Sports Personalization Starts Before The Start Line

Personalization has become a revenue engine in sports, and most of the thinking is built around fans: the right highlight, the right offer, the right seat. Participation organizations have something richer to personalize around, which is the participant's own progress. Our approach starts from intent rather than segments, keeps a slow participant profile separate from a fast activation layer, and ships the plays today's data can support while the platform work catches up.

Personalization is now an expectation, not a feature

In IBM’s 2025 survey of more than 20,000 sports fans across 12 countries, 30% identified personalized content as a top priority for AI. These are fan expectations, not direct evidence of participant behavior.

In pro sports, that has turned into AI-adapted emails that change content and tone for each fan, real-time offers triggered by audience segments, and personalized highlight reels produced on request. The common thread is that the organization knows enough about each fan to make the next message feel chosen rather than broadcast.

Participation has richer signals than fandom

A fan's profile is mostly preferences: favorite team, favorite player, where they sit. A participant's profile is a story with dates. They registered on a specific day, for a specific distance. They trained, or didn't. They finished in a time, or deferred. They came back, upgraded to a longer race, or lapsed. In certification and membership models, they move up levels, renew or let their status expire. If you are designing those status journeys, start with participation sports membership models that sell the year, not the race.

Those signals provide lifecycle context that preferences alone cannot. Their predictive value needs to be tested against your own participant cohorts. Organizers already hold many of them, but in different places.

The window to use them is also unusually long. In the first half of 2026, 16.4% of US race registrations came more than 120 days before race day, and 55.4% of marathon registrations did. That's months in which the organizer has the participant's money and attention. Today most of that time is filled with generic reminder emails.

Start with intent, not segments

Most personalization programs start by building segments. We start one step earlier, with intent: what the person is trying to do right now, and how long that will stay true. Intent comes in six kinds, and each one fades at a different speed.

Intent	What it looks like in participation	How fast it fades
Explicit	A search for "first half marathon", a distance filter, a form answer	Hours to days
Implicit	Three visits to beginner training plans, time spent on one course page	Hours to weeks
Next step	What the organizer can infer: after a first 10K, a half; after a sprint triathlon, an Olympic distance	Tied to the lifecycle
Upstream	The search term, social post, AI assistant answer or partner link that brought them in	One session
External	A heat wave, a course change, a travel disruption	Days to weeks
Aspirational	The first-timer who could be coaching others in five years	Long horizon

When several are present at once, the order matters. Honor explicit intent first, then implicit, then the upstream context, then the inferred next step as the default. External signals mostly suppress rather than promote. Aspiration stays in the background and never overrides what the person asked for.

Two dimensions do most of the work in participation: where someone is and where they are in their progression. A runner in a northern city who just finished a summer 10K needs something different from one in a warm climate who races year-round. Everything else modifies those two.

Separate what you believe from what you show

The most important design decision in a personalization program is to keep two things apart. The participant profile is slow: it records what you believe is true, and it changes only on verified events such as a finish, a purchase or a renewal. The activation layer is fast: it decides what to show in this session, based on what is happening right now.

Real-time signals change the surface for a window. They never rewrite the profile.

That split keeps the profile trustworthy. A heat warning in race week should pause the race-day hype on the site for a few days. It shouldn't change anything you know about the runner. Every decision the engine makes then comes down to three moves: route the person somewhere better, foreground the right thing in place, or suppress what competes with it. In practice that means one primary action and one secondary per page, instead of the five competing calls to action most event sites show today.

Build for the anonymous majority first

Most visitors to an event site aren't signed in, and a growing share block cookies or browse privately. A personalization program that only works for known participants leaves most of the audience with a generic site. We design the baseline first, using location, language, season and entry page, then layer sharper targeting on top for the smaller group of signed-in, high-intent visitors. The baseline has to work when you know nothing.

We also measure movement, not only conversion. Many visitors aren't ready to register today. Return visits, deeper reading and a saved training plan are the leading indicators that they will.

Where personalization pays in participation

We think about four moments, each with its own question:

Before registration: which event is right for me? Distance, location, date and difficulty matter more than brand messaging. Recommending the right event to the right person is personalization at its most commercial.

