NFL audience

Role

Innovation and Strategy (AAF MSU)

Highlights

Designed and led quantitative and qualitative analysis. Introduced AI tools to the team. Built the synthetic persona infrastructure used throughout the campaign.

Tools

Qualtrics. Claude. Python. Streamlit. Figma. NotebookLM

Duration

5 months

[data-wf-bgvideo-fallback-img] { display: none; } @media (prefers-reduced-motion: reduce) { [data-wf-bgvideo-fallback-img] { position: absolute; z-index: -100; display: inline-block; height: 100%; width: 100%; object-fit: cover; } }

NFL Audience Design · Part 1

AAF NSAC 2026 · MSU · Innovation & Strategy Lead

Audience
Design

A solid advertising campaign starts with knowing who you're talking to. This project set the strategic foundation for an NFL campaign by treating audience definition as both a creative and scientific problem, triangulating qualitative insight, quantitative modeling, and machine learning to turn a brief term into three precise behavioral profiles a creative team could write to and a media team could target.

Survey responses 0 cleaned & analyzed

In-depth interviews 0 qual & quant coded

Synthetic personas 0 built to be talked to

Survey DesignOLS RegressionK-Means ClusteringDecision TreeQualtricsNotebookLMStreamlitClaude APIMachine LearningFigmaSynthetic PersonasFan SegmentationSurvey DesignOLS RegressionK-Means ClusteringDecision TreeQualtricsNotebookLMStreamlitClaude APIMachine LearningFigmaSynthetic PersonasFan Segmentation

The Brief Challenge

Who is a
casual fan?

The NFL brief named the target: casual fans. The research translated that into three precise definitions both teams could act on.

cas · u · al fan true be · liev · er

casual
fan

/ˈkaʒ(u)əl fæn/

n. as received

"Fans with lower engagement levels and potential for increased fandom. Individuals who have some connection to the NFL but are not yet deeply invested in the league or its programs."

Research output · encoded

k-means · n=934

Revised Entry

Three distinct behavioral groups inside this term.

Disengaged Fan

Not opposed to the NFL. It just hasn't found them, and there's no particular reason it would have. Not in the feed, the game's not on their radar, and a campaign about football will slide right past them. What might actually work is one about the friendships, the rituals, and the social world that's built up around the game without them noticing. That's a conversation they might stay in.

Engaged Skeptic Fan

There every Sunday, opinions about the broadcast team, and they could tell you which ads were clearly made by someone who'd never watched a game in their life. Fan level nearly 8. Institutional credibility barely 3. Fully in on the sport. The league is still on probation. A campaign that opens with how much the NFL cares about communities is going to get a raised eyebrow. Proof, not announcement, is what moves them.

True Believer Fan

Already sold. NFL_Cares at 7.28, Credibility at 6.88. Not someone who needs convincing. They'd sign the kids up for Play 60 tonight if the path were clear enough. It isn't, and that's a different kind of problem than not caring. Two kids and a Saturday that's already fully allocated, so the campaign's only job is to make the next step take thirty seconds.

Research Approach

Triangulation: three methods, each catching what the others miss. When the survey pointed somewhere, the interviews explained why. The focus groups tested whether those themes held up in a room.

1,437

Survey

1,437 responses gave the scale to model behavior, not just describe it. Every question was designed as a codeable variable, making regression, correlation, and clustering possible after collection.

In-Depth Interviews

51 interviews gave the numbers texture. When respondents explained their relationship with football in their own words, a finding became a brief. The qualitative layer is what turns data into direction.

Focus Groups

Seven focus groups surfaced something individual interviews cannot: how people perform their relationship with football when others are watching. Social dynamics change the answers. For Gen Alpha specifically, group settings under supervised conditions were also the more ethical research path, given the age of respondents.

How We Asked

A survey designed
with intention.

Defining the audience is the first creative act. The survey was designed with every question as a codeable ordinal variable, built to model behavior, not just count it.

