Answer Customers Faster with AI · Lesson 7 of 7

Capstone: measure your time to first reply

15 minutes to set up, then two minutes a day for two weeks. One number tells you whether this is working.

---

You've built the answer bank, the voice instruction, the routine, the review replies and the after-hours plan. Now you find out whether customers actually hear back faster. Not with a feeling. With one number.

Time to first reply is how long a customer waits between sending a message and getting a real answer from a person. Not the auto-reply. The first human reply, even if it's a question back.

It's the right number for a small business because:

• Customers notice it more than anything else about how you answer.
• It's easy to measure by hand. Your email and texts already have timestamps.
• It's honest. If the drafts are slow to check, or the hand-offs sit for a day, it shows.

You'll measure it for one week the old way, then one week the new way, and compare.

Open the Reply Time Tracker template. It's a simple table: one row per customer message.

For each message, you'll write:

• Date
• Channel (email, text, form, social, review, voicemail)
• Type (A, D or P from Lesson 1, or "other")
• Time received
• Time of first human reply
• Wait (in hours, rounded to the nearest quarter hour is fine)
• Within business hours? (yes or no)
• AI draft used? (yes, edited, or no)
• Fact fixed? (did you have to correct a fact in the draft?)

You don't need every message. If you get a lot, track the first ten each day. Keep a simple paper copy by the phone for voicemails.

If you can, run Week 1 before you switch on the new routine. Answer the way you always have, and just log the times. If you've already started using the routine, that's fine. Look back through last month's sent mail and log 20 or so past messages from the timestamps instead.

At the end of the week, work out two things:

• Typical wait. Line up all the waits from shortest to longest and take the middle one. (This "middle value" is the median. It's better than an average here, because one message that waited three days won't swamp the rest.)
• Same-day share. Out of all the messages that arrived during business hours, how many got a reply the same day?

Write both at the top of the tracker.

Now run everything from this course: the answer bank, the Customer Replies project, the three daily batches, the hand-off list and the auto-reply. Keep logging the same way.

The extra columns matter here:

• AI draft used? tells you how much of the inbox the answer bank covers.
• Fact fixed? tells you how good the answer bank is. Every "yes" is an entry to update.

At the end of Week 2, work out the same two numbers.

Write three short answers at the bottom of the tracker:

• Did the typical wait go down? By how much?
• Did the same-day share go up?
• What slowed you down? Usually it's one of three things: questions not in the answer bank, hand-offs waiting on one person, or drafts piling up between batches.

Then pick one fix for the next two weeks:

• Lots of "no draft" rows? Add the missing questions to the answer bank.
• Lots of "fact fixed" rows? Clean up those answer bank entries and re-upload the file.
• Hand-offs waiting a day or more? Name a backup person for each kind on your Hand-Off Card.
• After-hours messages waiting until noon? Move your first batch earlier, or look at Option 2 in Lesson 6.

Be honest if the number didn't move. That's useful too. It tells you the slow part was never the writing. It may be the sorting, or one busy person, and you now know where to look.

You don't need to track every message forever. Instead:

• Once a month, track one full day and compare the typical wait to your Week 2 number.
• Every month, review the answer bank (Lesson 2), and replace the file in your AI tool.
• Every quarter, reread your voice instruction and hand-off list with the team. Prices, people and policies change.

Use your three templates and tick every box:

• [ ] The Answer Bank has at least your top 10 questions, every fact checked by a person and dated, and a "never promise" note on each.
• [ ] The Reply Voice and Hand-Off Card is saved in your AI tool as a project, Gem or agent, on a business account, and everyone who answers customers knows the hand-off list.
• [ ] Every reply in Week 2 was read and sent by a person.
• [ ] Your review replies are up to date, and none argues or invents.
• [ ] Your after-hours auto-reply, voicemail and Google hours all say the same thing.
• [ ] The Reply Time Tracker has two weeks of entries, the typical wait and same-day share for each week, and one fix chosen.

You started with a busy inbox and answers that lived in people's heads. Now the answers are written down, the AI drafts in your voice, a person checks and sends every reply, the hard ones go straight to the right person, and you have a number that tells you if customers are hearing back sooner.

That written answer bank will keep paying off. It trains new staff. It feeds any AI tool you switch to. And as customers start sending their own AI assistants to ask your questions, it's the clear, written source those assistants will need.

Next steps, when you're ready:

• Connect Your AI to Your Business Tools to have drafts appear right in your inbox, approvals on.
• Write Your Team's AI Rules in an Afternoon to turn the "never paste" list and person-sends rule into a policy everyone signs.
• Save Your Best Instructions: Projects, GPTs and Skills to share your Customer Replies setup across the whole team.
• Lock the Doors: AI Security Basics for a Small Office to make sure the accounts behind all this are locked down.

Previous: After hours: honest auto-replies and AI agents · Back to the course