If there's one thing to spend your unused HubSpot credits on right now, it's the Data Agent.

If you're a HubSpot Partner, you can see credit usage across your managed client portals. Observation across the board: credit spend is shockingly low. In my view, that can only mean one of two things:

• HubSpot customers don't know what's possible with AI credits.
• The cost-control tools (like credit caps) aren't common knowledge, so companies default to "better not — we don't want a surprise invoice."

Point 1 is business impact you're leaving on the table. Point 2 is a setting you simply need to configure. Both are solved by one feature that should be active in virtually every HubSpot environment: the Data Agent.

What does the Data Agent do? (and why you'll save time as early as tomorrow)

The Data Agent works through what HubSpot calls Smart Properties: intelligent CRM fields on your contact and company records that get enriched with external data. The agent populates these fields automatically based on triggers — a form submission, a workflow, or a fixed schedule.

In practice: sales no longer has to start from a nearly empty CRM record. The Data Agent pre-fills it based on web research, the prospect's website, shared documents, and call transcripts.

That also means you can enrich existing customer profiles — not just new leads. Think of information from past interactions or project documentation that you can use to surface upsell and cross-sell signals.

The result: data entry is still eating approx 15–20% of your sales reps sales time on "data-entry tax" Sales time = revenue.

Don't sleep on the Data Agent

This morning I sat down with one of our Solution Architects and asked him to walk me through the most impactful use cases for the HubSpot Data Agent. His immediate response:

"Having seen the Data Agent in practice at 5 of our customers now, I'm completely convinced it should be part of every HubSpot environment."

That same morning, we both came across new features that genuinely caught us off guard. More on that later.

AI cases with the Data Agent we have already set up

A few examples of what we've solved with businesses using the Data Agent — broadly applicable to other B2B companies:

• Routing leads to the right sales manager — Segment by industry or organization type, apply your territory mapping (see example above), and send leads to the right team automatically.
• Sales triggers for outbound — Open vacancies at prospects, announced projects, press releases, certifications achieved.
• ICP classification — Not just hard data points (revenue, headcount, location), but scoring leads on qualitative signals too.
• Product or project triggers — Are relevant applications or solutions in your market being mentioned on a prospect's site? The agent enriches those data points on the customer record.

Pattern #1 to break: the age-old Marketing & Sales friction

You probably know this argument. I've lost count of how many times I've heard it over the years:

Sales: "Hey marketing team, if you just add more fields to the website forms, we don't have to enter so much data in the CRM ourselves."

Marketing: "Yes, but that hurts our conversion rate."

And Marketing is right. Back in 2020, Eloqua and Unbounce already showed that the biggest drop-offs in conversion happen after 3+ form fields and again after 7+ form fields. In 2026, with AI making us even less patient, that gap has only widened.

HubSpot used to have progressive forms — you could queue form fields and skip ones you already had data for, swapping in a different question instead. You'd still enrich the record, but without hurting conversion. The catch: it only worked if you already knew the lead. The tension remained: Marketing wants the highest view-to-submission rate; Sales wants no manual admin.

The Data Agent genuinely removes that tension. Ask less on the form, let the agent fill in the rest.

Pattern #2 to break: the vicious cycle of "we'll stay away from AI for now — our data is already unreliable enough"

You'll hear a lot of consultancy firms say CRM is not fertile ground for AI if your data isn't in order so you need to fix that first (ok i partially agree) but that doesn't mean in the mean time you shouldn't do anything with AI. On the contrary.

AI can already solve a lot of data quality/gap issues even with imperfect data quality. Use AI to get your data in order, instead of waiting until your data is somehow perfect.

Ok but what does it cost?

