Scoring Model That Sales Ignores
A company's lead score is a single number 0–100. Demographic points were set years ago by a departed admin: 'Title contains Manager' = +20, 'Country = US' = +10, plus behavioral points for email opens (+2 each, uncapped) and form fills (+10). Sales says the scores are 'garbage' and routes by gut. Analysis shows 70% of high scores come from repeated email opens by a small set of contacts, and that job-title matching is case-sensitive and misses 'Sr. Manager' and 'Director'. Routing assigns leads round-robin to 12 reps regardless of territory or segment. MQL threshold is 50. Only 8% of MQLs convert to opportunity.
1. Diagnose why the scoring model fails and propose a corrected design.
Read the worked answer
Two distinct failures: a broken behavioral signal and a broken demographic signal, compounded by a threshold nobody validated.
Behavioral: uncapped open points mean a single engaged contact can inflate indefinitely; opens are also a weak, privacy-degraded signal (image blocking, Apple MPP). 70% of high scores from a few contacts is the tell. Fix: cap behavioral points per category and per period, decay or expire points, and weight higher-intent actions (demo request, pricing page, content download) far above opens. Consider dropping opens or weighting them near zero.
Demographic: case-sensitive title matching misses real titles. Replace substring matching with normalized, case-insensitive rules or a picklist mapped from the CRM, and include seniority bands (Director, VP, C-level) not just 'Manager'.
Threshold: 8% MQL-to-opportunity means the bar is miscalibrated. Rebuild the threshold empirically - look at the score distribution of leads that actually became opportunities and set the cutoff where precision is acceptable to sales.
Governance: document the model, assign an owner, and review quarterly. A departed admin's untested rules running unowned is the root cause.
2. How do you fix routing so sales trusts the handoff?
Read the worked answer
Round-robin across 12 reps ignoring territory and segment guarantees misassignment and erodes trust regardless of score quality.
- Route on fit first. Territory/region and segment (enterprise vs SMB) should determine the eligible rep pool; then distribute within the pool (round-robin or capacity-weighted).
- Respect ownership. If an account already has an owner in the CRM, route to that owner rather than round-robin, to avoid duplicate outreach.
- Tie routing to score. Only route at/above the validated MQL threshold; below-threshold leads stay in nurture. Routing everything trains reps to ignore the queue.
- Handle exceptions. Define fallback (unmatched territory → named queue), SLA for acceptance, and reassignment rules for no-touch within X hours.
- Close the feedback loop. Send rep disposition (accepted/rejected, reason) back into scoring so rejected-but-high-score leads inform recalibration.
The judgement: routing quality and scoring quality are coupled. Fixing one without the other won't restore trust.