---
title: "MDLH: Enrichment gap scoring"
url: "https://atlan.com/demos/mdlh-enrichment-gap-scoring/"
excerpt: "Measure exactly where your enrichment gaps are — by asset type and by domain — before your next sprint begins."
description: "Generate the governance scorecard your enrichment program should have started from. Score every asset type across five dimensions — descriptions, tags, certifications, ownership, and custom metadata — then slice the results by data domain to see which teams are furthest behind. If your AI agents are working blind because tables have 3% description coverage, this is the diagnostic step that tells you where to focus first."
format: "Video"
duration: "PT4M17S"
video: "https://videos.ctfassets.net/nwa1c00rtgxb/6lwLSi1rRqTRY8xfue7lX2/a61226cee1c16dc1e00dc624adfe6f05/MDLH_-_Enrichment_gap_scoring.mp4"
thumbnail: "https://images.ctfassets.net/nwa1c00rtgxb/25g56cVHrMdOCgUN72rSjW/e53efc4d134ffceb913be049f3c4918b/mdlh-enrichment-gap-scoring.webp"
content_purpose: ["Product Overview"]
target_persona: ["General", "Data Engineer", "Data Analyst", "Data Governance Lead"]
journey_stage: ["S2 - Discovery", "S3 - Solution Design", "S4 - Business Case", "C1 - Onboarding", "C2 - First Value", "C3 - Adoption"]
use_case_context: ["Training"]
product: ["Context For AI - Asset Profile 360"]
published: "2026-03-13"
updated: "2026-06-02"
content_type: "video transcript"
transcript_source: "sheet"
---

# MDLH: Enrichment gap scoring

Transcript of the video at https://atlan.com/demos/mdlh-enrichment-gap-scoring/

Your catalog has hundreds — maybe thousands — of assets harvested from across your data estate. Most of them have no description, no owner, no tags. You can't fix what you can't measure. One query gives you the full governance scorecard. We've watched governance teams run enrichment sprints for months — onboarding new domains, certifying critical tables, tagging sensitive assets — with no way to measure how far they've come. They finish a sprint and still can't answer: which domains are furthest behind? Which asset types are blocking discoverability? Once you run this query, you'll have the baseline every enrichment effort should have started from.

We're enriching metadata across a large estate — adding descriptions, assigning owners, applying tags, certifying critical assets — but we have no easy way to measure how far we've come. Without a completeness baseline, we can't identify which domains are furthest behind, which asset types are blocking discoverability, or whether our enrichment efforts are actually moving the needle. The Metadata Lakehouse turns our Atlan metadata into queryable SQL tables. In this walkthrough, we'll use it to generate a completeness scorecard — by asset type and by domain — so we know exactly where our enrichment gaps are before we start our next enrichment effort.

Inside of Snowflake, we have built out a query that looks at the completeness of our metadata on our assets. Here's what it measures. This completeness query measures five dimensions for each asset type: description coverage, tag coverage, certification coverage, ownership coverage, and custom metadata enrichment. Each one produces a percentage—the share of assets in that type that have the attribute set. The query is scoped by default to the asset types we're most concerned about for governance: Tables, Schemas, Glossary Terms, Data Domains, Data Products, and BI assets like Tableau Dashboards and Workbooks. We can add or remove asset types to meet our needs.

When we run it, here's what comes back. The output provides a row per asset type with coverage percentages across all five dimensions. We look at column description coverage first — in most orgs, this is the lowest number on the board. If it reads 2-3%, that’s a problem. Tags and certifications may tell a different story: if our team has been running playbooks for PII tagging or certification campaigns, we'll see that reflected here. This is our enrichment baseline. Every sprint, every source integration, every AI readiness initiative should start here. Here's why this matters beyond governance hygiene.

AI context engines and source systems both consume this metadata. If your tables and columns have 3% description coverage and missing tags, your AI tool is working blind — and any source system pulling enriched metadata from Atlan is getting an incomplete picture. This query is the diagnostic step before any enrichment program, any source push, any AI context build-out. It takes about 10 seconds to run. Now let’s break it down by domain to see where the gaps are concentrated. Now a second query gives us the same picture, but sliced by data domain — a row per domain with an overall enrichment score that averages across tags, glossary term assignments, readme coverage, and descriptions.

This is particularly useful for decentralized teams that need to track enrichment accountability at the domain level. One row worth watching specifically: "No Domain Assigned." That row captures all the assets in our estate that haven't been assigned to any domain. It's both an enrichment gap and a governance signal — those assets are entirely outside of our domain structure. The query is also flexible — let’s look at how to scope it to exactly what we need. Once we’ve seen the domain picture, the last step is scoping the queries to exactly what we need.

We can limit to specific asset types, connectors, owners, certification status, or even date range. If we're running an enrichment sprint focused on a specific domain or source system, the flexibility of the Metadata Lakehouse lets we narrow the scorecard to exactly what our team is accountable for. Since this is all SQL queries, our can even connect them directly to our BI tool of choice, allowing we to create up-to-date dashboards that track our enrichment and readiness. That's the full scope of what this query can do. We now have a governance scorecard that shows exactly where our metadata gaps are — by asset type and by domain.

That's the starting point for any enrichment sprint, any source integration readiness check, and any AI context initiative. The gaps are no longer a guess. Run it, pick our priorities, and start filling them in. If you want to see how to use the Metadata Lakehouse to run impact analysis — understanding how changes move through our data estate and making sure nothing you ship breaks a critical asset — that walkthrough is linked right below. [END YT CTA]