While writing this article, a cold-outreach email landed in my inbox — genuinely coincidental, but completely relevant:

English translation:

Dear Ralf,

I came across your profile on LinkedIn. Quick question: how many of the NACE/SBI codes in your CRM are still accurate? Based on our analyses, the average is 58% — the remaining 42% is running behind reality, with direct impact on segmentation, scoring, and marketing automation. Here's what we do concretely: you upload your VAT numbers, we return them enriched with corrected NACE codes, current decision-makers, verified email addresses, and mobile numbers. 4M+ Benelux entities, 93% accuracy, refreshed daily. Match rate typically sits between 70–85%, and you only pay for what we actually find (€2 per match + €0.10 processing per row). A 500–1,000 record test is ready within 2–3 business days, typically €600–1,500. No platform license, no long-term contract. Interested in running your own data through our engine?

My first thought: the Data Agent in HubSpot does this too — and cheaper. And I'm not paying a service fee to an external party.

• Each enriched field costs 10 credits ≈ €0.10. (Small caveat: this may depend on your license — don't hold me to the exact figure.)
• Haven't used your credits yet? Then it costs you nothing extra and only adds value.
• Tip: Turn off "auto credit increase" and you'll never get a surprise invoice.
• Tip: Previews are free — you can test whether it works before spending a single credit.

So what was so mind-blowing?

During the brainstorm with our Solution Architect, we came across a new feature: Portal Specific Recommendations (HubSpot → Data Agent → Recommendations).

Not generic advice — specific use cases for your business. HubSpot knows your organization, your products and services, your deals, your customers. The Data Agent then interprets which enrichment could be relevant specifically for you.

Example — a company with greenhouse horticulture and sports/recreation as target markets. HubSpot's recommendations based on that context:

• Greenhouse climate optimization (light/heat/shade) — identifies new potential customers for coatings in real time, based on greenhouse growers struggling with excessive heat or poor light exposure.
• Field marking — outsourced or in-house — checks whether a sports club handles its line marking itself or contracts it out to the municipality, to avoid wasted sales effort.

To be honest: most of this you could think of yourself. But:

1. There are data sources and angles you'd never consider on your own. It's a useful source of inspiration.
2. There are very few commercial functions that ever sit down together — Marketing and Sales are silos — let alone to think strategically about enriching the customer profile and what that could unlock in terms of opportunity and time saved. These recommendations give you a concrete reason to test, and to bring something tangible to your commercial teams or co-workers.

SaaS and software companies have had tools for over 15 years to identify what stack a company runs — which CMS, CRM, or other tooling. That gave them a natural intelligence advantage about the customer situation. The Data Agent gives companies in manufacturing and business services the ability to close that gap.

But how reliable is this really?

Devil's advocate: not with all data. Reliability depends entirely on how reliable publicly available information is — or on data you already hold about your customers (which can include documents).

The upside: Remember previewing is free? You can test every use case against your existing customers and contacts in the CRM without spending a single credit.

How to get started

1. Check your credits — How many monthly credits you have available, and how many have you used? If the counter reads "unused," you can start enriching data today at no additional cost. In fact, you'd be using credits you're already paying for.
2. Audit your forms — Which fields could be enriched and removed from the form entirely — improving conversion — or handed off to the agent instead of to sales?
3. Check Portal Specific Recommendations — Before building anything custom, see what company-specific recipes the Data Agent has already identified for your business.
4. Define your use cases, or pilot one — Do a back-of-the-napkin business case. Time saved per sales rep (at €40–60/hr) versus 10 credits per enriched field.
5. Estimate your monthly volume — New records × fields × 10 credits. If it fits within your credit cap, automate it. If not, run a monthly workflow that triggers when credits reset.
6. Review and monitor — Is the value actually showing up? Keep costs under control, but don't cap your impact with a ceiling that's too low.

Where is your organization getting stuck with AI?

The Data Agent is one example of starting small and delivering immediate value. But most organizations don't struggle with the feature — they struggle with the step toward AI-first working: where do you start, what do you automate first, and how do you bring your team along?

If you're in the middle of that: send me a message or connect. I'm happy to think through how to approach this concretely and affordably — without turning it into a multi-year project.