EDGE3/Intelligence Loop/Social Listening
EDGE3 Methodology

The always-on input. Drift caught between survey cycles.

Survey, parent, weekly snaps — those are scheduled inputs. Social listening runs continuously in the background, monitoring only what the athlete chose to make public, on accounts they self-reported. It's the layer that catches drift before a coach knows there's a problem.

What it captures

Four signal types. Each named in plain language. The model handles the math. Coaches see the read.

Signal 01

Follower-graph drift

Who is following the athlete — and who the athlete is following back.

  • A coach at another school starts following one of your players. Recruiting probe — immediate flag.
  • The player follows a rival program. Soft signal of curiosity.
  • The player unfollows your program or HC. Hard signal — immediate alert.
  • A teammate unfollows the player. Locker-room friction signal.
Most predictive signal in the layer. Rival recruiting interest shows up here weeks before any commitment language.
Signal 02

Engagement pattern

What the athlete is engaging with publicly — likes, replies, shares.

  • The player likes posts from someone who consistently criticizes your program.
  • The player replies positively to a transfer-portal account.
  • The player engages with a different school's NIL collective.
Pattern over event. Only alerts when a 14-day pattern crosses threshold.
Signal 03

Search-trend velocity

Public search interest in the player's name. Not from the player — from the recruiting market around the player.

  • Spike in Google searches for the player's name — other staffs doing their homework.
  • Cross-correlated with media coverage, rival-staff follows, recruiting cycles to filter noise.
Tells coaches when the recruiting market is heating up around their player — usually before the player knows.
Signal 04

Frustration markers

The cryptic-post layer. Plain-language patterns that indicate disengagement.

  • Cryptic posts about playing time, role, or staff decisions.
  • Direct or implied transfer-portal language.
  • Subtweet patterns toward teammates or staff.
  • Tone shift from team-positive to team-neutral over 14 days.
Only flags when 3+ markers land within 30 days. Single posts don't trigger alerts.

What it does NOT do

The compliance posture, in plain language. Listed first because every AD asks.

Not part of EDGE3 social listening
  • No DMs or private messages — public posts only
  • No posting on the athlete's behalf — read-only
  • No facial recognition. No image classification. No audio.
  • No browsing history — only what's posted publicly
  • No data shared outside the athlete's coaching staff
  • No NIL brand-deal velocity tracking — we don't measure it

How it shows up for the coach

No social platform login. No raw scores. Plain-English insights in the EDGE3 dashboard.

Coach dashboard · example
What changed · last 7 days
A coach at TCU started following the player. The player liked three posts from a fan account that's been calling out the offensive scheme. Cryptic post on Sunday about "playing time."
What it means
Recruiting interest from outside is heating up while the player is showing public frustration with the offense. Combined risk score moved from 0.42 to 0.61 — Yellow.
What to do
15-minute check-in this week. Anchor on the route-tree conversation from the October survey response — the player told you system fit was the reason for choosing the program. Reaffirm the development plan.
No AI metrics. No threshold scores. No technical jargon. Just what changed → what it means → what to do.

Cadence

Three layers running in parallel. The model re-scores the moment any signal threshold is crossed.

Continuous
Public-post + follower sweep
Every 6 hours. New posts, engagement events, follower-graph deltas pulled from self-reported accounts.
Daily
Pattern scoring + threshold check
Search-trend velocity pulled. Four signal types computed against rolling baseline. Model re-scores if any threshold crosses.
On detection
Coach push notification
Coach notified within hours of a hard signal — rival coach follow, unfollow event, agent contact.

How it feeds the models

Each signal maps to a feature column already wired into the model architecture. Coaches don't see this — it's the math under the hood.

Signal
Feature column
Models affected
Rival coach follows player
social_external_pressure (heavy weight)
Roster Retention Risk
Player unfollows program / HC
social_sentiment + social_external_pressure
Roster Retention Risk + Trajectory
Engaging with critic accounts
social_sentiment (negative)
Performance Trajectory
Search-trend spike on player
social_external_pressure
Roster Retention Risk
Frustration markers (3+/30d)
social_frustration_flag
Roster Retention Risk · immediate Yellow trigger
New agent in influence chain
social_external_pressure
Roster Retention Risk

What we promise the coach

EDGE3 doesn’t show coaches more data. It shows coaches what changed, what it means, and what to do — and only when there’s something worth knowing. Social listening is the layer that makes that possible without the staff opening a single social media app.

Companion docs

Where this layer connects to the rest of EDGE3.

Methodology
Continuous Intelligence Loop
The four input streams together — survey, parent, snaps, social. →
Methodology
Survey Scoring
Response → dimension → model feature column. →
Roadmap
Next Steps
Phase 2 priorities for the social-listening layer. →