objection-library

Objection Library

What this does

Reads all of your sales conversations and returns a ranked library of the objections buyers actually raise - each one in the buyer's own language, counted by frequency, and tagged to the stage of the funnel where it shows up. It also records how reps currently respond and whether that response tends to move the deal forward.

The point is coverage. Teams usually build objection handling from the handful of objections that stung most recently. This gives you the full set, weighted by how common each one really is.

Why this is a Fable-recommended job

Objections are scattered across many hours of talk. To rank them honestly you have to hear all of them and count across the whole set, which means holding every transcript at once. Claude Fable 5 can take the full call library in a single run and surface patterns no one person sitting through calls would tally.

If Fable is not available or the budget is spent: batch it. Run 8-10 calls at a time, produce a partial objection list per batch, then merge and re-rank the partial lists in a final pass.

Evidence to gather first

• Sales and demo call transcripts across the funnel (discovery through negotiation)
• Objection notes from reps or deal desk
• Support and onboarding calls, if pre-sale concerns surface there
• Lost-deal notes

Read .agents/product-marketing-context.md first if it exists, so objections can be mapped against your positioning and competitors.

Process

1. Confirm the call set and window. Report how many calls are included.
2. Read every call. Pull each moment where a buyer raises a concern, hesitation, or blocker - stated as a question, a worry, or a flat no.
3. Normalize duplicates: the same underlying objection phrased three ways becomes one entry, with the variant phrasings kept as examples.
4. Count how many calls each objection appears in and tag the funnel stage.
5. Capture the current rep response for each and note where a response seems to land or fall flat.
6. Produce the library.

Output format

• Summary - calls analyzed, window, top objections by frequency and by funnel stage.
• Objection table - objection (in buyer language) | funnel stage | number of calls | representative quote with source | current response | how well it seems to work.
• Gaps - objections with no consistent response, or where the current response is weak.
• Recommended handling - a suggested response for the top objections, grounded in what already works in the winning calls.

The evidence standard

Every objection entry must quote at least one real buyer moment and cite its source (call name, date, or line). Use the buyer's words, not a tidied-up paraphrase. If frequency cannot be counted reliably from the material, say so rather than inventing a number. Recommended responses should be built from responses that appear to work in your own calls before any that are added from general practice - and anything added from general practice must be labeled as such.