All templates · Template

🧲 Inbound Lead Qualification and Proposal Follow-up

A B2B consulting firm handles about 150 inbound leads and 30 proposals per month through a shared inbox, manual HubSpot entry and gut-feel scoring, with proposals drafted from an ageing Word template and follow-ups left to memory. The redesign puts an agent layer on intake, enrichment, scoring and drafting so response happens in minutes, while partners keep exclusive control over pricing, scope and every outbound message.

Open this template in the studio →

The as-is: how it works today · 12 steps

The to-be: humans + AI agents · 14 steps, 7 AI-powered

Why this redesign

The firm loses deals for two structural reasons: latency at the top of the funnel and inconsistency at the bottom. Both are caused by humans doing work that carries no judgement. Retyping leads, looking up company data and remembering follow-up dates are pure coordination tasks, so they go fully to agents. Scoring moves from gut feeling to a written rubric the agent applies identically to all 150 leads a month, which makes the criteria debatable and improvable for the first time. Proposal drafting is where the biggest hour saving sits, 90 hours a month, but it is also the highest-risk step, so the agent only assembles from an approved content library and published rate card, and every draft arrives with a visible assumptions list. The two hard constraints shape the whole design: partners keep an unmovable gate before any proposal leaves, and no message reaches a prospect without a human pressing send. The agents own memory, retrieval and first drafts; the partners own price, scope and relationship. Nothing about the firm's reputation is delegated to a model.

Quick wins of this redesign

• Route all lead channels into one HubSpot queue with a timestamp so response latency becomes visible before any AI is introduced.
• Write the fit rubric on one page with partners; even applied manually it removes most of the scoring inconsistency.
• Turn the best three recent proposals into an approved section library and rate card, the prerequisite for reliable drafting.
• Enable agent enrichment of new leads first: it saves 20 minutes each with zero external exposure.

Risks to manage

• A proposal agent drawing on unapproved content could invent a case study or credential, which is why the library guardrail and the partner gate are both mandatory.
• Partners may rubber-stamp AI drafts once volume rises, eroding the very control the constraint was meant to protect.
• Rubric-based scoring can quietly encode a bias toward familiar sectors and starve the pipeline of new market segments.
• Call transcription needs explicit client consent; skipping it creates a privacy and trust problem worse than the time saved.

Suggested pilot plan

Run a six-week pilot on inbound leads only, with one sales coordinator and two partners. Weeks 1-2: consolidate the intake queue, write the fit rubric, deploy the enrichment and scoring agents in inform-only mode alongside current practice. Weeks 3-6: switch on drafted first replies with human send, and pilot the proposal agent on ten proposals using the approved library, keeping the partner gate untouched. Measure four things: median first-response time, agreement between agent score and partner judgement, partner edit effort per proposal, and follow-ups sent on schedule. Success after roughly 60 leads and 10 proposals means median reply under 1 hour, score agreement above 80 percent, proposal review under 45 minutes, and zero unapproved messages sent. If proposal quality does not hold, keep the intake half and revert drafting to human.