Use Perplexity to research a narrow ideal-customer profile, collect the cited public sources behind each suggested business, and verify every detail on the company’s own site before outreach. Then record only verified prospects in your CRM, send a relevant human-written message, and measure replies and meetings. Perplexity can speed up research; it cannot guarantee leads or give permission to contact people.
If you need a list of people to message, Perplexity is the wrong starting point. It can return names and websites quickly, but speed does not make a company a fit, a contact detail current, or an unsolicited message welcome.
It is useful for a harder job: shrinking the time between a clear ideal customer profile and a verified list of businesses worth researching further.
That distinction protects your reputation. It also produces a list your sales team can explain, improve and measure.
Start with a narrow, observable buyer condition
“Businesses in Cape Town that need marketing” is too broad to research or contact well. Define a condition you can confirm from public information.
For example:
- Cape Town architecture practices that have announced a commercial project but show no project-specific landing page;
- Johannesburg solar installers serving commercial property managers, with a dated case study but no clear consultation path;
- Gauteng safety consultancies that publish technical guidance and need the page to earn more relevant third-party mentions.
Write the target in one sentence before opening Perplexity. Include the city, sector, buyer, trigger and evidence you expect to find. This makes the first research result testable rather than decorative.
Ask for evidence, not a finished lead list
Perplexity can search the web and return source URLs. Its documented filters can narrow research by region, language, recency and domain. Use those controls to find starting points, then ask for the evidence behind each result.
Try a prompt like this:
Find Cape Town firms in commercial solar installation that publish case studies for warehouses or office parks. Return the company name, website, public project evidence, location evidence and the source URL for every claim. If any field is not supported by a source, write “not verified.”
For Johannesburg, switch the city and the business condition. Do not ask it to invent an email address, estimate revenue, infer personal details or rank companies by “best.” Those answers sound useful while creating work you cannot defend.
Perplexity’s own prompt guidance recommends acknowledging gaps rather than guessing. Put that rule in your prompt. A blank field is better than a polished mistake.
Verify every candidate on the company’s own site
Treat the answer as a research queue. Open the cited page, then check the business’s own website and public profile for five things:
- The company actually serves the location and type of client you named.
- The project, service or change signal is current enough to matter.
- The website has a genuine gap that IDJoy can address.
- There is a legitimate public route to make contact.
- A message could refer to a specific fact instead of a generic pitch.
Remove any business that fails one of those tests. A smaller list with a clear reason for every entry is more useful than a spreadsheet of 200 scraped names.
Turn research into a human sales brief
Create a simple record for every verified business:
| Field | What to record |
|---|---|
| Account | Business name and website |
| Local proof | Cape Town, Johannesburg or wider service-area evidence |
| Trigger | A dated page, campaign, expansion or content gap |
| Source | The URL and date you checked it |
| Hypothesis | One precise way the business may lose qualified enquiries |
| Next action | Review, referral path or a personalised outreach message |
Keep the source URL beside the note. If a salesperson cannot open the evidence and explain why the business is on the list, the record is not ready.
This approach also keeps AI in the right role. It summarises public context. A person decides whether there is a useful, respectful reason to contact the company.
Make the first message useful enough to deserve a reply
Use the verified evidence to offer a practical observation, not an automated brochure.
For example: “I noticed your Johannesburg HAZOP service page explains the process well, but the consultation route is hidden below a long technical section. We mapped a shorter decision path for specialist firms. Would a one-page review be useful?”
That message is specific, but it does not pretend you know their conversion rate. It gives the recipient a clear reason to say yes or no.
Avoid bulk mail, copied praise and claims that an AI tool “found” a hidden problem. Perplexity may help you discover public information; your judgement must carry the conversation.
Keep POPIA in the workflow
Lead generation and profiling can involve personal information and direct marketing. Public information is not a blank cheque to collect, enrich and message people without considering the rules that apply.
Before outreach, check your lawful basis, consent requirements, opt-out process, retention period and the difference between a business address and a person’s personal information. The Information Regulator’s direct-marketing guidance is the appropriate starting point; get specialist legal advice for a workflow that depends on personal data.
The practical rule is simple: research public business context narrowly, retain only what you need, and make it easy for a recipient not to hear from you again.
Measure qualified next steps, not list size
Track the stages that reveal whether the research is working:
- verified accounts added;
- messages sent after human review;
- positive replies;
- booked conversations;
- opportunities that match your service;
- work won and the original source of the account.
If the list produces replies but no suitable conversations, tighten the trigger or target. If it produces no replies, inspect the relevance and offer before buying more AI credits.
Use the same principle for being found in Perplexity
There is a second opportunity here. The businesses you want to win are also using AI search to research suppliers. A clear service page, consistent local information, original evidence and credible third-party links make it easier for an answer engine to understand what your business does.
No link or page guarantees a Perplexity citation. What you can build is a site with facts worth checking, a local footprint worth trusting and a conversion path that turns a referred visit into an enquiry.
IDJoy can test how eight answer engines currently describe your business, identify what blocks them, and plan the evidence and link-building work worth doing next. See the AI search visibility scopes.