Same question, fortieth time today
What are your opening hours? Do you deliver to my area? How long does delivery take? Can I change my booking? Do you do a trade discount? Where do I send my returns? What is the difference between the standard and the premium option?
Your team answers these by email, phone, live chat, Facebook and Instagram messages, all day. Each answer takes a minute or two. Each one is polite and correct. And between them they crowd out the conversations that need a person: the upset customer, the complex quote, the order that went wrong.
You have probably added an FAQ page. People still ask.
Why the answers exist but people still ask
It is easy to conclude that customers do not read. The more useful explanation is that answers are in the wrong place, in the wrong form, at the wrong moment. The FAQ page is three clicks away from the product page. The delivery information is in a PDF. The answer depends on the customer's postcode, which a static page cannot know.
And asking is simply quicker than searching. If your chat box or email address is visible, people will use it, which is fair enough. The fix is not to hide the contact options. It is to make the quick answer arrive as fast as the question.
There is also a knowledge problem on your side. The real list of repeated questions is rarely written down. Your team knows it by feel, but nobody has counted, so the FAQ page often answers the questions you expected rather than the ones people ask.
What the repetition costs
- Staff time on routine answers that could go on complex or valuable conversations.
- Customers waiting in a queue behind simple questions.
- Slightly different answers from different people, including occasional wrong ones.
- Questions outside office hours waiting until the next day, when some customers have already bought elsewhere.
- Support work that grows at the same pace as the business.
How SpiderHunts takes the repeat questions off your team
- Export a sample of your real conversations from email, live chat, social inboxes and helpdesk, and cluster them by question, so we see the actual top questions and how they are phrased.
- Check that each common question has one correct, approved answer. Where answers are missing, out of date or contradictory, we flag them for you to decide.
- Build an assistant using a large language model (such as Anthropic Claude or OpenAI models) with retrieval from your approved content only, so it answers from your website, policies and product data rather than general knowledge.
- Connect live data where answers depend on it: delivery areas by postcode, stock levels, opening times, booking availability.
- Put it on the channels customers use: a website chat widget, WhatsApp, Facebook Messenger, and as suggested replies in your email or helpdesk.
- Set clear limits: anything outside the known topics, anything involving a complaint, payment or account change, or any customer who asks for a person goes straight to your team with the conversation attached.
- Review conversations weekly at first, correct wrong or weak answers at the source content, and add new questions as they appear.
| Type of question | Who handles it |
|---|---|
| Opening hours, delivery areas, policies | Assistant, from your content |
| Product comparisons and specs | Assistant, from product data |
| Order or booking specific | Assistant with system lookup, or team |
| Complaints and refunds | Your team, with context passed over |
| Anything unclear or new | Your team, and added to review |
Sometimes the right fix is not a chatbot at all. A clearer delivery page, a postcode checker or a better confirmation email can remove a question entirely. When that is the case, we say so and build that instead.
A support day with the noise taken out
Customers get routine answers straight away, at any hour, on the channel they chose. Your team's queue is shorter and made of conversations that need them. Answers are consistent, because they come from one approved source that you control.
The conversation logs also become a live picture of what customers are confused about, which is useful to marketing, product and operations, not only support. If a question suddenly spikes, something has changed, and you find out quickly.
Choosing a chatbot platform becomes a simpler decision once this work is done, because you know exactly which questions it must handle and which systems it must reach. We can build on a platform you already use, or build a custom assistant where the platform cannot reach your data.
Is your team stuck on repeat?
- Your team can list the top questions from memory.
- You have an FAQ page and people still ask the same things.
- Routine questions arrive faster than your team can clear them.
- Out-of-hours questions wait until the next working day.
- Different staff sometimes give different answers to the same question.