What to Do in the Seven Days After Launch
Last updated:
Read everything
For seven days, read every conversation. It is the single most valuable thing you can do and nobody does it after week two.
You will learn more about what your customers want than from any survey you have run.
Day by day
- Day 1: confirm it appears everywhere it should; ask it five questions yourself
- Days 2–3: read every conversation; note wrong or missing answers
- Day 4: fix the content behind the worst three; recrawl
- Days 5–6: check whether those questions now answer correctly
- Day 7: review the four metrics and decide what to fix next week
What to look for
- Questions it could not answer at all
- Answers that were technically right and unhelpful
- Visitors who asked for a person and how quickly that happened
- Anything where it sounded wrong for your business
- Questions you did not expect, which are the most interesting
The unexpected questions are the valuable ones. They tell you what your customers are actually uncertain about, which is rarely what you assumed.
Fix content, not the bot
Almost every wrong answer traces back to a page that is missing, unclear or out of date. Fixing the page fixes the answer and helps everyone else who visits.
Resist the urge to work around it in configuration. The content fix is the durable one.
Tell your team
Whoever handles enquiries should know it exists, what it can answer and how handovers reach them. Discovering it from a customer is a poor introduction.
Share the first week's unanswered-question list with them too — they will recognise most of it immediately.
Frequently asked questions
What if the first week goes badly?
Should we announce it?
How many conversations should we expect?
When can we stop reading everything?
About to put a chatbot live?
Block an hour a day for the first week. It is the highest-return hour in the whole project.