Skip to main content

Personalized Ad Targeting - How It Works

Overview​

The ad targeting system analyzes user behavior, interests, and activity to show them the most relevant ads. Similar to Facebook/Google ads, it builds a comprehensive user interest profile and matches ads based on that profile.

How User Interests Are Determined​

1. Favorite Species (High Weight)​

  • Source: UserFavoriteSpecies table
  • Analysis: What fish species the user has marked as favorites
  • Example: User favorites "Bass" → Shows ads for bass fishing gear

2. Most Caught Species (High Weight)​

  • Source: CatchDetails from FishingLogEntries
  • Analysis: What species the user actually catches most often
  • Example: User catches 50 bass vs 5 trout → Bass-related ads prioritized

3. Gear Brand Preferences (Very High Weight)​

  • Source: GearSetups → Rod, Reel, Lure → Brand
  • Analysis: What brands the user actually uses in their setups
  • Example:
    • User has 3 setups with Shimano reels → Shows Shimano ads
    • User uses Rapala lures → Shows Rapala ads
    • User has St. Croix rods → Shows St. Croix ads

4. Fishing Methods (Medium Weight)​

  • Source: FishingLogEntry.FishingMethod
  • Analysis: What fishing methods the user uses most (Trolling, Casting, Fly Fishing, etc.)
  • Example: User mostly trolls → Shows trolling-specific gear ads

5. Location/Geography (Medium Weight)​

  • Source: UserProfile.HomeWater, UserProfile.Location, FishingLogEntry coordinates
  • Analysis: Where the user fishes most often
  • Example: User fishes Lake Michigan → Shows ads for Great Lakes fishing gear

6. Social Engagement (Low-Medium Weight)​

  • Source: PostLikes, Comments, PostHashtags
  • Analysis: What content the user engages with (hashtags, topics)
  • Example: User likes posts about "ice fishing" → Shows ice fishing gear ads

7. User Role & Activity Level (Low Weight)​

  • Source: User.Role, CharterCaptain, TournamentRegistrations
  • Analysis: Is user a charter captain? Tournament angler? How active?
  • Example: Charter captain → Shows commercial fishing gear ads

8. Marketplace Activity (Low Weight)​

  • Source: MarketplaceListings (purchased/sold items)
  • Analysis: What products they buy/sell
  • Example: User buys used rods → Shows rod ads

Relevance Scoring System​

When selecting an ad, the system calculates a relevance score (0-100):

Scoring Breakdown:​

  1. Species Match (30 points)

    • Do ad's target species match user's favorites/most caught?
    • Example: Ad targets "Bass", user catches bass → +30 points

  2. Brand Match (25 points)

    • Does ad's brand match brands user uses?
    • Example: Shimano ad, user uses Shimano gear → +25 points

  3. Fishing Method Match (20 points)

    • Does ad's target method match user's preferred methods?
    • Example: Ad for trolling gear, user trolls → +20 points

  4. Location Match (15 points)

    • Does ad's target location match where user fishes?
    • Example: Ad targets "Great Lakes", user fishes Lake Michigan → +15 points

  5. User Role Match (10 points)

    • Does ad target user's role (charter captain, tournament angler)?
    • Example: Ad for charter captains, user is charter captain → +10 points

Minimum Score: Ads need at least 30 points to be shown (prevents irrelevant ads)

Example: How Rapala Ad Gets Shown to User​

User Profile:​

  • Favorite Species: Bass, Walleye
  • Most Caught: Bass (45 catches), Walleye (12 catches)
  • Gear Brands: Shimano (reels), St. Croix (rods), Rapala (lures) ← Key!
  • Fishing Methods: Trolling, Casting
  • Location: Lake Michigan, Michigan
  • Role: Tournament Angler

Rapala Campaign:​

  • Target Species: "Bass, Walleye"
  • Brand: Rapala
  • Target Methods: "Trolling, Casting"
  • Target Locations: "US, Michigan, Great Lakes"
  • Target Roles: "TournamentAngler"

Matching Process:​

  1. Species Match: ✅ User catches bass/walleye, ad targets bass/walleye → +30 points
  2. Brand Match: ✅ User uses Rapala lures, ad is for Rapala → +25 points
  3. Method Match: ✅ User trolls/casts, ad targets trolling/casting → +20 points
  4. Location Match: ✅ User fishes Lake Michigan, ad targets Great Lakes → +15 points
  5. Role Match: ✅ User is tournament angler, ad targets tournament anglers → +10 points

Total Score: 100/100 → Perfect match! This ad will be prioritized.