Questions

Variables captured

Dimensions

935

Clean observations

Platform

Qualtrics professional survey

Timeline

Sep–Nov 2025

Target

Michigan State University community

Initial responses

1,437 · 935 after quality control

Target Variable

interest_level (1–10 scale)

NFL Perception & Values

7 Likert scales

Community care, player credibility, social issues, health support, values alignment

Content Consumption

2 ordinal scales

Weekly frequency, likelihood to watch next game

Personal Connection

3 binary

Favorite team, favorite player, neither

Program Awareness

7 binary

Play 60, NFL Flag, Inspire Change, and related initiatives

Engagement Drivers

9 binary

Content preferences: inspirational, youth, community, highlights, etc.

Demographics

11 variables

Age, gender, race, geography, parental status

In-Depth Interviews · 51

"You can just tell when a brand is genuine. Right off the rip. You can tell they don't really know the audience they're trying to reach."

Maya, 22

"I don't pay attention to their messages. You watch the game and you get over it. I don't know what they stand for."

Jordan, 21

"I would be more involved if it appealed more to females. But it really doesn't. I feel like it's not geared towards females."

Marie, 54

NotebookLMFigmaQual Coding

Focus Groups · 7

Focus groups surfaced something individual interviews cannot: how people perform their relationship with football when others are in the room. Social dynamics shift the answers in ways that make the findings richer and more honest.

Gen Alpha · Methodological Note

In-depth interviews with respondents under 13 required parental consent and supervision per ethical research guidelines. Primary qualitative data was supplemented with secondary research from Mintel and MRI-Simmons to validate Gen Alpha media consumption and youth sports engagement patterns.

Mintel MRI-Simmons Supervised IDIs

What the Data Said

The gap isn't
in their heads.

Four analytical layers. The most useful thing they produced: a contradiction. Women don't distrust the NFL more than men. They just see it less. That changed the brief, the media strategy, and ultimately the campaign concept.

Distribution by Generation

Gen Z

434

Millennial

219

Gen X

109

Boomer+

Distribution by Gender

Male

389

Female

529

Young, female-skewed, and racially homogeneous.

Gen Z, Millennial, and female fan signal is solid. Worth naming: the sample skews white, so conclusions about Black and Hispanic fan experiences need additional validation. That transparency is more useful than pretending the gaps aren't there.

→ Reliable signal for Gen Z and female targeting

Correlation Matrix · Key Variables

Age

Fan Level

NFL Cares

Credibility

Content Freq

Age

1.00

0.09

0.03

0.06

−0.17

Fan Level

0.09

1.00

0.50

0.39

0.59

NFL Cares

0.03

0.50

1.00

0.54

0.29

Credibility

0.06

0.39

0.54

1.00

0.24

Content Freq

−0.17

0.59

0.29

0.24

1.00

Content frequency is the linchpin variable.

The strongest predictor of fan engagement: how often someone sees NFL content in their feed. r=0.59, highest in the matrix. Trust metrics also compound: lose credibility in one area and adjacent variables drop with it.

→ More touchpoints = higher engagement

The gender gap is behavioral, not perceptual.

The instinct is a trust campaign. Convince women the NFL cares. The regression disagrees: women don't distrust the NFL more than men. They just see it less. That's a distribution problem, not a persuasion problem. You're not changing minds. You're getting into the feed.

→ Meet them where they are

OLS Regression · Women's NFL Interest (n=454)

59.7%

R²: model explains 59.7% of variance

What drives women's interest in the NFL?

Watchability of content

39%

NFL community care belief

24%

Content frequency in feed

23%

Player credibility

14%

Decision Tree · High-Intent Fan Identification

Golden Cohort Criteria

Fan_Level > 9.5
Age ≤ 27.5
Content_Freq > 4.5

70.7%

High-intent probability · n=75

Low content frequency collapses high-intent probability to 32.1% regardless of fan level or age. Passion without exposure does not convert. Feed presence matters more than enthusiasm, useful when allocating media budget.

→ Feed frequency predicts conversion

Impact Simulation · What drives casual fan growth?

Baseline

5.76

+ Trust Boost

5.86

+ NFL Cares

6.02

+ Content Freq

6.32

+ Attention ↑

7.15

Predicted fan level · Scale 1–10

Attention is the multiplier.

Attention is the highest-return variable. A one-point increase predicts fan level 7.15, vs. 5.86 for trust-first approaches. Earn attention first. Nobody changes their mind about content they've already scrolled past.

→ Lead with entertainment. Embed purpose.

