You added a chat widget to your website. Maybe a WhatsApp bot too. Ask yourself one question: what has it actually booked?
If the honest answer is "not sure", you are not alone, and you are not wrong to feel uneasy. Most AI advice is written for banks and big retailers, not for an owner-run business with a full diary and a small team.
We sell AI website agents, so take this with that in mind. But the test below works on any tool, including ours. If a tool can't pass it, cancel it.
The return
You pay a monthly fee for a chatbot and can't say what it has brought in.
That is the most common AI story right now. BCG's 2025 study of 1,250 senior executives found 60% of companies report minimal revenue and cost gains from AI. Only 5% get value at scale, and they see 1.7 times the revenue growth of that 60%.
The difference is rarely the tool. It is whether anyone decided what the tool was for.
Customers ask AI first
Your website traffic is flat or falling, but you're not sure why.
Part of the answer is that people now get answers without clicking. Bain's February 2025 research found about 80% of consumers rely on AI summaries or other no-click results in at least 40% of their searches. About 60% of searches now end without a visit to any website, and Bain estimates organic traffic is down 15% to 25%.
So the visitors who do arrive matter more. And whether AI tools mention you at all matters more still.
POPIA and the widget
Your website chatbot collects names, numbers and questions. Do you know where they go?
Under POPIA, your business stays the responsible party for that information, even when a software provider runs the bot. As Schindlers Attorneys summarise, section 18 requires you to tell people what you collect and why, generally before you collect it. Section 19 requires reasonable technical and organisational safeguards. Sections 20 and 21 require any operator, such as a chatbot provider, to keep the information confidential and secure.
A chatbot is a data collection point. Treat it like one.
This is general information, not legal advice. Speak to your attorney about your own obligations.
Who answers at 9pm
A rival seems to win the jobs you quoted on, and they reply faster.
Speed has always mattered. In the study behind Harvard Business Review's "The Short Life of Online Sales Leads", researchers sent test enquiries to 2,241 US companies. Only 37% replied within an hour, and the average reply took 42 hours. A separate study of 1.25 million sales leads found firms that replied within an hour were nearly seven times as likely to qualify the lead as firms that waited an hour longer.
The study is old and American, but the pattern holds anywhere: the first business to answer properly usually gets the conversation.
A ten-minute self-check
Under all four worries is the same gap: you have the AI tool, but nobody can say what it has changed. Before you buy or cancel anything, write down three lines:
- What AI tools touch your customers: website chat, WhatsApp, email, phone.
- The one number each one should move, and last month's figure.
- Who checks that number, and where customer data from those tools is kept.
If all three come easily, you're ahead of most businesses. Stop reading here.
If a line stalls, that's where to start. Pick one tool, one owner and one number, and track it for 30 days. If the number doesn't move, the tool goes.
One number, measured honestly, tells you more than any sales pitch. Including ours.
Sources
- BCG press release, 30 September 2025: AI leaders outpace laggards
- Bain & Company, February 2025: Goodbye clicks, hello AI
- SmartCompany, July 2011: Are you letting online leads go cold? (reports the HBR study's figures)
- Oldroyd, McElheran & Elkington, "The Short Life of Online Sales Leads", Harvard Business Review, March 2011
- Schindlers Attorneys: The conditions of lawful processing of personal information