Quick answer
HubSpot smart properties are custom CRM fields that AI fills for you: you write a prompt, pick a data source (web research, the company website, another property, or activities and transcripts), and HubSpot's Data Agent researches the value for every record. Each fill costs 10 HubSpot credits per record (≈ €0.09 / $0.10) — charged even when no answer is found. Use them for research and interpretation; keep anything you can compute in free calculated properties or workflows.
HubSpot smart properties are CRM fields that AI fills in for you. You write a prompt, pick a data source, and HubSpot's Data Agent researches the answer for every record you point it at.
They're one of the more useful things HubSpot has shipped in a while. They're also one of the easiest ways to spend a lot of credits without much to show for it.
In this guide, you'll learn exactly how smart properties work, what they cost, where they beat calculated properties and workflows (and where they don't), and how to roll them out without surprises on your invoice.
What are HubSpot smart properties?
HubSpot smart properties are custom CRM properties that get their values from AI instead of manual entry, forms, or formulas. You attach a natural-language prompt and a data source to a property, and HubSpot's Data Agent generates the value for each record.
Think of them as a research assistant that reads the record, looks things up, and writes the answer into a field you can then use in lists, reports, and workflows.
They work on contacts, companies, deals, tickets, and custom objects. And because the output lands in a regular property, everything downstream in HubSpot treats it like any other field.
Smart properties belong to the Data Agent, one of HubSpot's Breeze AI agents. You'll see them referred to as "Breeze smart properties" or "Data Agent smart properties" — same feature.
Key takeaway: smart properties don't calculate. They interpret and research. That distinction decides almost everything about when you should (and shouldn't) use them.
How HubSpot smart properties work
A smart property has three parts: a prompt, a data source, and a field type. Here's what each one does.
1. The prompt
You describe what you want in plain language. For example:
- "Summarise what this company does in one sentence."
- "Does this company sell to other businesses, to consumers, or both?"
- "List the main objections raised in recent sales calls."
You can reference other properties inside the prompt using tokens, like "Does this company have an office in company.city?" That makes prompts adapt to each record.
2. The data source
Each smart property uses exactly one data source:
| Data source | What it reads | Best for |
|---|---|---|
| Web research | Searches the internet using the company domain (or name + email for contacts) | Firmographics, tech stack, competitor use, market signals |
| Company website | The company's primary domain only | "What do they do?", product lines, pricing model |
| Property data | One other property on the same record (including the first PDF in a file property) | Classifying free-text fields, extracting facts from uploaded docs |
| Activities and transcripts | Logged calls, emails, meeting transcripts, notes | Objections, next steps, buying signals, decision makers |
3. The field type
Smart properties can output single-line text, multi-line text, numbers, single checkboxes, multiple checkboxes, radio selects, and dropdown selects.
Structured types (dropdown, radio, checkboxes) are your friend here. Giving the AI a fixed set of options produces far more consistent results than free text, and the values are immediately usable in lists and reports.
Three ways to create one
From scratch. Create a new property, tick the "Data Agent prompt" option, and add your prompt and data source.
With Data Agent. Describe what you want ("Does this company have locations in Europe?") and Data Agent generates the property name, field type, prompt, and data source for you.
From an existing property. Take a custom property you already use in reports and segments and add a prompt to it. This is the least disruptive route if you're replacing a manual process.
How values get filled
Smart properties don't update live. A value is written when a fill runs, and it stays until the next fill. You can trigger fills:
- Manually from a record or index page ("Smart fill")
- In bulk from a segment or list
- On a schedule (daily, weekly, monthly) for a chosen segment
- On record creation
- From a workflow using the "Fill Smart Property" action (currently in beta)
Every filled value shows its sources. Click the AI icon on the cell and you'll see the web pages or CRM records the answer came from. That audit trail matters more than it sounds, and we'll come back to it.
What HubSpot smart properties cost
This is the part most guides skip, and it's the part that gets teams in trouble.