Barrier ROI Analysis

▲ +0.003 Hard to Watch (access)
▲ +0.003 Rules Confusing (education)
▼ −0.004 Cost (price sensitivity)
▼ −0.005 No Time (scheduling)

Invest in accessibility and education. Price sensitivity is a stated rather than actual barrier for most respondents.

NFL Audience Design · Part 2

Team Exploration · ML Quiz

Can machine learning
classify a fan?

During the exploratory phase, before strategy had crystallized, I built a quiz as a shared reference point that aligned how the whole team thought about the audience.

Central Question

Can machine learning help classify fan engagement levels to enable personalized program messaging?

ML Pipeline

Data Collection

→

Data Cleaning

→

Schema Validation

↓

Feature Engineering

←

Train-Test Split

←

Normalization

↓

Model Evaluation

←

Model Building

←

Feature Selection RFE

↓

FastAPI

→

Streamlit

→

Final Web App UI

Python · Scikit-learn FastAPI Streamlit

Live Demo · nflfantype.streamlit.app

Try the live quiz

From Clusters to Characters

They weren't invented.
They were found.

K-means clustering showed that the casual fan cohort is not one thing. It's three distinct groups living under the same label, each with a different relationship to the NFL and a different argument that might move them. The clusters came first. The personas were named from what the data showed.

Step 1 · What the data found

Cluster	Fan Level ÷10	NFL Cares ÷10	Credibility ÷10	Content Freq ÷5	Attention ÷5	% of Casuals
True Believers	8.64	7.28	6.88	3.90	3.23	19%
Engaged Skeptics	7.91	4.67	3.64	3.97	2.98	35%
Disengaged	3.83	3.83	3.59	1.94	1.88	46%

Dark green = strong signal · Dark red = weak signal · K-means clustering on casual fan cohort (n=934)

Step 2 · From behavioral signature to person

Each cluster had a behavioral signature. The question: which real person does this profile describe, what frustrates them, and what would actually move them?

Cluster · True Believers

19%

of casual fans

Fan Level

8.64

NFL Cares

7.28

Credibility

6.88

Content Freq

3.90

This cluster became

Couch Coordinator

NFL_Cares 7.28, Credibility 6.88. This cluster's relationship with the NFL is sorted. Season ticket holder, two kids in FLAG. Nothing to convince him of. What he needs is sign-up in under ninety seconds. Saturday is fully spoken for and any form longer than that gets closed.

Cluster · Engaged Skeptics

35%

of casual fans

Fan Level

7.91

Content Freq

3.97

NFL Cares

4.67

Credibility

3.64

This cluster became

GDG

Game Day Girly

Fan level nearly 8, content frequency near 4, but credibility at 3.64 and NFL_Cares at 4.67. In on the game. Still deciding about the league. A campaign opening with NFL community commitments gets a polite raised eyebrow. What moves her is unproduced, specific proof, which is also the hardest kind to fake.

Cluster · Disengaged

46%

of casual fans

Fan Level

3.83

NFL Cares

3.83

Content Freq

1.94

Attention

1.88

This cluster became

Tiny Teammate

Content frequency 1.94, attention 1.88, floor of the cohort, got there by accident. Plays FLAG because her friends do. Physically inside the NFL ecosystem. No relationship with its content. A football campaign speaks directly past her. The conversation that lands is about her team, her friends, and the fact that they're all doing this together.

Step 3 · The personas in full

Hover or tap to reveal each persona

↑ True Believers cluster

Couch Coordinator · 40 yrs · Gen X

Lions season ticket holder since '08. Two kids in NFL FLAG. Wants programs to work, not to be sold.

Proud DadSunday FanPragmatist

He'd do it. He's not resistant, he's just busy. Make the path short enough to complete from his couch during the Lions game and he's in.

↑ Engaged Skeptics cluster

GDG

Game Day Girly · 22 yrs · Gen Z

Came to football through tailgate culture. Hosts game day every Sunday. Has very specific opinions about the halftime show.

Tailgate PlannerAuthenticatorValues-Driven

She's got a sixth sense for when something's been approved by a committee. NFL messaging that reads like marketing copy gets scrolled past in under a second.

