How to Use Fallback Reports to Find Your Missing Q&A
Your chatbot is already telling you what's missing. Fallback reports show the exact questions visitors ask that you haven't answered yet.
Every chatbot has blind spots. Questions visitors ask that you never thought to add. The good news: your chatbot is tracking every single one of them.
What's a Fallback?
When someone asks your chatbot a question and it can't find a matching Q&A pair, that's a fallback. The chatbot still tries to help using the context it has, but it flags the interaction as unmatched.
These aren't failures. They're signals.
Where to Find Fallback Reports
In your DropBot dashboard, go to Fallback Submissions. You'll see a list of every question that didn't match one of your Q&A pairs, along with when it was asked and how many times.
Sort by frequency. The question that appears 12 times is more urgent than the one that appeared once.
What to Do With Them
Step 1: Look for patterns. Three visitors asking "do you do house calls" and two asking "do you come to my home" are the same question. Group them mentally.
Step 2: Prioritize by frequency. If 15 people asked about parking this month, that's your next Q&A pair. Don't start with the question that came up once.
Step 3: Add the Q&A pair. Write the answer in your voice, the way you'd tell a customer standing in front of you. Keep it short and specific.
Step 4: Check back in a week. That question should disappear from fallbacks. If it doesn't, your Q&A wording might not match how visitors phrase it. Adjust the question text to match how real people ask.
The 80/20 of Chatbot Accuracy
Most chatbots only need 10-15 Q&A pairs to handle 80% of questions. But that remaining 20% is where you lose people. Fallback reports help you chip away at that gap, one real question at a time.
You don't have to guess what's missing. Your visitors are telling you every day. You just have to look.
A Real Example
One pet groomer added 8 Q&A pairs at launch. After two weeks, fallback reports showed that "do you sedate dogs" was asked 9 times. She added it. The next most common fallback was about nail trimming pricing - she added that too.
Within a month, her fallback rate dropped from 35% to under 10%. She didn't need to brainstorm what visitors wanted. They told her.
Check your fallback reports this week. There's at least one Q&A pair in there you didn't know you needed.