Smart properties consume HubSpot credits. Each fill costs 10 credits per record, which works out to roughly €0.09 / $0.10 per record.
Two things to understand about that number:
You pay for blanks. If the AI can't find an answer and returns nothing, you're still charged. A prompt that returns "Unknown" on 40% of records costs exactly the same as one that succeeds every time.
Cost scales with records, not properties. One smart property across 10,000 companies is 100,000 credits, roughly €900 / $1,000. Three smart properties across the same list is triple that.
Included credits by tier
Every Hub subscription includes a monthly credit allowance. For Marketing, Sales, Service, and Content Hub, expect roughly 500 credits on Starter, 3,000 on Professional, and 5,000 on Enterprise. Data Hub includes more (about 5,000 on Professional and 10,000 on Enterprise). We cover how credits work across all the Breeze agents in our Agent Hub guide.
Do the maths against your database. A Professional allowance covers a few hundred fills a month. That's enough to enrich new inbound leads. It's nowhere near enough to run a smart property across your whole company list.
Beyond the allowance, you buy credit packs (1,000 credits for around €9 / $10) or enable overages. Check your billing settings for a maximum credit limit and set one before your first bulk fill. Without it, heavy usage can trigger a permanent capacity upgrade rather than a one-off overage.
Which plans include smart properties
Smart properties are available on Starter, Professional, and Enterprise across Marketing Hub, Sales Hub, Service Hub, Content Hub, and Data Hub. You also need the generative AI settings turned on in your account (and the customer conversation and files data settings if you want to use transcripts or PDFs as sources).
Need help sizing credit usage before you switch this on across your portal? Talk to Superwork about a HubSpot data quality review and we'll model the cost against your actual database.
Smart properties vs calculated properties vs workflows
Here's the question we get most from clients: "Should this be a smart property?"
Usually, the honest answer is no. Not because smart properties are bad, but because HubSpot already has free, deterministic tools for most of what people try to use them for.
| Smart property | Calculated property | Workflow-set property | |
|---|---|---|---|
| How the value is set | AI reads a prompt and a data source | Formula (math, rollups, dates) | If/then rules you define |
| Deterministic? | No — same input can give different output | Yes | Yes |
| Cost | 10 credits per record per fill | Free | Free |
| Reads unstructured data? | Yes — transcripts, web, documents | No | No |
| Updates automatically? | Only when a fill runs | Instantly | When the workflow runs |
| Best for | Research and interpretation | Numbers and aggregation | Rules-based logic |
The simple decision rule
If you can already compute the value from data in HubSpot, don't use a smart property. Days since last activity, deal size bands, lifetime value, number of open tickets — all of these belong in calculated or rollup properties. They're exact, free, and update instantly.
If the logic is a fixed set of rules, use a workflow. "If industry is X and employees > 200, set segment to Enterprise" doesn't need AI. Our workflows guide covers this pattern in depth.
If you need to find something outside HubSpot, or make sense of text, that's a smart property job. What does this company sell? Which CRM do they use today? What objection came up on the last call? No formula can answer those.
One practical warning from experience: smart properties handle multi-condition logic poorly. If your prompt starts to look like "if A and B but not C, then…", move that logic into a calculated property or workflow and let the smart property answer one clean question.
How they differ from standard Breeze enrichment
Standard Breeze Intelligence enrichment fills predefined fields (industry, revenue range, employee count, and so on) from HubSpot's own dataset. Smart properties answer your question, and can read your calls and emails.
They complement each other. Enrichment gives you the baseline firmographics. Smart properties add the fields that are specific to how you sell.
How they compare to Clay and dedicated enrichment tools
Tools like Clay chain multiple data providers together ("waterfall enrichment") and typically achieve higher match rates than any single source. They also give you review steps before anything writes back to HubSpot.
The trade-off is another tool, another contract, and another place your data lives. Smart properties win on simplicity: no integration, native to the CRM, and immediately usable in every HubSpot tool. Clay wins on coverage and control at high volumes.