↑ Disengaged cluster

Tiny Teammate · 12 yrs · Gen Alpha

Didn't care about football. Joined NFL FLAG to hang with friends. Loves it now because of the team, not the game.

Social TeammateFootball CuriousMom Decides

Sports content just isn't written for her. The door in is the friendships, not the game.

Audience Design as Infrastructure

Personas you can
talk to.

Most agencies put personas on a slide and leave them there. These are AI-powered and grounded in all three layers of triangulated research. Drop in a tagline, upload a concept, describe an activation and they respond the way the actual audience would.

Focus Group Recording

NFL Creative Lab · Synthetic Focus Group · Live · Web-Grounded

🧒🏽 Tiny Teammate

👩🏾 Game Day Girly

👨🏻 Couch Coordinator

What do you think of this poster? Kids playing flag football under stadium lights.

🧒🏽

TINY TEAMMATEooh the lights are sooo cool
but like why is everyone so serious
my team is always laughing and being silly
if they were actually having fun i would show Ava this fr 😭

What if we added a trending sound?

🧒🏽

TINY TEAMMATEWAIT YES
if it had that sound everyone is using rn
idk what its called but u know the one
my mom would still have to approve it tho lol

Does this tagline work? "Everyone belongs in the huddle."

👩🏾

GAME DAY GIRLYok wait
this is actually giving something
"huddle" is specific enough to feel real but open enough to not require football knowledge
ngl if this had the right casting behind it i would screenshot it

Who would you want to see in it?

👩🏾

GAME DAY GIRLYreal people not models 💀
the girl doing tailgate videos, the dad at his kid's first game
shoot it on an iphone in natural light and it's a whole different thing
the produced look is what makes it feel fake fr

Here's the Play 60 program overview: two kids running drills. Would you sign your kid up?

👨🏻

COUCH COORDINATOROk this is clean. I get what it's saying. But where do I actually go? That's the piece that's missing. A dad sees this during halftime, thinks "huh, maybe." Then forgets by the third quarter if there's no next step.

What would make you actually click or call?

👨🏻

COUCH COORDINATORShow me a real coach. Not an actor. Also: is it on Saturdays? I already got soccer. If it fits the schedule I'd actually talk to my wife about it. That's how it works.

Sample exchanges · AI-powered with web search · Grounded in 934 survey responses + 51 interviews

Protected portfolio

Enter
Access

This case study is password protected.
Enter the key to continue.

Incorrect password. Try again.

← alvanguyen.com

NFL audience

Audience
Design

Who is a
casual fan?

A survey designed
with intention.

The gap isn't
in their heads.

What drives women's interest in the NFL?

Can machine learning
classify a fan?

They weren't invented.
They were found.

Personas you can
talk to.

the insights

THE IDEA

STATIC

experientials

Commercial

AD TESTING

THE AAF MSU CREW

the insights

THE IDEA

BRAND DESIGN

STATICS

BRANDING MERCH

OUT-OF-HOME

SOCIAL

EXPERIENTIAL

the insights

STATICS

ACTIVATIONS

Branding

inspiration

TYPOGRAPHY

Color palette

Logo Process

MOBILE APP

KEY VISUALS

BOOKMARK ORIGAMI

AR EXPERIENCE

print ad

BREWING IDEAS...

NFL audience

AudienceDesign

Who is acasual fan?

A survey designedwith intention.

The gap isn'tin their heads.

What drives women's interest in the NFL?

Can machine learningclassify a fan?

They weren't invented.They were found.

Personas you cantalk to.

the insights

THE IDEA

STATIC

experientials

Commercial

AD TESTING

THE AAF MSU CREW

the insights

THE IDEA

BRAND DESIGN

STATICS

BRANDING MERCH

OUT-OF-HOME

SOCIAL

EXPERIENTIAL

the insights

STATICS

ACTIVATIONS

Branding

inspiration

TYPOGRAPHY

Color palette

Logo Process

MOBILE APP

KEY VISUALS

BOOKMARK ORIGAMI

AR EXPERIENCE

print ad

BREWING IDEAS...

Audience
Design

Who is a
casual fan?

A survey designed
with intention.

The gap isn't
in their heads.

Can machine learning
classify a fan?

They weren't invented.
They were found.

Personas you can
talk to.