Book Your Call
One call. Clear direction. Faster execution.
How a home services franchise turned its worst-rated channel into its best in 90 days.
Their first AI chatbot frustrated customers so badly that leadership wanted it switched off. Skillfi audited every failed conversation, chose the AI CRM platform that could do the job properly and rebuilt the assistant around what customers actually ask. It's how our AI CRM service works: the right tool first, then a setup that sticks.
A chatbot customers had learned to avoid.
Eighteen months earlier the franchise launched an early AI chatbot across its website. It gave generic or wrong answers, looped customers back to the start and couldn't book a job, give a price range or check on a visit.
"Agent" and "human" were the most-typed words in chat. One in four conversations ended with the customer leaving mid-chat, and many of them called a competitor instead. Chat had the lowest satisfaction of any channel, and the leadership team was ready to switch it off. The Chief Customer Officer had one quarter to prove AI could work.
Our customers had learned that typing into our chat was a waste of time.
Chief Customer Officer, client
Find out why it failed, choose the right platform, then rebuild around real questions.
We didn't patch the old bot. We read what customers were actually asking, scored the platforms that could answer from the company's own data and book real jobs, and only then rebuilt. Follow-ups and review requests run on marketing automation inside the same platform.
11,000 conversations analysed
We mapped the top 50 customer intents and every drop-off point, and found 63% of failed chats were about booking, pricing or job status.
4 platforms scored on what mattered
Options scored on AI answer quality, grounding in the company's own data, booking integration, multi-location support and reporting. One all-in-one AI CRM platform came out clearly ahead.
An assistant that books, prices and hands off
Answers from a cleaned knowledge base and local pricing guides, bookings straight into each location's calendar, and a one-step handoff to a person with the full conversation.
12 weeks, week by week
From a chatbot leadership wanted to switch off to the channel customers rate highest. Here's how the engagement unfolded.
- Week 1–211,000 chats
CX audit
Every chat transcript reviewed, customer intents ranked and the reasons for each failed conversation mapped.
- Week 34 platforms scored
Platform selection
Shortlisted platforms tested against the top 50 real customer questions and scored side by side.
- Week 4–6Top 50 intents
Assistant rebuilt
Knowledge base cleaned, local pricing guides added and booking connected to every location's calendar.
- Week 7Human in one step
Handoff & follow-up
One-step handoff to a person, frustration detection, SMS follow-up on unfinished chats and the CXO dashboard.
- Week 8–9Pilot live
Five-location pilot
Live in five locations with daily transcript reviews and weekly tuning of answers and handoff rules.
- Week 10–1242 of 42 live
All 42 locations live
Rolled out across the network, with the support team trained on taking over conversations smoothly.
From dead ends to booked jobs
Four layers that turned a chatbot customers avoided into the fastest way to get help.
Right answers, first time
- Answers from verified content only
- Local pricing ranges
- Job status lookup
Jobs booked in chat
- Live slots for every location
- Text confirmations
- Reschedule without a call
A person in one step
- Frustration detected early
- Full conversation passed over
- SMS follow-up on unfinished chats
CX the board can see
- Resolution & CSAT by location
- Top handoff reasons
- Bookings & revenue from chat
Ninety days after full rollout
Chats resolved without a person
And when customers do need a person, they reach one in a single step.
Ninety days ago my leadership team wanted the bot switched off. Now chat is our best-rated channel and our biggest source of new bookings.
Chief Customer Officer, clientIs your chatbot helping customers or driving them away?
We'll audit your conversations, choose the AI CRM that can answer from your own data and rebuild the assistant your customers actually need, with follow-ups running on marketing automation. Want your team confident working alongside AI? Add a hands-on Claude workshop.
- A full chat & CX audit
- A scored platform recommendation
- An AI assistant grounded in your data
- Human handoff & a CXO dashboard
30 minutes. No obligation.
Related Case Studies