All work
Confidential client · AI / MarTech

AI Customer Profile Prediction

An AI-powered MarTech tool that turns complex user data into marketing segments teams can actually act on, and trust.

Profile prediction dashboard with projected value distributions and sample user cards
The shipped product: marketers define an attribute, review predictions one at a time, and project segment size across the whole user base before applying it.
RoleProduct Strategist & Design Lead
TeamJunior UX/UI, PM, AI Researcher, Devs
ClientIndustry-leading AI MarTech company
FocusAI UX, trust, mentorship
faster audience segmentation vs. manual workflows
+75%improvement in targeting outcomes
+40%adoption of AI-assisted workflows in pilots
Days→mintime-to-segmentation
01Context

The opportunity, and the catch

This was built at a large AI MarTech company serving enterprise marketing teams.

Marketers want to target people by real behavior, preference, and intent. Doing that by hand across millions of visitors isn't realistic.

AI can close that gap, but most tools traded one problem for another: marketers couldn't tell whether the output was worth trusting.

  • Opaque: the predictions were hard to explain
  • Unverifiable: no clear way to test, correct, or override them
  • Risky: expensive to act on at scale without confidence in the accuracy
02Problem

A "look what we can do" demo, not yet a product

Before I joined, the team had built a simulation that generated live attribute predictions, demographics, hobbies, brand preferences, off a tester's own browsing.

The tech was genuinely impressive. The use case wasn't.

  • The insights came from biased simulations, not real customer behavior
  • There was no workflow beyond showing off what the model could do
  • The output was interesting, but not something a marketer could act on with confidence

It needed to become an actual marketing tool, not a tech demo.

03Goal

Build a workflow that lets marketers define an attribute, review and correct the model's predictions, see how each one was reached, and then apply it across a large audience with confidence.

The point wasn't just faster segmentation. It was earning enough trust that marketers would actually rely on AI to make a call.

04How it works

A human-in-the-loop prediction workflow

01

Define the attribute

Marketers start by defining what they want to learn, say, which TV display technology a user is most likely to prefer.

Define attribute screen: naming the attribute, setting values, and previewing predictions
Step 1: define the attribute and its expected values, with a way to preview predictions before committing.
02

Review sample predictions

They review a set of real user profiles with predicted values, confidence scores, and plain-language explanations. This is where they test accuracy, give feedback, and tune the model before it goes wide.

Sample prediction cards showing predicted attribute, confidence score, explanation, and thumbs feedback
Step 2: one card at a time: the predicted value, a confidence score, a plain-language why, and a thumbs up/down that feeds back into the model.
03

View distribution and apply

Once the logic is tuned, the system projects the attribute across the full user base. Marketers can size the segment and apply it to a campaign, like promoting OLED TVs to the people most likely to want them.

Site-wide projected value distribution across the full user base, ready to apply as a segment
Step 3: the tuned attribute projected across the whole user base, then applied directly as a targetable segment.
05Approach

Redefine the problem before designing any screens

Redefine the problem

I pushed back on the assumption that "simulation" was the experience. Through workshops and cross-functional sessions, I helped steer the team toward generating attributes from real behavioral data instead.

Create strategic value

That reframe turned a stalled sandbox feature into a real workflow: define an attribute, tune the prediction, segment the audience.

Align and mentor the team

I got PM, engineering, AI research, and UX pointed at one product direction, and mentored the junior designers through IA, interaction patterns, and stakeholder conversations.

End-to-end flow from tag creation and prediction refinement to site-wide projection
The workflow I mapped: tag creation and prediction tuning on the left, site-wide projection and confident application on the right.
06UX challenges

Designing for trust in AI

Challenge 01

Making the workflow easy to pick up

I streamlined attribute creation, automated value generation, added CSV upload, and paired setup with live prediction examples, so the workflow taught itself as marketers moved through it.

Iteration on the side-panel layout for defining an attribute and previewing predictions
One of several patterns we iterated on with the junior designers: how the define-and-preview side panel walks a marketer through the flow.
Challenge 02

Getting marketers to trust the predictions

I surfaced confidence scores, explanations, and live previews so marketers could see how a prediction was reached before betting a campaign on it.

Real-time tagging extension predicting attributes live as a user browses a retail site
The live preview: a real-time panel predicts attributes as a user browses, so marketers can watch the model reason on a real site before trusting it at scale.
Challenge 03

Getting feedback without making it feel like work

I cut the overload by showing one card at a time, and kept the feedback actions light, so marketers could improve accuracy without it becoming a chore.

Before and after of the review flow: the original dense multi-card grid next to the redesigned one-card-at-a-time review with plain-language explanations
Before and after: the original flow reviewed predictions in a dense grid; the redesign walks through one card at a time with a plain-language why and one-tap feedback.
07Assumptions tested
  • What makes a marketer trust a prediction on first look?
  • Which data points actually help them verify accuracy?
  • Random previews or filtered demographic ones, which is more useful for evaluation?
08Results
  • Built trust in the AI tagging workflow
  • Cut time-to-segmentation from days to minutes
  • Delivered a system that's flexible, scalable, and auditable
  • Helped marketers apply AI insights with more confidence and control
Takeaway

Trust in AI takes more than accurate predictions. It needs transparency, control, and a clear way for a person to check what the model is doing.

Keeping marketers in the loop didn't just make it more accurate. It made the whole thing easier to trust, adopt, and scale.

Next project
Kidsync