Data Chronicles Part 2: Smart Properties in HubSpot

You know that feeling when a tool you’ve dreamed of suddenly happens — like, where was this six months ago?

That’s Smart Properties in HubSpot.

On paper, they sound like a minor feature — “AI-filled CRM fields with customizable prompts” — but in practice, they’re an entirely new way to enrich, score, recommend, and route without the overhead of yet another enrichment tool duct-taped to your CRM.

It’s like having a tiny RevOps robot living inside every record, pulling context from behavior, transcripts, and firmographics — and writing stuff just when you need it, not every time someone blinks.

Let’s walk through what they are, how to use them, 15+ high-impact examples,  and some under-the-hood insights that’ll save you from a messy implementation.

What Are Smart Properties (Really)?

Smart properties are AI-powered CRM fields that fill themselves with custom prompts.

You define:

• The object (Contact, Company)

• A property type (text, number, enum, etc.)

• A prompt that references existing data or transcript/web inputs

Then you decide when and how they fill:

• Per record (manually or on assignment)

• In bulk (from a filtered view or list)

• Automatically (via workflow triggers)

They consume HubSpot AI credits, so yes, there’s a cost — but it’s predictable, manageable, and capped by plan-level limits or overage settings.

Creating Smart Properties: Data Agent, Templates & Input Sources

If you’re starting from scratch, HubSpot’s built-in Data Agent (Breeze Assist) makes it absurdly easy to get going without prompt anxiety.

Option 1: Create with Data Agent Assist (Breeze)

In the smart property setup flow, you can just describe what you want the property to do — like:

“I need a property that surfaces common objections in the sales process.”

HubSpot will suggest:

• A prompt structure

• Matching data source (e.g., call transcripts)

• Recommended property format (text, number, etc.)

It’s not quite no-code, but it’s no-headache. Perfect for getting started or experimenting before fine-tuning the prompt manually.

Option 2: Use Pre-Built Prompt Templates

Once you pick your data source, HubSpot offers suggested prompts based on the context — so if you’re not sure how to phrase “Summarize call objections” or “Score ICP fit,” these templates can give you a head start.

You’ll see a “Suggest prompts” link next to the prompt field. Click it. Use it. Tailor later.

Selecting the Right Data Source

HubSpot supports four main data sources when creating a smart property:

⚠️ Important: You can only select one property as the data source if you're using “Property data.” 

Live Preview Built In

Once you’ve chosen your source and written your prompt, you can preview the results on real records before you save or trigger a fill.

HubSpot shows you:

• The record name (dropdown)

• The preview output from the AI

• How the prompt behaves on actual CRM data

This preview is your best QA step — always test your prompt against 2–3 known records before deploying across a view or list.

Why Smart Properties Beat Bulk Enrichment

Old school enrichment is a blunt instrument — you enrich everything, whether or not it’s useful. It’s noisy, expensive, and usually stale within weeks.

Smart properties flip that. You enrich only when it matters.

• Scoped to high-value actions: e.g., viewed pricing page, replied to SDR

• Custom prompts per field: “Based on job title and behavior, assign persona”

• Real-time enrichment: Run them manually or via automation at key lifecycle moments

It’s enrichment-as-a-service, baked into your CRM, with none of the Zapier glue or API wrangling.

15+ Use Cases Worth Stealing

Marketing / Content

• Relevance Engine – Recommend 3 articles a contact hasn’t read yet

• Topic Affinity Score – Score interest in topics like “Pricing” or “Implementation”

• Campaign Match – Classify contact into persona bucket based on UTM + behavior

• SEO Gap Nudge – Suggest internal link anchors for underlinked posts

Sales

• Hot Intent Trigger – Set "Ready to Talk" if contact hits key pages + replies

• Meeting Prep Brief – Top 5 pages + 3 likely objections, summarized

• Buying Role Guess – Classify Economic vs. Technical buyer from job + behavior

• Deal Risk Signal – Risk level from idle days + sentiment

Support / Success

• Churn Watch – Risk score based on NPS trend, tickets, and usage dips

• Upsell Hint – Add-on rec based on peer behavior and feature usage

• Reply Assist – One-liner recap from last 5 support interactions

Ops / Data

• ICP Fit Score – Score 0–100 based on industry, size, revenue

• Lifecycle Autopilot – Suggested lifecycle move based on activity

• Source of Truth Selector – Pick best value across conflicting fields

• PII Scrub – Flag and redact sensitive data from notes

Smart Properties Only Work If the Data Exists

Here’s the unsexy truth: no matter how clever your prompt is, AI can’t synthesize insights from a black hole.

Smart properties only work when the data you’re referencing is:

• Stored in HubSpot (as properties or attachments),

• Available via transcripts (for call-based prompts), or

• Publicly accessible online (for company-level web research).

And if it’s not? Well… the model has nothing to work with. It’ll either hallucinate, return “insufficient data,” or — worse — give you false confidence in a half-baked answer.

Examples:

• Call transcript prompt? → If you don’t store call transcripts in HubSpot, the AI’s working off air.

• Company website analysis? → If Company Domain Name is blank, web research falls flat.

• ICP scoring via industry/size? → If those fields are empty or inconsistent, your score is noise.

Good inputs = great outputs.

If you're serious about operationalizing AI inside HubSpot, you need to treat data hygiene and availability as part of your smart property strategy. Think of it like feeding a chef — you can’t ask for a gourmet meal with no ingredients in the fridge.

So before you spin up that genius prompt:

• Audit which objects actually contain the data you need

• Standardize key inputs (especially for workflows and scoring)

• Turn on call recording + transcription if you plan to summarize objections or pull meeting briefs

• Store files consistently on the right records (e.g., PDF proposals, scopes, etc.)

 AI doesn't guess well when it’s starving.

Make sure you’re feeding it something useful.

Permissions and Limitations

As you might guess, there are some limitations, especially if you don't have the proper access to manage.

• To create or edit a smart property: you need Super Admin or Edit property settings permission

• To enrich/run smart properties: you need Super Admin or Data enrichment access

And heads-up:

• You cannot apply validation rules to ALL smart properties, only some

• You cannot mark smart properties as Sensitive Data

Also — if your prompt depends on LinkedIn (e.g., “Search LinkedIn profile...”), it’s likely to be filtered or return nothing. HubSpot will skip these requests due to policy restrictions.

Final Thought: You’re Not Automating. You’re Creating Context.

This isn’t about doing “AI” because it’s trendy. It’s about surfacing context that helps reps close faster, marketers personalize better, and workflows make smarter decisions.

Think of smart properties like little AI concierges living inside your CRM — waiting for the right moment to drop a killer insight, score, or suggestion.

You’re one prompt away from magic. Just make sure it’s scoped, previewed, and backed by good data.