Here's a pattern we see constantly in real portals: a team sets up a beautiful Data Agent prompt to enrich contact records. It works great for a week. Then it starts erroring out on half the database because some records are missing a job title, a company size, or an industry field. The prompt either stalls or charges credits and returns nothing useful.
That's the problem this update solves. And it's a real one.
What This Update Actually Is
HubSpot shipped optional and required token configuration inside Data Agent prompts. When you build a prompt and add a CRM token, you now see a toggle: required or optional.
That's it. No new interface, no extra setup. Just one toggle per token that gives you real control over when a prompt fires and when it doesn't.
Before this change, all tokens behaved the same way when blank. That left teams choosing between two bad options: write vague prompts with no CRM context, or accept that prompts would fail on messy records. Neither option is good.
Why HubSpot Shipped This
No CRM database is perfectly complete. That's not a failure of your team. It's just reality. Humans fill out forms with different levels of detail. Sales reps log calls and skip fields. Imported lists have gaps. Even the cleanest portals have partial records.
Data Agent is built to enrich and analyze CRM data at scale. But if every blank field stops a prompt cold, the tool becomes unreliable on real-world data. HubSpot needed to give practitioners a way to say: this context is nice to have, but this context is non-negotiable.
There's also a credit dimension here. HubSpot Credits power Data Agent runs. Burning credits on a prompt that was always going to fail without a critical piece of data is wasteful. The required token behavior stops the prompt before it charges anything, which is a meaningful protection for teams running enrichment at volume.
How to Use It Step by Step
- Navigate to Data Agent inside your HubSpot portal and open an existing prompt or create a new one.
- Add a CRM token to your prompt by selecting the token from the property picker as you normally would.
- When the token appears, locate the required/optional toggle next to it.
- Set tokens to required for any field where the prompt is meaningless without the data. Example: if your prompt scores a contact based on job title, job title is required.
- Set tokens to optional for any field that adds context but isn't a dealbreaker. Example: company revenue adds depth to a prompt but the prompt can still run without it.
- Save and test the prompt against a record you know has gaps. Confirm that optional-token prompts run and required-token prompts error out cleanly without charging credits.
- Review your existing prompts and apply this logic retroactively. Any prompt you already built treats all tokens as required by default, which may be why some of them fail on partial records.
What It Touches in Your HubSpot Strategy
This update looks small on the surface. It isn't. It touches every part of your portal where Data Agent runs, and that list keeps growing.
Data enrichment at scale: If you're running batch enrichment across thousands of contacts, deals, or companies, partial records no longer kill your whole run. Optional tokens let the prompt do what it can with what's available. Required tokens protect your credits where the output would be garbage anyway.
Workflow-triggered prompts: Many teams trigger Data Agent inside workflows, either on enrollment or on property change. Without token flexibility, those triggers often fired and failed silently on records missing key fields. Now you can build reliable automation logic around what the prompt actually needs.
Key Takeaway
Audit every existing Data Agent prompt you have. Any prompt built before this update treats all tokens as required by default. That means prompts that could have run on partial records have been failing unnecessarily. Go back and set optional flags on every non-critical token.
Credit management: HubSpot Credits aren't free, and they aren't infinite. Required tokens acting as a gate means you're not burning credits on prompts that were always going to return unusable output. For teams running high-volume enrichment, this is a real cost control mechanism.
This update pairs directly with the earlier Data Agent expansion that let you query your entire CRM in a single prompt. The more data sources Data Agent touches, the more variation you'll encounter in field completeness. Token flexibility is infrastructure for everything that comes next.
Key Takeaway
Think of required tokens as your prompt's quality gate. If a prompt would produce a misleading or useless result without a specific field, mark that field required. Don't let bad AI output pollute your CRM just because a record is incomplete.
Data quality and property hygiene also come into play here. If you notice that most of your required tokens are erroring out, that's signal: your CRM has field coverage gaps that need attention. If you haven't done a full property and data audit recently, our HubSpot portal audit checklist is a good starting point before you invest more time in Data Agent prompt architecture.
Who Should Care Most
This update matters most to the humans who are already using Data Agent at any meaningful scale. Here's who should prioritize it:
- RevOps and CRM admins running batch enrichment across large contact or company databases with inconsistent field coverage
- Marketing ops teams triggering Data Agent prompts inside enrollment workflows where record quality varies by source
- Sales ops leaders who've set up prompt-based lead scoring or qualification summaries and noticed inconsistent output on certain segments
- Portal admins watching HubSpot Credits and looking for ways to reduce waste on failed or low-quality prompt runs
- Growing companies that acquired contacts through list imports, trade shows, or third-party integrations where field completeness is lower than inbound-generated records
If you're on Starter and just getting started with Data Agent, this is still worth understanding now. It shapes how you should structure prompts from the beginning so you don't have to rebuild them later.
George's Take
I've been inside enough portals to know that messy data isn't an exception, it's the baseline. Most databases have 20 to 40 percent of records with at least one key field missing, and that's not because the humans managing those portals are doing a bad job. It's because data enters a CRM from too many sources with too many formats. What I love about this update is that it stops treating incomplete data as a failure state. It treats it as reality. Required tokens are your quality control layer. Optional tokens are your flexibility layer. Together, they let you build prompts that are honest about what they need and graceful about what they don't.
“Stop building prompts that pretend your CRM is perfect. Mark the tokens that actually matter as required, give everything else an optional flag, and let Data Agent do its best work on real-world data.”
This update is one piece of a much larger shift in how Breeze and Data Agent are expanding across the platform. If you want the full picture of where HubSpot's AI layer is heading, read our April 2026 Breeze overview to see how all the pieces connect.
If you want help auditing your existing Data Agent prompts, cleaning up your token logic, or building a credit-efficient enrichment strategy across your portal, that's exactly what we do at Sidekick Strategies. Book a strategy call and let's look at what you're working with.
Frequently Asked Questions
What does marking a Data Agent token as optional actually do?
When a token is marked optional, the Data Agent prompt runs even if that field is blank on a record. The prompt simply works with whatever context is available. This lets you run enrichment across records with incomplete data without stopping the entire process.
What happens when a required token is blank in a Data Agent prompt?
If a required token is blank when the prompt runs, Data Agent returns an error and does not execute the prompt. Critically, it also does not charge HubSpot Credits for the failed run, which protects your credit budget from being spent on prompts that can't produce useful output.
Do I need to update my existing Data Agent prompts after this release?
Yes. Prompts built before this update treat all tokens as required by default. That means any prompt with a token that could be optional has been failing unnecessarily on partial records. Review your existing prompts and set optional flags on any non-critical fields to improve coverage and reliability.
Which HubSpot plans include this Data Agent token configuration feature?
This feature is available across Marketing Hub, Sales Hub, Service Hub, Content Hub, Data Hub, Commerce Hub, and Smart CRM at Starter, Professional, and Enterprise tiers. HubSpot Credits are required to run Data Agent prompts regardless of tier.
How do I decide which tokens should be required versus optional?
Mark a token required if a blank value would make the prompt output meaningless or misleading. For example, a job-title-based scoring prompt needs job title. Mark a token optional if it adds useful context but the prompt can still produce a valid result without it, like company revenue or LinkedIn URL.
Can optional and required tokens be mixed in the same Data Agent prompt?
Yes. You can configure multiple tokens in a single prompt and set each one independently as required or optional. This gives you precise control over which data is a hard prerequisite for the prompt to run and which data is additive context that improves the output when present.