For most European B2B SaaS teams under a few thousand target accounts, smart properties are enough. Above that, or when match rate matters more than convenience, evaluate a dedicated tool.
6 practical use cases for HubSpot smart properties
These are the applications where we've seen smart properties pay for themselves.
1. Pre-call research
Prompt: "Summarise what this company does in one sentence, based on their website."
Source: Company website
Type: Single-line text
Reps stop tabbing out to research and start every call with context. This is the lowest-risk starting point because a slightly imperfect summary still saves time.
2. ICP fit classification
Prompt: "Based on this company's industry, size, and what they sell, classify their fit for a B2B SaaS RevOps consultancy as Strong, Partial, or Weak."
Source: Web research
Type: Dropdown (Strong / Partial / Weak / Unknown)
This turns gut feel into a field you can route on. Pair it with a calculated score that combines the AI classification with deterministic signals like deal size and engagement — the same layering we recommend in our lead scoring guide.
3. Tech stack and competitor detection
Prompt: "Which of the following does this company use: HubSpot, Salesforce, Pipedrive, Microsoft Dynamics? Select all that apply."
Source: Web research
Type: Multiple checkboxes
Useful for competitive displacement campaigns and partner programmes. It won't catch everything (companies don't always publicise their stack), but it's far cheaper than a technographics subscription for a targeted list.
4. Call objections and next steps
Prompt: "List the main objections raised by the prospect in recent calls, in bullet form."
Source: Activities and transcripts (calls, meetings)
Type: Multi-line text
This one changes deal reviews. Instead of asking reps to remember, managers see objections across the pipeline and can spot patterns by segment or product line.
5. Buying signal extraction
Prompt: "Based on recent emails and calls, has the prospect mentioned a budget, a timeline, or a decision maker? Answer Yes or No for each."
Source: Activities and transcripts
Type: Multiple checkboxes (Budget / Timeline / Decision maker)
Effectively automated BANT capture without asking reps to fill in more fields.
6. Ticket classification
Prompt: "Classify this ticket as Bug, Feature request, Billing, How-to, or Other based on its description."
Source: Property data (ticket description)
Type: Dropdown
Cheaper to run than most external classification tools, and the result feeds straight into ticket routing and reporting.
7 mistakes that waste credits (and how to avoid them)
The pattern we see repeatedly: a smart property works perfectly on a 50-record test, then a team runs it across 3,000 records and a third come back blank. Credits gone, trust in the feature gone.
Here's what usually went wrong.
1. Dirty domain data
Web research relies on the Company Domain Name property. If it's blank, wrong, or points to a parent company, the AI has nothing to work with. Clean your domains before any bulk fill — our guide to fixing bad data in HubSpot walks through exactly how.
2. Prompts that do too much
"Summarise the company, identify their tech stack, and rate their ICP fit" is three properties, not one. Long compound prompts produce vague answers at scale. One question per property.
3. No "Unknown" option
For dropdown and radio fields, always include an Unknown or Not found option. Without it, the AI either forces a wrong answer into your categories or returns nothing. Both cost the same.
4. Referencing LinkedIn
Prompts that mention LinkedIn are filtered and often return no value. Point the AI at the company website or general web research instead.
5. Skipping the preview
Every smart property has a preview that runs your prompt against a chosen record before you commit. Use it on five to ten records of varying quality, not just your best-known accounts.
6. Bulk-filling without a credit limit
Set a maximum credit limit in billing before your first bulk run. Then check the credit estimate HubSpot shows before you confirm a fill. Ten seconds of attention prevents a surprise line item.
7. Trusting the output blindly
Smart properties are AI outputs. They will occasionally be wrong. Any field that feeds scoring, routing, or compensation needs a human review step and a periodic accuracy check. Use the sources view to spot-check where answers came from.
How to roll out smart properties without regret
Here's the process we use with clients.
Step 1: Pick two or three high-value properties. Start with pre-call summary, ICP fit, and one transcript-based field. Skip anything you could calculate.
