HubSpot's earlier emails tailored content to demographic and interest-based cohorts, but each recipient still received messaging designed for a group. AI enabled the team to infer an individual's likely job to be done and make a one-to-one recommendation at a scale that manual copywriting could not support.

Traditional personalization targets finite cohorts
Individual intent can differ within the same segment
AI can infer a recipient's likely job to be done
Automation makes one-to-one recommendations economically feasible

“So it's really personalization based on segmented cohorts.”
Emmy Johnson · 16:30

“And then use AI to actually do instead of cohorted personalization, one to one personalization at scale.”
Kip Bodner · 19:00

#email personalization#segmentation#jobs to be done#generative ai

Myth Buster29:00

The Recommendation Drove the Lift, Not the Personalized Copy

The first experiment assumed that highly personalized prose would produce the biggest conversion improvement. Testing instead showed that the decisive factor was accurately inferring the recipient's job to be done and recommending content that could help complete it.

Personalized wording was not the primary conversion driver
Accurate intent inference mattered more
The recommended content needed to solve the inferred job
Test the underlying value proposition separately from generated copy

“But the thing that's really most important is being able to accurately guess what the job is this person is looking to do, and then…”
Emmy Johnson · 29:00

“And what we found was really where the step function conversion rate improvements came is not necessarily in how personalized this text is, but really…”
Emmy Johnson · 30:30

#recommendation relevance#email copy#jobs to be done#cro

Hot Take· 1

Hot Take10:00

AI Products Cannot Be Perfect Before Real Users Touch Them

The speakers argue that AI systems need real-world interaction data before they can be optimized. Teams should launch at an acceptable baseline, set expectations, and then improve the model through repeated feedback and training.

Do not wait for a perfect model before launch
Collect feedback from real user interactions
Iterate and retrain after release
Set expectations that the first version will improve

“Until you actually release it into the wild, you're not going to be able to get it to perfection because an AI model really needs…”
Emmy Johnson · 10:30

“Know it's not gonna be great when you first launch it, set the expectations, and know that you're gonna get it to great as you…”
Kip Bodner · 12:00

#shipping#model training#user feedback#iteration

Explainer· 4

Explainer02:00

How HubSpot Prioritized More Than 100 AI Marketing Ideas

HubSpot needed to allocate finite resources across more than 100 proposed AI experiments. The team scored ideas by their potential impact on demand or brand awareness and by how broadly employees could use them, then focused on the strongest 10 to 15 candidates.

Collect ideas from the entire marketing organization
Estimate each idea's top-of-funnel impact
Estimate breadth of internal use
Prioritize ideas that score highly on both dimensions

“So we had to come up with a way to prioritize the use cases, the ideas that people were bringing forward so that we could…”
Emmy Johnson · 02:00

“And then we plotted the top 10 to 15 use cases within that two by two.”
Emmy Johnson · 04:30

#ai strategy#prioritization#marketing operations#experimentation

Explainer08:30

Why HubSpot Put Its AI Team Inside Marketing Technology

HubSpot concluded that AI experimentation needed a centralized owner capable of moving quickly. It placed the function within marketing technology, whose team designed the intake process, scored ideas, and coordinated execution.

Centralize ownership of early AI initiatives
Choose a team with technical and marketing capabilities
Give one group responsibility for intake and prioritization
Start with a focused person before expanding the team

“We need to really start building out a team that can take this on and centralize it somewhere within the marketing team so that we…”
Emmy Johnson · 09:00

“But we got started, we kind of had like one person who is focused and spending time on our top things, right?”
Kip Bodner · 10:00

#marketing technology#team design#ai operations#hubspot

Explainer13:00

Why Educational Intent Is More Than 10 Times Larger Than Buying Intent

HubSpot distinguishes people ready to buy software from those who want to try it and those merely seeking education. Educational demand is much larger, so the company uses nurturing emails to move these contacts toward product trials or purchases.