Between registration and race day: am I ready? Training content, course information, travel and kit advice, timed to how far away the race is and how experienced the participant is.

Race week: what do I need to know right now? Start times, logistics, weather and changes, delivered to the people they affect.

After the finish: what's next? The moment of highest emotion and the moment most organizers go quiet. The next goal, the next distance, the result shared, the membership offered, ideally within hours of crossing the line, while the feeling is still there, not in an email a week later.

Timeline of a participant's year with four moments to personalize: before registration, up to race day, race week, after the finish

The longest window, from registration to race day, often runs 120 days or more. The moment after the finish is where most organizers go quiet.

There's money attached to each moment. In the US, 13.1% of registrants buy add-ons at registration, email drives about 12% of registration revenue, and events with peer-to-peer fundraising raise eight times more than donation-only races. These are platform-level observations, not causal evidence that personalization produced them; use a holdout to test the effect in your own program.

There's also a partner moment that fan models miss. When a training center, a club or a coach owns the next step, personalization means routing the participant to the right local partner with real availability, not sending them to a generic store locator. The same partner moments are where participation sports sponsorship moves from logos to verified audience segments.

Not every play needs the same data, and that's the part most roadmaps miss. We sort lifecycle plays by what they depend on. A completion celebration, an anniversary of someone's first finish, a training check-in and a nudge toward a local club can run on data most organizers already hold.

Next-step recommendations, lapsed-participant reactivation and churn-aware renewals need one unified profile. Anything that quotes savings or lifetime spend needs commerce data on one system of record. And a few plays, such as training groups or reasons people drop out, need data nobody captures yet.

Ship the first column while the platform work behind the other three runs.

What gets in the way

In our experience, personalization programs rarely fail because of the tools. They fail because of the data.

The same person exists several times over, with a different email address in the registration platform, the shop and the training app. Consent is captured differently in each system. Content is written as pages rather than structured pieces, so there's nothing to swap in or out for different audiences. And the team that owns the website, the team that owns email and the team that owns the CRM work from different calendars.

Ownership is the other gap. We often find two or three identity efforts running in parallel: one inside a CRM program, one in a data warehouse project, one proposed by a platform vendor. Each is reasonable on its own. Together they leave nobody accountable for the participant profile, and the path from raw data to a usable profile stays unclear. Commerce is frequently mid-migration too, which means any play built on purchase history has to wait for the new system of record.

The profile in the middle does the work. Every activation channel is only as good as the identity and consent behind it. That identity layer is what a participant data platform built as a loop is designed to produce.

Label AI-generated content and assistants clearly. Participants should know when they are interacting with AI, what information it uses, and when a person takes over.

What leaders should do now

Run an intent mapping session. List every explicit, implicit, upstream and external signal your stack already sees, and decide which layer acts on it. Then map what arrives from search, social, AI assistants and partner sites, and what gets lost on the way in.

Name one owner for the participant profile, and agree the path from source systems to warehouse to activation before you buy more tooling. Whether the activation layer is HubSpot, Salesforce Marketing Cloud or another platform matters less than whether it's fed by one clean profile with consent attached. See how we run journey orchestration across channels.

Measure how people actually progress. A cohort analysis of the time from first event to second, and from one distance or level to the next, replaces guesswork with a lifecycle you can design for. Then ship the anonymous baseline and the plays today's data supports, and structure your content into modules a platform can reassemble.

Measure with holdouts. Keep a small group that gets the generic experience, so you know what personalization is actually adding, and track forward movement alongside conversion.

The road ahead

Analysts expect 60% of brands to use agentic AI for one-to-one interactions by 2028. In participation sports, that will mean training nudges, race-week assistants and next-event recommendations generated for each person. Before those assistants go live, settle the guardrails for AI agents in participation sports.

The organizations that benefit will be the ones that did the unglamorous work first: one record, clear consent and content built to be reassembled. Personalization isn't a campaign. It's what happens when the data finally agrees about who someone is.

Start with a digital estate diagnostic: identify the participant signals you hold, who owns them, and the first lifecycle moment worth testing. See how this applies across adventure and endurance sports organizations.