requests from its published IP ranges.
A simple allow rule looks like this:
User-agent: PerplexityBot
Allow: /
A specific allow rule may not override a broader firewall or content delivery network (CDN) block. Check the full path from the bot to the page.
Perplexity-User
Perplexity-User can fetch a page in response to a user action. Perplexity says it generally ignores robots.txt for these user-requested visits. This user-triggered fetch is different from routine crawling by PerplexityBot.
It does not replace normal indexing. A business that wants recurring discovery should still allow PerplexityBot.
Verify the request
A user-agent name can be copied. Perplexity publishes current IP ranges for both agents and recommends matching the name with the IP address in web application firewall (WAF) rules.
Do not build a permanent firewall rule from a copied list in a blog article. Open the current IP-range links in Perplexity’s crawler documentation when you check your rules; the published ranges can change.
If you need to read server or WAF logs, ask the person who manages your hosting to check the timestamp, source IP, user-agent, requested URL, response status and whether the request was allowed, blocked or challenged. A successful response shows page access under those conditions, not an answer citation.
If a verified request was blocked or challenged, ask the host to identify the rule, compare the source IP with Perplexity’s current published ranges and adjust the rule where appropriate. Then retest the same URL and record its response status.
Check access before rewriting content
Run these checks on the pages that describe your main services:
- The page returns a successful status code.
robots.txtdoes not block the path.- The page has no
noindexdirective. - The canonical points to the correct clean URL.
- The main answer appears in crawlable HTML.
- A bot does not receive a challenge page instead of the content.
- The page does not redirect through a loop.
- The sitemap includes the final URL.
- Internal links point to the final URL without tracking parameters.
Make key service pages clear and useful
Perplexity’s documentation explains its crawler roles and that Search answers include source links. It does not provide a checklist that guarantees a page will be cited. The following are editorial checks for helping a prospective customer understand a service page; treat any effect on Perplexity answers as a hypothesis to test.
A service page that says “we deliver innovative solutions tailored to your needs” does not tell a prospective customer what the service includes or who it suits.
A useful service page answers concrete questions:
- What does the service include?
- Who is it for?
- Which locations does the company serve?
- What does it cost or what changes the price?
- What evidence supports the method?
- Which standards, regulations or constraints apply?
- When should a buyer choose another option?
- What changed in the most recent update?
Specific content helps a person assess whether the service fits. It may also give a search system clearer material to retrieve, but this guide does not claim that any individual page element causes a citation.
Include details that matter to South African buyers
Local details help readers judge whether a provider serves their needs. Whether those details change Perplexity’s answer depends on the query and the sources it retrieves.
Useful local evidence includes:
- Prices in Rand with a checked date.
- South African service areas.
- Local payment, delivery or procurement conditions.
- Relevant legislation or standards.
- Local case studies.
- South African buyer questions.
- First-hand observations from projects in Pretoria, Johannesburg, Cape Town, Durban or another served market.
Do not add “South Africa” to every heading and call it local expertise.
Explain what changes because the business operates here.
Use one definitive page per intent
Compare the buyer intent before creating another page. If two pages answer the same need, combining them can reduce the work of keeping pricing, proof and updates aligned. Keep pages separate when the service, audience, location or decision differs.
For example, one pricing page can answer close variations such as:
- web design prices South Africa;
- website design costs South Africa;
- web design rates South Africa;
- cost of web design in South Africa.
This is reader-centered site-planning advice, not a documented Perplexity ranking factor or a guaranteed citation method.
Add evidence that survives comparison
If a prompt asks for a comparison, a provider’s public information should make its offer assessable. The examples below are content ideas, not documented Perplexity ranking factors.
Useful evidence includes:
- Public package scope.
- Ownership terms.
- A project process.
- Named industries and locations.
- Client-approved case studies.
- A comparison table with stated criteria.
- An original checklist or test.
- A founder or expert byline.
- Publication and update dates.
- Links to primary sources.
Avoid invented performance numbers, unsupported “best” claims and testimonials without context.
A precise limitation tells a buyer when a service may fit and when to consider another option:
IDJOY builds lead-generation websites for South African service businesses. It is not positioned as an enterprise ecommerce platform agency.
Strengthen sources outside the website
When you inspect captured answers, note whether the cited URLs come from a directory, business profile, media page, review site or professional organisation. Do not assume these sources appear for every prompt.
Correct:
- Old phone numbers.
- Previous domains.
- Wrong locations.
- Services the business no longer offers.
- Duplicate profiles.
- Different spellings of the company name.
Build legitimate corroboration through work that has value outside AI search:
- Contribute expert comments to an industry publication.
- Publish a case study with a client partner.
- Maintain accurate professional profiles.
- Join relevant industry organisations.
- Speak at a local event and publish the material.
- Earn reviews that describe the service delivered.
Do not buy fake news placements or bulk directory links.
These steps can make public business information more consistent for people and other systems to inspect. Perplexity does not document them as ranking factors, and they do not guarantee a citation.
Build a Perplexity prompt baseline
Use a fixed prompt sheet instead of relying on one screenshot. This guide reports no Perplexity test results. Use the table as a template.
