A legal tech company buys HubSpot's AI, switches it on, and waits. The dashboards populate. The recommendations appear. And the results are underwhelming — generic prompts, a competitor list that doesn't quite match, content suggestions aimed at a buyer the company stopped selling to two years ago.
The natural conclusion is that the tool was oversold. The natural conclusion is wrong. The tool is reading the CRM underneath it, and the CRM is not ready.
Readiness comes before the tool
Every AI feature HubSpot ships is downstream of CRM data. The Breeze agents read what's in your Smart CRM to decide what to suggest, and the same is true of the recommendation and enrichment layers. When the data is clean and complete, the output is sharp. When it isn't, the AI produces confident, specific, wrong answers — and nothing in the interface warns you.
This is not a HubSpot quirk. It is the defining failure mode of enterprise AI. Gartner estimates poor data quality costs organizations an average of $12.9 million a year, and notes that many data and AI initiatives fail for exactly that reason.
For a legal tech company, the problem is sharper still: a CRM that can't tell a law firm from a corporate legal department from a competing vendor gives the AI nothing to segment on. And because the category is crowded with similar-sounding products, vague records push the AI toward the best-known names when it generates recommendations — which usually means a competitor, not you.
Five signals your CRM isn't AI-ready
Five patterns reliably signal that a CRM can't yet feed AI — each one breaking something specific:
- Inconsistent or blank industry and segment tagging — the AI can't suggest prompts or content by buyer type.
- Duplicate contacts and orphaned companies — deal patterns become unreadable, and one buyer looks like three.
- A pipeline that still reflects a previous era — competitor lists surface vendors that no longer matter.
- An ICP set at onboarding and never updated — every recommendation targets a buyer you've outgrown.
- Key fields are empty or filled with free-text chaos — there's nothing reliable to score, route, or personalize on.
The sixty-second test
You don't need a formal audit to know which side of this you're on. Pull ten records at random — five contacts, five companies. For each one, ask whether an outsider could tell who the buyer is, which segment they belong to, where they sit in the pipeline, and why the record exists at all.
If a new hire couldn't read the record, the AI can't either.
If most of the ten pass, the foundation is probably sound and the AI investment is worth making now. If most fail — blank industries, mystery deals, contacts attached to no company — the readiness gap is real, and it is the thing to fix before any AI purchase, not after. The companies that skip this step are the ones that conclude, six months later, that the tool doesn't work.
Fixing it, in order
The fix is unglamorous and sequential, and the order matters more than the speed. Enriching records before fixing the tagging taxonomy, for instance, only fills a broken structure faster.
-
Start with structure: a real tagging taxonomy that distinguishes the segments the business actually sells to.
-
Hygiene — merge duplicates and set association rules so each contact is tied to a company.
-
Reset the pipeline to reflect the last twelve months of real selling rather than the assumptions of a previous era.
-
Redefine the ICP with the current sales team in the room, not the version written at onboarding.
-
Finally, make a handful of fields required and let HubSpot's enrichment — drawing on a database that indexes over 100 million company domains and 380 million email addresses — fill the remaining gaps automatically.
Done in that order, readiness is a project a legal tech company finishes in a few focused weeks. Done out of order, or skipped entirely, it becomes the quiet reason an AI investment underperforms for a year before anyone connects the two.
The tool was never the variable that decided the outcome. The foundation was.
If you find this content insightful, consider subscribing to our LinkedIn Newsletter and following our company page for our latest articles!
