What We're Building: See Which Questions Your Visitors Ask Most
Your chatbot knows what visitors care about. We're building analytics to surface those insights - which Q&A gets asked most, what's trending, what's missing.
Every time a visitor asks your chatbot a question, that's data. Right now, DropBot answers the question and moves on. We want to do more with that.
What We're Building
Q&A analytics. Specifically: which questions get asked most, how often, and what questions visitors ask that you haven't added yet.
Think of it as a heat map of what your visitors care about. If "What are your prices?" gets asked 50 times a month but "Where are you located?" gets asked 3 times, that tells you something about what to put front and center on your website.
Why This Matters for Small Businesses
Most small business owners guess what their customers want to know. They build their website based on what they think is important.
Analytics would flip that. Instead of guessing, you'd see exactly what visitors ask. Maybe you thought people cared about your company history. Turns out they just want to know if you're open on Saturdays.
The Technical Challenges (Honestly)
This sounds simple but it's not.
When a visitor types "how much does it cost" and another types "what are your prices" and a third types "pricing info please" - those are all the same question. DropBot uses semantic matching to find the right Q&A, but each match has a different confidence score.
Do we count a low-confidence match the same as a high-confidence one? Probably not. But where do we draw the line? At 70% confidence? 80%? We're still figuring this out.
Then there's the "unanswered questions" feature - showing you what visitors asked that didn't match any Q&A. This is gold for business owners (it tells them what to add). But it requires careful filtering. Spam, gibberish, and off-topic questions would muddy the data.
The Business Challenge
Here's the tricky part: DropBot's free plan allows 100 conversations per month. That's not a lot of data to draw meaningful conclusions from.
If we show a chart that says "Pricing was your top question this month (asked 8 times)," is that statistically meaningful? Not really. But it's still useful directionally.
We need to design the UX so it's helpful at small scale without being misleading. Maybe we skip charts entirely for small volumes and just show a simple ranked list.
What This Unlocks
Once you can see what visitors ask, you can make better decisions. Update your homepage to answer the top question. Add Q&A you're missing. Remove ones nobody asks about.
Your chatbot becomes more than a support tool - it becomes a research tool. That's the direction we're headed.