For each prompt, record:
| Field | Record |
|---|---|
| Prompt | Exact wording |
| Date | Day and time |
| Product and mode | Product surface and the mode shown in its interface; use the same one for each comparison |
| Session conditions | Signed-in or signed-out state and location setting, if known |
| IDJOY mentioned | Yes or no |
| IDJOY website cited | Yes or no |
| Other IDJOY source cited | Profile, directory or article |
| Competitors named | Names and order |
| Facts correct | Correct, incomplete or wrong |
| Sources | Every cited URL |
| Page gap | Missing pricing, service, location, proof or comparison |
| Action | Specific change to make |
For each prompt, start a new conversation in the same product and mode, then save the full answer and its visible source links. Record the date and session conditions above. If you cannot identify a setting, write “unknown” rather than guessing. Product results can still vary between runs.
Optional prompt library for a South African service business
This library is optional. For a first baseline, choose four prompts: two non-branded discovery questions, one buyer decision question such as pricing or trust, and one branded accuracy question. Add more only if the first set gives you a reason to investigate another topic. Adapt the examples to your industry.
Discovery
- Which web design companies serve small businesses in Pretoria?
- Who offers web design for South African service businesses?
- Which companies build lead-generation websites in Gauteng?
- Which South African studios combine brand identity and web development?
Pricing
- What does a five-page business website cost in South Africa?
- Compare web design rates in South Africa.
- Which South African web designers publish their prices?
- What should a website quote include before I sign?
Trust
- Which Pretoria web designers let the client own the domain and source files?
- Which providers explain their process and revision limits?
- Which South African agencies publish detailed case studies?
- How can I tell whether a local web design company is trustworthy?
AI visibility
- Which South African companies help businesses get found in Perplexity?
- How can a South African business improve AI search visibility?
- Which local agencies publish an AI visibility testing method?
- Who can improve a website for ChatGPT, Gemini and Perplexity discovery?
Branded accuracy
- What services does IDJOY provide?
- Where is IDJOY based?
- What are IDJOY’s website prices?
- What evidence supports IDJOY’s AI visibility service?
If you run this sample, state the location in the prompt instead of assuming the product uses the device location. Treat these questions as hypotheses; adjust them using actual customer conversations or search data before using them to plan content.
Diagnose the result
The business is absent and no page is cited
Treat this as an observation about the tested prompts, not proof of an indexing or selection cause. Check that relevant pages are accessible and answer the buyer question, then repeat the same prompts under the recorded conditions.
A competitor is named and your article is cited
In these captured answers, the page was cited while the business offer may remain unclear. Check the cited page and whether its service scope and evidence answer the buyer’s question.
The business is named but the facts are wrong
Find the cited source. Correct the source with the strongest apparent influence, then update the website and major profiles.
The business is cited only for branded prompts
In this prompt sample, branded questions returned the business name while non-branded questions did not. That does not establish a general discovery pattern. Check whether the relevant service page answers the non-branded question, then compare another set of captures before changing the content plan.
The business appears once and disappears later
Record the change between captured runs. A small sample does not establish a stable ranking or explain why the answer changed.
Measure traffic from Perplexity
Analytics may show referrals from Perplexity when a user opens a cited link. Referrer behaviour can vary by product, browser and privacy setting.
Track:
- Sessions whose analytics source or referral report shows a Perplexity hostname; record the exact hostname shown and do not assume analytics identifies every visit.
- Landing pages.
- Engaged sessions, if available; use the definition shown by your analytics platform.
- Form starts.
- Form submissions.
- Calls or WhatsApp actions where measured.
- Lead quality.
- Assisted conversions.
Ask new leads where they first heard about the business. A person may discover the company in Perplexity, verify it on Google and return directly later.
What not to buy
Guaranteed Perplexity rankings
No source reviewed for this article establishes a fixed organic position that a business can reserve. Do not accept a guaranteed placement claim without verifiable terms describing what is being sold and where it appears.
Paid submission to organic answers
Perplexity has announced publisher partnerships, including a Publishers Program. The public announcement reviewed here describes a partnership, not a self-serve directory application for South African small businesses. Ask Perplexity directly about current partner opportunities; do not treat the absence of a public submission route as proof that no private arrangement exists.
An llms.txt file sold as a full strategy
Perplexity publishes an llms.txt index for its own documentation. That does not prove that adding the file to a business website creates ranking or citation gains.
Use it as an optional documentation index when it helps users or agents. Do not mistake it for evidence, crawler access or authority.
Bulk AI articles
A summary of common advice may give a buyer little new information. Prioritise material that answers a real customer question with a specific fact, comparison, method or local explanation.
A 30-day Perplexity plan
Week 1: access and a four-prompt baseline
- Check
PerplexityBotaccess. - Review firewall logs.
- Choose four prompts using the starter mix above. The twenty-question library is optional.
Week 2: source map
- Group cited sources by type.
- Correct wrong business details.
- Find prompts where competitors have stronger evidence.
- Choose one commercial gap.
Week 3: publish the better source
- Improve the main service page.
- Add pricing, scope, process or evidence.
- Publish one supporting article only when the intent is distinct.
- Link the article to the service, proof and enquiry route.
Week 4: verify and schedule the retest
- Confirm the page is crawlable.
- Check the sitemap and internal links.
- Record publication and update dates.
- If you choose to monitor changes, repeat the same prompt set on a stated schedule and record the product mode and conditions.
Where IDJOY fits
IDJOY’s AI search visibility work covers crawler checks, prompt baselines, service-page and content revisions, internal links and optional measurement.
IDJOY cannot force Perplexity to cite or recommend a page.
Reports separate captured answers from identifiable referral visits and recorded enquiries; analytics may leave some referrals unknown.
Review the AI search visibility service, then use the AI recommendation scorecard to compare the next test with the baseline.