Chatbots are easy to buy and easy to regret. The ones that work share a few traits, and you can check for them in an afternoon, before spending anything.
1. Do you get the same questions again and again?
Open your last 100 customer messages. If 30 or more are variations of the same ten questions (price, opening hours, delivery, booking), a bot can take a real load off your team. If every message is unique, it cannot.
2. Are the answers written down?
A chatbot is only as good as the documents behind it. If the answers live in your staff’s heads, write down the top 20 first. That exercise is useful even if you never build a bot.
3. Do answers depend on private account data?
“What are your hours?” is easy. “Where is my order?” needs a connection to your order system. Both are possible, but the second is a bigger project, so scope it separately.
4. What happens when the bot does not know?
Decide in advance who takes over, during which hours, and how they see the conversation so far. A bot that traps customers in a loop costs you more than it saves.
5. How will you know it works?
Pick two numbers: the share of chats resolved without staff, and the number of leads captured. Review them monthly and use the unanswered questions to improve the bot.
Scoring yourself
- Four or five yes: a good candidate. Start with your top 20 questions.
- Two or three: begin with a clear FAQ page and WhatsApp quick replies.
- Zero or one: wait until enquiry volume grows.
If you are in the first group, read how we build assistants that answer only from your own documents.
Practical guides on websites, AI, SEO and growth from the Toolmato Agency team.