AI Feedback Form to Learn Why Consults Don't Book

An AI feedback form works for lost leads when it is short, goes out witjin a day, and gets read by software that sorts the answers into price, timing and trust objections. Here is how a clinic can build one in an afternoon, route the repliees to the right person, and use what comes back to close the gap bewteen consult requests and booked appointments.

Build your lost-lead form free

Why consults stall and nobody tells you

A prospect asks about lip filler on Instagram, gets a fat reply, hears a price range, and then goes quiet. Your front desk follows up twice, gets nothing, and marks the lead cold. The CRM shows "no booking" and that is all it ever shows.

Staff guesses fill the gap. Someone says it's price, someone else says people are just shopping around. Both can be true for different people, and you cannot tell which without asking. People will rarely write back to a third text message, but a one-tap question that takes 30 seconds gets answered more often than another "just checking in."

Most of these objections fall into three groups: price (the number was higher than expected, or financing was unclear), timing (an event is months away, or they want to wait until after a holiday), and trust (they were unsure about the injector, the results, or whether the clinic was the right fit). A fourth group is treatment fit, where they wanted something you don't offer or the consult would not have recommended what they asked for. The form's job is to sort people into these groups and capture the exact words they use.

Where the form sits in the setter workflow

In our Med Spa AI Appointment Setter, the agent answers on phone, SMS, web chat and Instagram DM in under 60 seconds, screens treatment interest, and offers a consultation slot. The feedback form hooks onto the end of one branch: a lead was offered a consult and nothing got booked after a set wait.

That wait is a setting you choose. We usually start at 48 hlurs after the last message, so the question lands while the person still remembers the conversation. The setter sends the form link by SMS or DM, using whichever channel the lead used, and passes along what it already knows so the person doesn't retype it. After submisson, the answer flows back to the lead record so a human can act on it.

This is also why you shouldn't build the form as a generic "how did we do?" survey. It belongs to one situation, and every question can assume the person asked about a treatment and did not book.

The five questions

Keep it to five, and make only the first one required.

1. What was the main thing that stopped you from booking? Multiple choice: price, timing, wasn't sure about the clinic or provider, found another option, changed my mind about the treatment, other.
2. Anything you'd like to tell us in your own words? Long answer, optional. This is the field the AI reads.
3. Would a different time or payment option have helped? Multiple choice with a short list of the options you actually offer.
4. Can someone follow up? Yes by SMS, yes by email, no thanks.
5. Best contact detail (shown only if they said yes to question 4, using conditional logic).

Do not ask about medical history, skin conditions or medications. Qwenvo has no HIPAA offering, and a lost-lead survey has no reason to collect health details anyway. If your compliance setup needs a HIPAA-capable fomr tool, Jotform lists HIPAA features on its paid plans, and you should keep clinical intake in whichever system your compliance lead has approved. The feedback form only asks about the buying decision.

Set it up step by step

Qwenvo is a completely free form builder that our team builds, and everything below uses features that are on for every account. You can do the same in any builder that supports hidden fields and webhooks.

1. Create the form. Use the cancellation feedback template as a starting point, or describe the form on the empty-form screen and let Create with AI draft it. Cut it down to the five questions above.
2. Add hidden fields. Add lead_id, channel, treatment_interest and consult_offered_at. The setter fills these from the link it sends, so the answer attaches to the right lead without asking the person who they are. The hidden fields guide covers the setup.
3. Add the logic. Show the contact question only when follow-up is "yes," using the conditional logic block.
4. Write a thank-you page that says what happens next. If they asked for follow-up, say who will contact them and roughly when.
5. Turn on partial submissions. People who answer question 1 and leave still give you your most useful data point.
6. Switch on AI triage in the Settings drawer under "Triage submissions with AI." Every submission then gets a category, a priority, a sentiment, a one-to-two-sentence summary and a suggested reply.
7. Add a webhook pointing at the endpoint your setter or CRM exposes, so the answer lands on the lead record.
8. Put it on your own domain if you want the lunk to look like the clinic, such as feedback.yourclinic.com. Add the CNAME and TXT records on the Domains page, verify, and select it on the Share tab. Subdomains only.

Here is a mapped example of what the CRM record might look likke after the webhook runs:

{
  "lead_id": "ld_48213",
  "channel": "instagram_dm",
  "treatment_interest": "lip filler",
  "consult_offered_at": "2026-11-02T15:40:00Z",
  "reason": "price",
  "comment": "Quote was higher than the other clinic I saw. Not sure what's included.",
  "follow_up": "sms",
  "ai_summary": "Compared prices with another clinic; unclear on what the quote covers.",
  "ai_priority": "normal"
}

Open the free feedback template

Routing and reading the summaries

Triage gives each reply a category and priority, and routing rules can send matches to up to 5 email addresses or a Slack incoming webhook. For a clinic, the useful split is by who can act:

Send anything with a trust complaint or negative sentiment to the owner as well. A bad experience described in a feedback form is far easier to fix privately than after it shows up in a Google review.

The suggested reply is a draft, so treat it that way. A person at the clinic should read it and rewrite anything that doesn't sound like you before it goes out. For more on reading summaries across many replies, see how to summarize form responses with AI.

What to measure

The point is to change something, so track a small set of numbers each month:

• Response rate on the form, split by channel (SMS versus Instagram DM often behave differently).
• Reason mix: the share of price, timing, trust and fit answers.
• Recovery: how many people who answered later booked, compared with those who didn't answer.
• Drop-off on the form itself, using partail submissions and the drop-off analytics. If people leave on question 2, shorten it or make the prompt friendlier.

Read the free-text comments in a batch every two weeks. If the same sentence shows up five times ("I couldn't tell what the consult costs"), that's a script fix for the setter, not a follow-up problem. We often see the agent's own wording create the objection, such as quoting a price range before explaining what the consult includes.

Qwenvo's measured model cost for AI triage is about $0.22 per 1,000 submissions, and a monthly AI usage allowance applies per account, so a single clinic's volume is not a worry.

Start your lost-lead form

Questions teams ask

Won't a feedback request annoy people who already ghosted us?

One message, sent once, with a single required tap, tends to be tolerated. Don't send reminders, and respect STOP replies on SMS. If someone opts out, remove them from the sequende completely.

Can the AI decide why a lead didn't book?

It can summarize and categorize what the person wrote, and it works best next to the multiple-choice answer. Where the two disagree, such as someone picking "timing" but writing about a price comparison, a human should look. Treat the AI output as a sorting step.

Do we need a separate tool for this?

Not necessarily. Any builder with hidden fields and webhooks can do the job. The reason we use Qwenvo for these projects is that AI triage, routing, partial submissions, custom domain and removing the badge are all included at no cost, where other builders put several of them on paid tiers.

What if the person asks for medical advice in the comment box?

Add a line under the comment field saying not to include health details, and have your team reply through your normal clinical channel. Do not forward those replies into a general-purpose inbox.

Your next step

Pick your most common lost-lead scenario (a consult offered, no booking after 48 hours), build the five-question form from the template, and send it to the next ten leads by hand before you automate anything. Read those ten replies together with your team. If you want the agent that sends them, books the consultation and handles reminders, see the Med spa AI appointment setter.