Conversational Lead Qualification · US Franchise Brokerage
A conversational bot that let franchise leads qualify and book themselves, with an 80% show rate on bot-scheduled intro calls against a 60% historical baseline.
- Client
- Confidential (US franchise brokerage)
- Sector
- Franchising / Lead generation
- Role
- Service & Conversational Designer, from discovery to build and measurement
Overview
A US franchise brokerage bought leads from a third-party marketplace and qualified them by phone. Historically it took more than 200 of these leads to close one deal, and a large share of the broker's calendar went to intro calls that never happened. Together with the calling team, the broker and the marketing team, I designed a conversational experience that lets prospective franchisees qualify themselves, get routed to the right next step and book an intro call on their own time.
Challenge
- Lead quality was uneven and contact was expensive: callers spent most of their time chasing people who were unreachable, not ready or not financially eligible.
- In 2020, 83 of 208 intro calls booked by callers were no-shows, which meant hours of senior broker time lost every month.
- Prospects are cautious. They are being asked to share financial information with a stranger, and they receive unsolicited business pitches all the time. Trust had to be earned before any question was asked.
Approach
- Discovery. Facilitated workshops with the broker and the marketing team to map the current qualification process, using a CSD matrix (certainties, assumptions, doubts) to separate what the team knew from what it only believed, and ran semi-structured interviews with prospective franchisees to understand motivations, fears and decision timing.
- Intent research. Mapped real franchise search queries by funnel stage, knowledge level and sentiment to learn the vocabulary and the questions people bring to the conversation.
- Conversation design. Created "Franbot", a named search assistant that introduces the broker, and mapped the full sequence: identity check, broker credentials and testimonials, how the service works, interests, search history, location, liquid capital, investment timing and scheduling. Each answer routes to one of three outcomes: book an intro call, book a later check-in with the franchise coordinator, or leave with curated resources. At any point, the prospect can switch to a human by email or phone.
- Explaining before asking. The hardest moments were money and trust, so the script teaches before it asks. It clears up the question people rarely voice (the broker is paid by the franchisors, never by the candidate, like an executive recruiter) and frames liquidity with a familiar model: franchises need about 30% liquid capital, like a down payment on a home, and the rest can be financed.
- Tone of voice. User feedback showed some prospects felt the bot did not respect their time. The dialogue was rewritten around that tension: explain up front why each question matters and that answering is the fastest route to a useful conversation. Disqualification was written to keep the door open ("just because it's not the right time doesn't mean it's the wrong direction").
- Pre-bot journey. Data showed the real drop-off happened before the conversation started, and emails were being caught by spam filters. I redesigned the entry as an SMS-first journey that introduces the broker through a credible public profile before inviting people into the bot.
- Validation. Planned and ran an unmoderated end-to-end usability test with recruited participants matching the real profile, measuring clarity, ease, comfort in sharing financial data, understanding of the broker's role and likelihood to attend the call.
- Measurement. Built the funnel tracking (link sent, clicked, started, finished, scheduled, showed) and compared bot-scheduled leads against the caller-only baseline in periodic status reports.
Solutions & Deliverables
Rule-based qualifying bot, designed and built by me on Landbot, with scheduling built in; conversation flows and dialogue scripts; disqualification paths that send non-eligible leads to self-serve resources with no manual follow-up; human escalation path; SMS and email entry sequences; usability test plan and findings; funnel dashboard and quarterly status reports; a family of bot variants built from the same template (marketplace leads, ads, website quiz and a general qualifier), so a new lead source could be served at low cost.
Results & Impact
- 80%
- show rate on bot-scheduled intro calls, against a 60% caller-only baseline
- 2x
- interested, qualified leads with bot plus caller
- 50%
- fewer intro calls to get there
The bot booked better calls with less of the broker's time, and the data it produced changed how the brokerage thought about lead sources.
Details
In the Q4 pilot, leads who scheduled through the bot showed up to 80% of intro calls, against a 60% show rate for caller-only scheduling in 2020. Where the bot was involved at any point, the show rate was 71%. Applied to the 2020 volume, the bot-only rate would have meant about 42 fewer no-shows, roughly 21 hours of the broker's intro-call time given back.
Bot plus caller produced twice as many interested, qualified leads with 50% fewer intro calls than a caller working alone. Leads qualified by the bot kept advancing in the sales pipeline, which showed the bot was filtering for quality.
Lead sources behaved very differently: about 10% of marketplace leads engaged with the bot, against about 40% of LinkedIn leads. That shifted the recommendation from buying more marketplace leads to building a proper marketing pipeline, with the bot as its qualification layer. The funnel also made the next bottleneck visible: most leads never clicked into the conversation, so the following phase focused on the pre-bot experience.
The 2021 bot was rule-based. Rebuilt today, I would keep the same qualification logic and escalation rules as guardrails and let an LLM agent handle the open questions prospects ask along the way (fees, risk, how the broker is paid), with a prompt grounded in the brokerage's own material, a clear handoff to a human when confidence is low, and evaluation on real transcripts before rollout.
The pilot sample was small (17 intro calls booked through the bot), so these figures are an early signal, not a benchmark.
Skills
Conversational Design, Service Design, Stakeholder Workshops, User Interviews, Usability Testing, UX Writing, Funnel Analytics.