Real-World Scenarios​

Scenario 1: New User (No Data)​

  • Profile: Empty (just signed up)
  • Ads Shown: Generic ads with no targeting, or ads targeting "All Users"
  • Score: ~50 points (neutral)

Scenario 2: Bass Angler​

  • Profile:
    • Catches bass 80% of the time
    • Uses Abu Garcia reels
    • Fishes small lakes in Texas
  • Ads Shown:
    • ✅ Bass fishing gear (high score)
    • ✅ Abu Garcia products (high score)
    • ✅ Texas fishing gear (medium score)
    • ❌ Saltwater gear (low score - filtered out)

Scenario 3: Charter Captain​

  • Profile:
    • Charter captain
    • Catches salmon, trout
    • Uses Penn reels
    • Fishes Lake Superior
  • Ads Shown:
    • ✅ Commercial fishing gear (high score)
    • ✅ Penn products (high score)
    • ✅ Great Lakes gear (high score)
    • ✅ Charter boat equipment (high score)

Scenario 4: Gear Enthusiast​

  • Profile:
    • Has 10+ gear setups
    • Uses multiple brands (Shimano, Daiwa, St. Croix)
    • Catches various species
  • Ads Shown:
    • ✅ Ads for brands they use (high score)
    • ✅ Premium gear ads (they're gear-focused)
    • ✅ New product launches (they're early adopters)

Frequency Capping​

To prevent ad fatigue, the system limits how often a user sees the same ad:

  • Lifetime Cap: Max impressions per user per campaign (e.g., 10 times)
  • Daily Cap: Max impressions per user per day (e.g., 3 times per day)

Example: User sees Rapala ad 3 times today → Won't see it again today, but might tomorrow.

Privacy & Consent​

  • Anonymous Users: Only see generic ads (no targeting)
  • Logged-in Users: See personalized ads based on their data
  • Opt-out: Users can disable personalized ads (future feature)
  • Data Usage: Only uses data user has consented to share

Performance Optimization​

Caching​

  • User interest profiles are cached (rebuild daily)
  • Reduces database queries during ad serving
  • Improves response time

Batch Processing​

  • Interest profiles rebuilt nightly
  • Campaign relevance scores pre-calculated
  • Real-time updates for recent activity

Future Enhancements​

  1. Lookalike Audiences: Find users similar to existing customers
  2. Retargeting: Show ads to users who visited advertiser's website
  3. Behavioral Signals: Track clicks, views, time spent on content
  4. Machine Learning: Use ML to improve relevance scoring
  5. A/B Testing: Test different ad creatives to see what works best
  6. Conversion Tracking: Track when ads lead to purchases/signups

API Usage​

Get Personalized Ad​

GET /api/ad-serving/next?placement=Feed&userId={userId}

What Happens:

  1. System builds/retrieves user interest profile
  2. Finds active campaigns
  3. Filters by basic targeting (location, demographics)
  4. Calculates relevance scores for each campaign
  5. Selects campaign with highest score
  6. Returns ad creative
  7. Records impression

Example Response​

{
  "adCreativeId": "creative-id",
  "headline": "New Rapala X-Rap - Catch More Bass!",
  "primaryText": "Perfect for trolling Lake Michigan",
  "imageUrl": "https://cdn.rapala.com/banner.jpg",
  "destinationUrl": "rapala.com/products/xrap",
  "clickTrackingUrl": "/api/ad-serving/click/{creativeId}",
  "format": "Banner"
}

Summary​

The system creates a 360-degree view of each user by analyzing:

  • ✅ What they like (favorite species)
  • ✅ What they catch (fishing logs)
  • ✅ What gear they use (gear setups)
  • ✅ Where they fish (location data)
  • ✅ How they fish (fishing methods)
  • ✅ What they engage with (social activity)
  • ✅ Who they are (role, activity level)

Then it matches ads based on relevance scores, ensuring users see ads that are:

  • Relevant to their interests
  • Useful for their fishing style
  • From brands they already use or might like
  • Not repetitive (frequency capping)

This creates a win-win: Users see relevant ads, advertisers reach the right audience, and you make more money from higher-performing ads! 🎣💰