---
title: "The scoring model wasn't broken — 90,000 contacts had no industry"
type: case-study
evidence_type: proof
category: "Data Quality & Enrichment"
publisher: "LeanScale"
date_modified: 2026-08-08
word_count: 371
topics: ["revenue-operations"]
canonical_url: https://knowledge.leanscale.team/customers/root-causing-crm-data-gaps-behind-bad-scoring/
source: "LeanScale Knowledge Hub — https://knowledge.leanscale.team"
license: "Free to quote and cite with attribution to LeanScale."
---

# The scoring model wasn't broken — 90,000 contacts had no industry

**Evidence type:** proof (what happened)

A PE-backed financial-services platform blamed its lead-scoring model for skewed results. The audit found the cause underneath: tens of thousands of contacts with no industry or company association, plus mis-tagged segments. LeanScale fixed the data, and the deduplication process around it.

## The challenge

Lead scores were skewing in ways the model design did not explain. An audit traced the cause upstream: roughly 90,000 contacts had no industry value and roughly 80,000 had no associated company record, so fit scoring had almost nothing to score against. Thousands more contacts were tagged to the wrong segment entirely. Separately, a bulk deduplication run was merging records in ways that would have quietly corrupted reporting, and a rep departure was about to leave thousands of in-flight contacts without an owner.

## The approach

Root-cause the skew instead of re-tuning around it
Traced skewed scoring to missing firmographic data rather than model design — roughly 90,000 contacts with no industry and roughly 80,000 with no associated company — and reported the cause rather than adjusting weights to hide the symptom.

Segment correction at the source
Fixed roughly 1,100 mis-tagged segment contacts as the upstream cause of the skew, and cleaned segments containing firms outside the approved list.

Auditing a merge run before it reached reporting
Audited a bulk deduplication run of roughly 2,100 merges and caught around 90 problematic merges — including roughly 60 discarded qualified leads — before they corrupted the reporting they fed.

Email validation across the database
Ran an email-validation audit to separate deliverable contacts from decayed ones, so list size stopped flattering the engagement numbers.

Ownership continuity through a rep departure
Managed a rep departure end to end — roughly 5,500 contacts reassigned across departed owners and 14 workflows updated — so no in-flight leads went cold during the handoff.

A taxonomy so the gaps do not reopen
Scoping a company-hierarchy and ICP taxonomy so firmographic data has a structure to hang on going forward, rather than being re-cleaned every year.

## Outcomes

Root cause found rather than papered over
The scoring skew was traced to roughly 90,000 contacts with no industry and roughly 80,000 with no associated company, and to roughly 1,100 mis-tagged segment contacts upstream of it.

~90 bad merges caught before reporting
An audit of roughly 2,100 bulk merges caught around 90 problematic merges, including roughly 60 discarded qualified leads, before they reached the dashboards.

No leads dropped through a rep exit
Roughly 5,500 contacts were reassigned across departed owners and 14 workflows updated during the transition.

## Quotes

> I know that you guys often help us with really like tactical implementation work, especially on the data side with [our data lead]. After hearing from the C suite this morning, like, I need help with the total roadmap.
>
> — Customer, Marketing & RevOps lead


## Canonical

https://knowledge.leanscale.team/customers/root-causing-crm-data-gaps-behind-bad-scoring/