Step 2: Clean the inputs. Audit Company Domain Name coverage. Confirm call recording and transcription are on if you plan to use transcripts.
Step 3: Set the guardrails. Turn on the required AI settings, restrict smart-fill permissions to a few named admins, and set a maximum credit limit.
Step 4: Test on 50 records. Use a mix of good, average, and messy records. Aim for at least 80% useful results before going further. If you're below that, fix the prompt or the data, not the volume.
Step 5: Scale to a segment, not the whole database. Run your first bulk fill on active pipeline or new inbound leads, where the value is highest and the cost is bounded.
Step 6: Automate for new records. Once a property proves itself, set it to fill on record creation. This is where the included credit allowance actually stretches: enriching new leads as they arrive, rather than backfilling history.
Step 7: Review monthly. Check fill rate, blank rate, and credit consumption in the Data Agent management view. If blanks stay high after fixing prompts and data, that job probably belongs in a different tool.
GDPR and data residency: what European teams should check
HubSpot offers EU data hosting for CRM data, but AI features add a second question: where does the inference happen?
Before switching on smart properties for customer data, check HubSpot's current data processing addendum and subprocessor list for how Breeze features process data. Your legal team will also want to confirm the lawful basis for web-based enrichment of individuals (companies are far less sensitive than contacts here), and whether any smart property feeds a decision that could fall under automated decision-making rules.
None of this blocks the feature. Most teams end up using company-level smart properties freely and being more careful with contact-level web research. Just make the decision deliberately rather than discovering it in an audit.
Frequently asked questions
What is a smart property in HubSpot?
A smart property is a custom CRM property that gets its value from HubSpot's Data Agent AI. You write a prompt and choose a data source (web research, the company website, another property, or logged activities and transcripts), and HubSpot fills the property for each record.
How much do HubSpot smart properties cost?
Each fill costs 10 HubSpot credits per record, roughly €0.09 / $0.10. Credits are charged whether or not a value is returned. Filling one smart property across 10,000 records costs about 100,000 credits, or roughly €900 / $1,000.
Which HubSpot plans include smart properties?
Smart properties are available on Starter, Professional, and Enterprise tiers of Marketing Hub, Sales Hub, Service Hub, Content Hub, and Data Hub. You need the generative AI settings enabled in your account.
Which objects support smart properties?
Contacts, companies, deals, tickets, and custom objects.
Do smart properties update automatically?
Only when a fill runs. You can fill manually, in bulk from a segment, on a schedule, on record creation, or through a workflow action (in beta). The value stays until the next fill.
What's the difference between a smart property and a calculated property?
Calculated properties use formulas on existing HubSpot data and are free, exact, and instant. Smart properties use AI to research or interpret information, including from outside HubSpot, and cost credits per record. Use calculated properties for anything you can compute; use smart properties for anything you need to find out or interpret.
Can smart properties read call transcripts?
Yes. Choose "Activities and transcripts" as the data source and select the activity types (calls, meetings, emails, notes). You need the customer conversation data AI setting enabled and call transcription turned on.
Are smart properties the same as Breeze Intelligence enrichment?
No. Breeze Intelligence enrichment fills standard firmographic and demographic fields from HubSpot's dataset. Smart properties answer a custom question you define and can read your own CRM activities. Many teams use both.
Final thoughts
HubSpot smart properties are worth using when you have a question no formula can answer: what a company does, which tools they run, what a prospect objected to on the last call. For those jobs, they're cheaper and simpler than any alternative that doesn't involve adding another tool to your stack.
They're not worth using for anything you can calculate, and they're expensive to use carelessly. Clean your domains, write single-purpose prompts with an Unknown fallback, test on 50 records, set a credit limit, and scale to a segment before you scale to the database.
Do that, and smart properties become one of the more reliable ways to get your CRM to fill itself in.
Thinking about where AI fits in your HubSpot setup? Book a RevOps review with Superwork and we'll map which of your fields should be calculated, which should be workflows, and which are worth spending credits on.