HubSpot Credits: We Don't Love Them Either. Here's How to Deal With Them.

Let's get this out of the way: we don't love HubSpot's credit system.

We've been in HubSpot for fourteen years. We've watched pricing models evolve, features get bundled and unbundled, and tools shift from "included in your subscription" to "that'll cost extra." Credits are the latest version of that evolution, and honestly? They're confusing. The pricing feels inconsistent. The visibility into what you're spending is limited. And the fact that two tools with the same unit price can cost wildly different amounts depending on how you use them is the kind of thing that catches people off guard.

But here's the thing. Credits are here. They're not going away… probably… for now. And if you're on HubSpot Pro or Enterprise, you already have them (3,000 per month on Pro, 5,000 on Enterprise). So the question isn't whether you like the system. It's whether you're getting value from the credits you're already paying for.

We went to a webinar recently where a HubSpot product advocate walked through how credits actually work, and it clarified a few things that are worth sharing. Because the way credits are presented on HubSpot's pricing page doesn't tell the whole story.

The Same Price Doesn't Mean the Same Cost

This is the single most important thing to understand about credits. Two features can have the same unit price and cost you completely different amounts.

Buyer intent costs 10 credits per company per month. If you're tracking 300 companies, that's 3,000 credits. Your entire Pro allotment, but manageable. And that number is predictable because you control how many companies you're tracking.

Smart properties (from Data Agent) also cost 10 credits. Per fill. If you build five smart properties on contacts and five on companies, and you try to backfill your entire database of 250,000 contacts and 25,000 companies... that's 2.75 million credits. Not a typo. Same unit price. Completely different universe of cost.

The difference is volume. Buyer intent operates on a bounded number of company records. Smart properties operate on every record you point them at. And if you set them to auto-fill without thinking through the math, they will consume your credits in hours and then stop working.

That's the pattern to watch for. Some credit-consuming tools are naturally bounded (you control the volume). Others can run away from you if you're not paying attention.

Which Tools Are Safe to Experiment With

Based on what we heard and what we've seen in client portals, here's how we'd categorize the credit-consuming tools:

Low risk, high signal. Turn these on.

Buyer intent is the easiest recommendation. 10 credits per company per month. Predictable volume. High value signal. If you haven't turned this on yet, you should. It tells you which companies are showing purchase intent based on their behavior, and at this price point, it's unlikely to blow through your credits.

Customer agent (the AI chatbot) costs 50 credits per resolution. And the key word is resolution. If the agent responds to a visitor but doesn't actually resolve the issue, it doesn't cost you anything. You only pay when it provides value. That's the closest thing to outcome-based pricing HubSpot has figured out so far, and it's a good model. Resolution rates across HubSpot's customer base are approaching 70%, with some companies breaking 90%. This one earns its credits.

Medium risk. Set limits and watch it.

Prospecting agent costs 100 credits per suggested lead. That adds up fast if you want it to surface hundreds of leads. But you set the monthly limit when you configure it, and it sticks to that number. So you control the spend. Start small, see if the leads are relevant, and adjust from there.

Content agent costs 1,000 credits per piece of content. That's the highest unit price. But it's almost entirely manual. You're telling it to write a blog post or create an asset. It's not going to run itself. The question here isn't "will it run away from me?" It's "is this worth 1,000 credits compared to writing it myself or hiring someone?" That depends entirely on your situation… honestly, I hate this one the most. Just… use your other AI tools to write this.

High risk. Be very careful.

Data agent (smart properties) is the one that can get out of control. 10 credits per fill sounds cheap until you multiply it by the number of records and the number of properties. Build smart properties intentionally. Set them to manual fill or trigger them from specific workflow conditions (like a deal reaching a certain stage). Do not set them to auto-fill across your entire database. Test on a small group first. Always.

Workflow breeze actions are the other volatile one. If you're used to building workflows (and if you're reading this blog, you probably are), you know they can enroll hundreds or thousands of records. Adding a breeze action to a high-volume workflow means every enrollment consumes credits. And unlike a regular workflow action that just runs, this one has a cost attached to every execution. Think carefully about volume before you add AI actions to any workflow.

Data Studio charges per use based on data set size. 25 credits for small (under 500K rows), 75 for medium (up to 5M rows), 200 for large (over 5M rows). It's mostly manual, so it's unlikely to run away. But if you're exporting or enrolling records frequently from large data sets, it adds up.

Credits Stop When They Run Out (Which Is Its Own Problem)

If you're using your included credits and you hit zero, the credit-consuming features shut off. You don't get a surprise bill. That's the good news.

The bad news is that whatever those features were doing just stops. If a smart property was mid-backfill, it stops wherever it is. There's no queue. It doesn't pick up again next month. And there's no easy way to see how far it got. So you could end up with half your database enriched and the other half empty, with no clean way to figure out where the cutoff was.

If a workflow with breeze actions runs out of credits, that workflow stops firing those actions. If customer agent runs out, it stops resolving tickets.

The features don't fail gracefully. They just stop. Which means if you're depending on any of these tools for something important, you need to be watching your credit usage, not discovering it after the fact.

How to Manage Your Credits

Here's what we recommend:

Go look at your credits page. Seriously. If you haven't been there, go now. In your HubSpot account, find the credits section under billing. You can see how many you have, how many you've used, and what's consuming them.

Set feature-level limits. You can set a maximum credit allocation per feature. If you don't want workflow actions or custom agents consuming credits at all, you can turn them off from this page. If you want buyer intent to run but you want to cap smart properties, you can do that. Use these controls. They exist for a reason.

If you buy additional credits, set a total monthly maximum. You can choose between pay-as-you-go (you pay per credit once you exceed your included amount, up to your set max) or auto-upgrade (your credit tier adjusts to match your actual usage). Pay-as-you-go with a hard cap is safer while you're still figuring out your patterns. Auto-upgrade makes sense once you understand your usage, but not in your first month of experimenting.

Don't let smart properties run unsupervised. We can't say this enough. Build them with intention. Test on a small segment. Set them to trigger from specific conditions, not blanket auto-fill. The value is real when they work. The cost is real when they run wild.

Start with buyer intent and customer agent. If you haven't used any credit-consuming tools yet, these two are the safest starting points. Low risk, high value, predictable cost. Turn them on, see what you learn, and build from there.

The Honest Take

Credits are HubSpot's attempt to align the cost of AI tools with the value they provide. The intention is good. Pay for what you use. Don't pay for what you don't. And in some cases (customer agent's per-resolution pricing, specifically), they've gotten it right. You're paying for outcomes, not just access.

But the execution is still messy. The pricing isn't intuitive. The visibility into usage is limited. The fact that features just stop working when credits run out (without a queue, without a graceful degradation, without a clear way to see what happened) is a real problem. And custom agents in Agent Hub are so new that you genuinely don't know what they'll cost until you've run them a few times.

HubSpot knows this. They're iterating. Usage-based pricing is going to evolve. Outcome-based pricing (where you pay based on results, not just activity) is the direction they're moving, but they're not there yet for every tool.

In the meantime, the best thing you can do is understand what you have, set the right guardrails, experiment at low volume, and pay attention to what's actually consuming your credits.

If you want help making sense of your credits, figuring out which tools are worth turning on, and setting up the guardrails so nothing gets out of hand, that's what we do.