Customer Follow-Up Skill
Automated post-service follow-up via any messaging channel. Sends a "How was your experience?" message after a job is completed. Positive responses get a thank-you and review request. Negative responses get escalated to the owner immediately.
The Problem
Most small service businesses never follow up after a job. The ones that do, do it manually and inconsistently. Result: no reviews (competitors outrank you on Google), unhappy customers churn silently, and problems that could be fixed quickly fester into bad Yelp reviews.
Dedicated customer feedback tools (Podium, Birdeye, ServiceTitan) cost $200-500/month. Most small businesses can't justify that.
How It Works
- Service is completed — someone triggers the follow-up (manually, on a schedule, or via a trigger file)
- Customer receives a message on their preferred channel: "How was your experience with [service]?"
- Customer replies naturally
- AI analyzes the sentiment: - Positive: Thank-you response + Google Review link - Negative: Empathetic response to customer + immediate alert to owner with details and urgency - Neutral: Gentle follow-up asking for specifics
Setup
1. Enable Inbound Messaging
The customer's reply channel must be enabled inbound:
- SMS (most common for service businesses): System > Server Settings > SMS
- WhatsApp: System > Server Settings > Messaging > WhatsApp
- Email: System > Server Settings > Email
2. Create Customer Contact Profiles
For each customer (or create dynamically via the API):
- Users & Privileges > Subscribers
- Set their phone number or email
- Inbound Enabled: Yes
3. Customize the Skill
Edit SKILL.md to set your business details:
- Replace
[business name]with your company name - Replace
[owner name]with the person who handles complaints - Replace the Google Review URL placeholder with your actual link
- Get your link: Google Business Profile > Share > "Ask for reviews"
4. Create the Follow-Up Job
A simple job that sends the initial follow-up message:
name: customer-followup-send
description: Send post-service follow-up to customer
tasks:
- name: followup
tool: message
properties:
subject: "How was your service?"
body: "Hi! We hope everything went well with your recent [service type]. We'd love to hear your feedback — just reply to this message. Your response helps us improve!"
contactNames: "${customerName}"
Trigger this job: - Manually via OneShot after completing a service - On a schedule — daily at 5pm, pulling customers serviced that day - Via trigger file — when a "job complete" file is dropped by your dispatch software - Via the AI assistant — "Send a follow-up to John Smith"
5. Install the Skill
Pre-installed at INTOUCH_HOME/skills/customer-followup/SKILL.md.
Usage Flow
Step 1 — Trigger sends initial message:
SMS to customer: "Hi! We hope everything went well with your
recent AC repair. We'd love to hear your feedback — just reply
to this message."
Step 2 — Customer replies:
Customer: "Tech was great, fixed it fast. Only issue was he
tracked mud on the carpet."
Step 3 — AI processes via @customer-followup:
FEEDBACK: NEGATIVE — REQUIRES ATTENTION
Customer: John Smith
Sentiment: Mixed (positive on service, negative on cleanliness)
Issue: Technician tracked mud on carpet
Urgency: Medium
RESPONSE TO CUSTOMER:
Thank you for the feedback! Glad we got your AC fixed quickly.
I'm sorry about the carpet — I've flagged this for Mike who
will reach out to you. We'll make it right.
ALERT TO OWNER:
John Smith (AC repair, today): Positive on speed and tech
skill, but tech tracked mud on the carpet. Medium urgency.
Customer is not angry but expects resolution.
Customization
Edit SKILL.md to change:
- Review platform — swap Google for Yelp, Facebook, Angi, HomeAdvisor
- Response tone — more formal for professional services, more casual for retail
- Urgency criteria — adjust what qualifies as Critical vs High vs Medium
- Auto-responses — change what the customer receives
- Escalation rules — route different urgency levels to different people
- Languages — Claude handles any language; adjust response templates for your market
Cost
- Ollama: Free (works well for sentiment analysis)
- Claude/GPT: ~$0.005 per response (text only)
- 100 customers/month: $0.50/month with Claude
Compare to Podium ($250/month) or Birdeye ($300/month).
Advanced: Automated Review Tracking
Pair with a YAML job that: 1. Runs daily 2. SQL queries customer feedback from a database 3. Claude generates a weekly satisfaction report 4. Posts to the owner via Slack or email
This gives you trend tracking — "satisfaction dropped 15% this month, mostly around wait times" — without a dedicated analytics platform.