Disqualification recommender

A disqualification recommender argues the negative case on an open opportunity and attaches the evidence. It does not change stage, and it does not write Closed Lost.

The output is a recommended no-bid or disqualify. Each reason must point at a CRM field, a call or email quote, or a closed-lost pattern you already recorded. The sales manager confirms, overrides, or sends the AE back for discovery. Until that confirmation, Salesforce stays exactly where the AE left it.

Salesforce, Gong, 6sense, and Clari sit in this class. Salesforce is the opportunity of record. Gong holds the conversation quotes you will cite. 6sense-style fit and intent sit on the account. Clari-style inspection sits on the pipeline. None of them should close the deal for you.

Treat this as qualify-stage hygiene, not late-stage rescue. Deal risk scoring watches committed pipeline later. This page is for deals that should never have been worked as if they were qualified.

Score four questions. A recommendation that cannot name the evidence for each question is not ready for a manager.

ICP. Does this account match the closed-won profile, not the whole CRM? Compare industry, size, product fit, and use case with accounts you actually won. If lookalikes were built from every open opportunity, the fit signal is polluted. Rebuild the reference set from closed-won, the same way ICP lookalike account discovery should. A 6sense-style fit score is an input. It is not a no-bid.

Budget authority. Has anyone who can spend engaged, or have they said there is no budget this cycle? Quote the call. "We have not asked" is incomplete. "There is no line this year" is a reason. Do not invent an economic buyer because the title looks senior.

Timeline. Is there a date tied to a business event, or an empty close date with "sometime next year" on the recording? Long enterprise processes look dormant while they are still moving. Silence is not a stall if the stated process is a nine-month security review. An empty date is a gap.

Known lost-deal patterns. Match the file against reasons you already lose: incumbent mandate, missing product capability, security review failure in that industry, education-only buyers, no budget. Pull those reasons from a win/loss reason classifier on closed-lost, not from a guessed taxonomy. A pattern match still needs a cite on this deal.

Put the cites in the recommendation itself:

• CRM field, value, and last-modified date
• Call or email quote with speaker, date, and timestamp
• Closed-lost reason code and the similar accounts, without treating those accounts as proof this one will lose

A predictive SQL conversion score ranks inbound for follow-up. It does not replace this review. A high score is not a qualified opportunity, and a low score is not a confirmed no-bid.

Blank Metrics, Economic Buyer, Decision Criteria, Decision Process, Identify Pain, and Champion fields mean you have not captured them. They do not mean the buyer lacks them.

MEDDIC auto-extraction can suggest fields from Gong recordings and leave blanks blank. A recommender that treats those blanks as "no" will kill deals that are early, deals in new segments, and deals where the AE has not run the next meeting yet.

Split every flag:

• Evidence of no. They said there is no budget. They said the incumbent is mandated. They asked for a capability you do not ship. Quote it.
• Evidence missing. Economic Buyer is empty. No champion named. Close date is blank. Champion last appeared on a call six weeks ago and you have no new recording. Ask the AE to fill the gap or schedule the meeting. Do not recommend disqualify on a blank.

Killing a strategic logo is the expensive version of the same error. Named accounts, board relationships, and land-and-expand bets often sit outside the mid-market closed-won cluster. The model will look right on ICP and still be commercially wrong. Route those logos to the manager with the cites intact. Do not hide the mismatch. Do not auto-close to make the forecast look clean.

The following is a made-up weekly review, not a measured program and not reported results.

An AE has Harbor Medical, a regional hospital group, in Stage 2. The VP named it on the account list. The recommender returns a no-bid:

• ICP: closed-won cluster is manufacturers and SaaS firms above a headcount the model learned from wins. Harbor is healthcare. 6sense-style fit is low.
• Budget authority: no economic buyer on any Gong call. The only contact is a clinical operations manager who said, "I am gathering options for later."
• Timeline: close date empty. No event named.
• Lost-deal pattern: other healthcare logos in CRM closed-lost with reason "security review stalled."

If Salesforce writes Closed Lost overnight, you have killed a named account on blanks and a pattern from other hospitals. The AE stops logging. The VP finds out in forecast.

The manager keeps the opportunity open. They do not rewrite the cites. They mark the ICP mismatch as expected for a named logo, convert the empty Economic Buyer from a "no" into a required CIO meeting by a date, and leave the security-review pattern on the file as a discovery question, not a verdict.

Two weeks later the CIO still has not joined. A new Gong quote: "We have a freeze through year end and we already renewed the incumbent." That is evidence of no. The manager confirms disqualify, writes the reason, and only then does the opportunity move. The recommendation did not move it.

Write the play before the first recommendation hits the queue.

Who owns the decision. The AE's manager, not the model, not sales ops, not a workflow. Ops can build the queue. Ops cannot confirm a no-bid.

What the recommendation must contain. The four scores, each with a cite or an explicit "missing." The proposed disqualify or closed-lost reason. Whether the account is on a named or strategic list. A place for the AE to rebut with information the system lacks.

What it must not do. Change Salesforce stage. Write Closed Lost. Clear the next step. Remove the deal from Clari-style inspection. Send the customer a breakup email.

Start with newly created opportunities and early stages, not the entire open pipeline. Prioritize deals with a Gong quote of no budget or no authority, accounts far from the closed-won ICP that are not on a named list, and files that match a repeated closed-lost reason with a cite on this deal.

Measure recommendations the manager confirmed, overrides with a written reason, recommendations sent back as incomplete, and any opportunity the system closed without a human. If that last count is not zero, turn the write-back off.

Salesforce, Gong, 6sense, and Clari do the same job from different angles: hold the record, the quotes, the fit, and the inspection. Do not rank them. Do not expect any of them to know a CIO conversation that never happened.

The recommender is doing its job when a manager can open the flag, read the cites, keep a named logo in play, and confirm a real no-bid without the CRM closing itself.