Buying intent represents a limited audience
Free products address people interested in trying software
Educational content reaches the broadest cohort
Nurturing connects educational interest to later commercial intent

“There's at least 10x more people in your market out there learning than they are evaluating right now.”
Kip Bodner · 14:30

“And that nurturing path we call first conversion nurturing. And it's an email flow that we have here at HubSpot.”
Emmy Johnson · 15:00

#demand generation#lead nurturing#buyer intent#content marketing

Explainer19:30

Inside HubSpot's AI-Personalized Email Pipeline

When a lead submits a form, HubSpot combines the company website, firmographic information, downloaded offer, and site behavior into a summary of the person's likely objective. An LLM imagines the ideal course, a database retrieves similar real courses, and the LLM selects and explains the best available recommendation in a personalized email.

Scrape the submitted business website
Combine firmographic, conversion, and behavioral data
Infer the recipient's likely objective
Describe the ideal resource without limiting the model to the catalog
Retrieve similar real courses from the content database
Select and explain the best match in a personalized email

“We put together a summary of what we think they are trying to accomplish based on all of those things.”
Emmy Johnson · 21:00

“I think this course right here is the best course for this person to help them do the job they want to do.”
Emmy Johnson · 22:00

#email automation#llm#recommendation engine#content matching#gpt-4

Tool· 1

Tool05:30

The Simple Intake Stack Behind HubSpot's AI Experiments

HubSpot solicited AI experiment ideas through Slack and collected them with a Google Form. The deliberately basic process let the team move quickly rather than spending time designing a perfect intake system.

Use existing communication channels to solicit ideas
Capture submissions in a lightweight structured form
Optimize the intake process for speed rather than polish

“And I think that we collected them just through a message over Slack to the marketing team. And then I think we collected everything through…”
Emmy Johnson · 06:00

“You wanted to move quickly. You did a basic way of collecting ideas, and you wanted to move as fast as possible.”
Kip Bodner · 06:00

#slack#google forms#idea intake#ai experiments

Takeaway· 3

Takeaway06:00

Why HubSpot Revisited AI Priorities Every Two Weeks

The team retained a recurring prioritization meeting after its initial ranking exercise. Reviewing the portfolio every two weeks allowed it to respond quickly when technologies or market conditions changed.

Treat AI priorities as a living portfolio
Review the ranking every two weeks
Allow new technologies and market changes to alter the order

“And actually, after we did the initial stack ranking, we decided to continue to keep these prioritization meetings every two weeks.”
Emmy Johnson · 06:00

“We wanted to have the ability to agily switch which use cases we were prioritizing so we could take advantage or respond to that macro…”
Emmy Johnson · 06:30

#agile marketing#portfolio management#ai strategy

Takeaway23:00

The AI Email Test That Increased Conversion by 82%

HubSpot reported major gains after replacing cohort-based nurturing with individually inferred content recommendations. Conversion rate improved by more than 80%, click-through rate by more than 50%, and open rate by nearly 30%, with the engagement gains suggesting that recipients also found the emails more useful.

Conversion rate improved by more than 80%
Click-through rate improved by more than 50%
Open rate improved by nearly 30%
Higher engagement indicated greater recipient value

“Our conversion rate increased by 82%. Huge step function gain.”
Emmy Johnson · 23:00

“Open rate, I think was close to 30%, maybe just a hair under.”
Kip Bodner · 23:30

#conversion rate#email marketing#experiment results#personalization

Takeaway32:00

The Two-Person Partnership Behind HubSpot's AI Email System

HubSpot paired an AI specialist with an email automation specialist who deeply understood nurturing flows and customer personas. The combination connected technical model development with the domain knowledge needed to judge recommendations and operationalize the experience.

Assign an AI specialist to the technical implementation
Pair that person with a subject-matter expert
Use domain knowledge to guide personas and automation
Build the system collaboratively rather than isolating AI work

“Pairing the two together has also been, I think, yeah, really important.”
Emmy Johnson · 32:30

“So you have a marketing uh someone who is just a subject matter expert in the automation work, the persona, the email. And then you…”
Emmy Johnson · 32:30

#cross-functional teams#ai implementation#email automation#domain expertise

Frameworks in this episode

AI Expert and Domain Expert Pairing

AI Use-Case Prioritization Two-by-Two

Buy-Try-Learn Intent Ladder

Job-to-Be-Done Personalization Pipeline

Ship-Train-Iterate AI Loop

Insights & moments

⚡Myth Buster· 2

Why Traditional Segmentation Is Not Truly Personal Email