An Instant Athlete Brand Audit Scores You in 30 Seconds

Cam Hazzard won the DUNKMAN Group A qualifier and pulled 120 million views across platforms in a few weeks. On our last call he asked one fair question: can we actually show a young athlete their brand score on the spot, or is it a slide deck? So we built the thing that answers it — live, at a public URL, in about 30 seconds — and this is the honest write-up of how.

1. The task

Build the front door of theathletespotlight.com: a free, instant personal-brand audit for a young athlete. Enter your name, your sport, your socials, upload a photo or two, and get a real score for the six things a sponsor or recruiter judges when they Google you — plus a rendered preview of what your own website could look like, and whether yourname.com is still available. If they want the real thing, it’s $30 a month and we build it. Cam is the figurehead; the athletes he’ll send become the proof.

It’s a findability engine, not a website builder

The audit is the front door to something bigger. The goal isn’t a pretty page — it’s making a young athlete findable everywhere someone checks: Google, ChatGPT and the other AI answers, and the quiet search a sponsor’s marketing manager runs before they reply. So the program builds the site, but also the entity schema, the clips repurposed into rankable articles, and a contact path a sponsor will actually use.

The proof already exists. At Dunk Camp 2026 we scored 76 dunkers on this same rubric, and we’ve built the real thing for a growing roster who now own their names online — Cam Hazzard, Dylan Haugen, Hoopin’ Nate, Miles “McDeezy” McDonald, Travis Reynolds, and Poland’s Piotr “Grabo” Grabowski, with more each month. Dylan already has a full Google Knowledge Panel — the boxed profile Google shows beside his name — built from his site, schema, and repurposed videos by this program. It’s all one network: the dunkers live on Dunker Spotlight, which shares an engine with the wider Spotlight Network.

And every build ships with a working way for a sponsor to reach the athlete. We set the domain’s email (MX) records so a contact-form inquiry lands straight in the athlete’s inbox — a real lead, not a form that goes nowhere. That’s why the build flow has each athlete tell us exactly where those messages should go.

2. Step-by-step process

1. Read the room (25 min). Pulled the Aug 3 Zoom with Cam, the July 1 launch handoff, and the strategy doc. The brief was explicit: 50/50 of gross, $30/mo, tie it to Keap, keep the audit cheap, and — Cam’s words — make it look good, not like a wall of AI text.
2. Wrote the spec before the code (20 min). One document defining every endpoint, the scoring rubric, the sample-site themes, and the security rules, so the build agents had a single contract to hit.
3. Built the engine, then attacked it (parallel). Three components — the audit backend, the sample-site generator, the intake widget — each built and then handed to an adversarial reviewer whose only job was to break it. That pass earned its keep: it caught a bug where the search API silently returned nothing, which would have graded every athlete “invisible on Google.” Fixed before it ever shipped.
4. Made it visual (the second pass). The first version was correct but read like a report. So we added a photo pipeline — upload, store, serve — and rebuilt the sample sites as full-bleed, magazine-grade pages with the athlete’s face as the hero, in three distinct styles.
5. Deployed and ran a real audit (debugging included). Stood up the AWS plumbing, then hit three real infrastructure walls in a row and worked around each (see section 3).
6. Verified end to end. Ran a live audit on Cam and on two other athletes, confirmed the score, the email gate, the domain check, the sample renders, and the contest counter — then screenshotted the cinematic result.

3. Critical decision-making

The judgment calls a lesser build would have missed:

• When the model got blocked, we didn’t stop — we made it fall back. Halfway through, the AWS account’s access to the newest Claude models started flapping (an account-level form gate). Instead of waiting on a console fix, we made the scorer provider-aware: try Claude first, automatically fall back to a second model that was available. The audit never goes down; it upgrades itself when access returns.
• When public URLs were blocked, we used the front door that worked. The account blocks public Lambda URLs outright. Rather than fight it, we put the same API behind API Gateway — the exact pattern our dashboard already uses — with zero code changes.
• We kept the athlete honest and the tool honest. The scoring uses only what Google actually shows plus what the athlete enters. If the internet shows nothing for their name, the audit says so plainly — the same gap we documented by hand in our personal-brand audit of a dunker the internet can’t see, because that gap is the entire reason the product exists.
• We refused to email a stranger. A small one: the sample site shows an aspirational email address as text, not a live link, because the domain isn’t the athlete’s yet.

4. Effort and cost comparison

Honest read: a human team would produce something comparable, and probably make better isolated design choices in places. What they would not do cheaply is the build-then-break loop on every component, or the willingness to re-architect twice in one sitting when the environment fought back.

5. What the agent can and cannot do

• Did autonomously: wrote the spec, built and adversarially reviewed all three components, provisioned the AWS stack, debugged three infra failures, deployed, ran live audits, and rendered/QA’d the sample sites.
• Needed a human: publishing the audit page onto the WordPress site — the safety layer correctly gated every admin-login path, and no one was logged into that site. So the audit went live at its own hosted URL instead, and the WordPress embed is queued.
• Couldn’t do: clear the account-level model-access form (a console action) — hence the automatic fallback.

6. Information ingestion inventory

• Aug 3 Zoom recording summary, July 1 handoff email thread, the strategy-session document.
• Live data sources wired into the product: Google SERP (via DataForSEO), domain availability (via AWS Route 53), and model scoring (via AWS Bedrock).
• The existing personal-brand builder repo and the Spotlight Network registry for house style.

7. Guidelines compliance scorecard

• Title under 60 chars: yes. Meta description under 160: yes.
• Hook = a specific person in a specific situation (Cam, post–Group A): yes.
• Figurehead voice, active voice, short paragraphs, H2/H3 structure: yes.
• No AI fluff (“unleash”, “elevate”): checked.
• 2–3 internal BlitzMetrics links: done — Spotlight Network, Dollar a Day, and a personal-brand audit example, linked inline.

8. What’s next

Publish the audit at theathletespotlight.com/audit and link it from the homepage; flip the primary model back to the newest Claude once the account form is active; wire the $30/mo signup to trigger the real build; and run the September Cam-vs-Dennis contest — his organic reach against a dollar a day — with the scoreboard the tool already keeps.

Common questions

The quick answers athletes ask before they start — tucked down here so anyone who just wants their score isn’t slowed down.

By Dennis Yu · v1, 2026-08-08