Case study · HubSpot
An in-context credit estimate in Data Studio showed customers what an AI action would cost before they ran it, and Data Studio activations rose about 12%.
Customers avoided paid AI features in Data Studio because they couldn’t predict how many credits an action would cost. CSAT was falling, with billing friction as a top cause, and some customers were paying for features they didn’t need.
I narrowed a broad adoption study to one fast question: does showing cost upfront increase activation? Then I shipped an in-context credit estimate next to each AI action, plus clearer copy about credits. That meant getting approval from Legal and other organizations despite pushback.
Business context: Data Studio launched at INBOUND ’25 as HubSpot’s workspace for building and cleaning datasets. Its AI features, such as Smart Columns, run on HubSpot credits under a new usage-based billing (UBB) model. Adoption of those features drives both customer value and revenue.
Customer problem: Customers couldn’t tell what an AI action would cost until after they had run it. Faced with an open-ended bill, many simply didn’t activate. Others spent credits inefficiently, for example paying for Smart Properties to solve problems that free enrichment could already handle.
Evidence: CSAT was declining with UBB friction as a top driver, and the issue had executive visibility. [CSAT figures, number of interviews or support tickets, and where users dropped off in the activation flow.]
“[Customer quote about not knowing what an action would cost]”
Problem statement: How might we help Data Studio users understand what an AI action will cost before they run it, so that they activate features with confidence instead of avoiding them?
I owned: Discovery and synthesis, the problem framing, hypotheses, success metrics and guardrails, the end-to-end UX of the credit estimate, and getting Legal and cross-org approval to ship it.
I partnered on: The UBB pricing model and credit-usage UX with the UBB platform team; updated in-product copy with Content Design; implementation with multiple Data Studio engineering teams.
As UX DRI for Smart Columns, I had already designed its credit usage, web search, UBB and error states. That put me close to the usage data. I gathered evidence of credit anxiety, including customers paying for features they didn’t need, and used AI tools (NotebookLM, Granola) to collect and synthesize research faster.
Key insight: Low adoption wasn’t only about feature value. The fear of an unpredictable bill was a problem on its own, and the system as a whole was risking customer trust and activation.
The first research plan tried to answer both “why don’t people adopt Smart Columns?” and “are people averse to credits?” at once. I recognized we were boiling the ocean and narrowed the scope to one fast question: does showing cost upfront increase activation? I set hypotheses, success metrics and guardrails for it.
[Concepts you tried, such as where and when the estimate appears, ranges vs. exact numbers, and copy variations, with notes on what you rejected and why.]
After Legal approved it, I launched an in-context credit estimate experiment for Data Studio activations in December 2025. Alongside it, I worked with Content Design to fix comprehension issues in the credit copy that were hurting monetization of new features.
How I got buy-in: I turned the UBB constraints and usage data into concrete customer problems, ran a problems and opportunities exercise across Data Agent and UBB, and set guardrails so stakeholders could approve a contained experiment instead of a permanent change. [Who was skeptical and what changed their mind.]
When a user sets up an AI action, such as a Smart Column, Data Studio shows the estimated credit cost right next to it. Users can decide with the cost in front of them, instead of finding out from their bill. [How the estimate is calculated and displayed, e.g. per row, as a total, or as a range.]
Working with Content Design, I rewrote credit copy that users were misreading, so they could understand what they’d pay for and why.
The estimate became a pattern for the UBB platform and helped set new content standards. Teams in Data Platform and Marketing Hub, including Events and Email/SMS, asked to adopt it.
Showing the credit estimate in context raised Data Studio activations by about 12% in early data after the full rollout on 1/12/26. [Baseline, final number, time period, and whether this came from an A/B test or a before/after comparison.] The pattern also shaped new UBB platform patterns and content standards.
Qualitative impact: My manager wrote that the work “identified and addressed systemic risk to customer activation and trust.”
Wider context: Since INBOUND ’25, more than 400 portals use credits in Data Studio (+292% since September 2025), and Smart Columns alone account for 320K+ credits.
“[Customer or PM quote about the work, used with permission]”