Attribution Modeling Beyond Last-Click — What DTC Brands Actually Need

The Last-Click Lie

A DTC brand spent $500K/month on marketing. According to last-click attribution: Google Brand Search drove 60% of conversions. Facebook drove 5%. They cut Facebook spend by 50%. Two weeks later, Google Brand Search conversions dropped 35%.

Why? Facebook drove awareness. Customers saw Facebook ads, remembered the brand, then Googled the brand name and bought. Last-click gave all credit to Google. When they cut Facebook, fewer people searched for the brand.

Last-click attribution is the default in every analytics platform because it's simple. It's also dangerously wrong for any business with a multi-touch customer journey.

Attribution Models Explained

Last-Click:
  Touchpoints: Facebook Ad → Blog Post → Google Search → Purchase
  Credit:      0%             0%          100%           ←
  Problem:     Ignores everything that led to the search

First-Click:
  Touchpoints: Facebook Ad → Blog Post → Google Search → Purchase
  Credit:      100%          0%          0%             ←
  Problem:     Ignores everything after initial awareness

Linear:
  Touchpoints: Facebook Ad → Blog Post → Google Search → Purchase
  Credit:      33%           33%         33%            ←
  Better:      At least acknowledges every touchpoint
  Problem:     Treats all touchpoints as equally important

Time-Decay:
  Touchpoints: Facebook Ad → Blog Post → Google Search → Purchase
  Credit:      15%           25%         60%            ←
  Better:      Gives more credit to recent touchpoints
  Problem:     Still under-credits top-of-funnel

Position-Based (U-Shaped):
  Touchpoints: Facebook Ad → Blog Post → Google Search → Purchase
  Credit:      40%           20%         40%            ←
  Best default: Credits both discovery and conversion touchpoints

Implementing Multi-Touch Attribution

Step 1: Capture Every Touchpoint

// Track all marketing touchpoints in a customer journey
interface Touchpoint {
  timestamp: Date;
  channel: string;       // "facebook", "google", "email", "direct"
  campaign?: string;     // UTM campaign
  medium: string;        // "cpc", "organic", "social", "email"
  source: string;        // "google", "facebook", "klaviyo"
  landingPage: string;
  sessionId: string;
  userId?: string;       // If logged in
  anonymousId: string;   // Cookie-based ID
}
 
// Capture on every page load
function captureTouchpoint() {
  const params = new URLSearchParams(window.location.search);
  const referrer = document.referrer;
  
  const touchpoint: Touchpoint = {
    timestamp: new Date(),
    channel: determineChannel(params, referrer),
    campaign: params.get("utm_campaign") || undefined,
    medium: params.get("utm_medium") || inferMedium(referrer),
    source: params.get("utm_source") || inferSource(referrer),
    landingPage: window.location.pathname,
    sessionId: getSessionId(),
    anonymousId: getAnonymousId(),
  };
 
  // Store in first-party storage (not just cookies — those get blocked)
  appendToJourney(touchpoint);
}

Step 2: Build the Customer Journey

// Reconstruct the full journey at conversion time
async function getCustomerJourney(customerId: string): Promise<Touchpoint[]> {
  // Merge anonymous and authenticated touchpoints
  const anonymousId = getAnonymousIdForCustomer(customerId);
  
  const touchpoints = await db.touchpoints.findMany({
    where: {
      OR: [
        { userId: customerId },
        { anonymousId: anonymousId },
      ],
    },
    orderBy: { timestamp: "asc" },
  });
 
  // Deduplicate and clean
  return deduplicateTouchpoints(touchpoints);
}

Step 3: Apply Attribution Models

function attributeConversion(
  journey: Touchpoint[],
  conversionValue: number,
  model: "linear" | "time_decay" | "position_based" = "position_based"
): AttributionResult[] {
  if (journey.length === 0) return [];
  if (journey.length === 1) {
    return [{ touchpoint: journey[0], credit: conversionValue }];
  }
 
  switch (model) {
    case "linear":
      const equalCredit = conversionValue / journey.length;
      return journey.map(tp => ({ touchpoint: tp, credit: equalCredit }));
 
    case "time_decay": {
      const halfLife = 7 * 24 * 60 * 60 * 1000; // 7 days
      const conversionTime = journey[journey.length - 1].timestamp.getTime();
      const weights = journey.map(tp => {
        const age = conversionTime - tp.timestamp.getTime();
        return Math.pow(0.5, age / halfLife);
      });
      const totalWeight = weights.reduce((a, b) => a + b, 0);
      return journey.map((tp, i) => ({
        touchpoint: tp,
        credit: (weights[i] / totalWeight) * conversionValue,
      }));
    }
 
    case "position_based": {
      // 40% first, 40% last, 20% distributed among middle
      const credits = journey.map((_, i) => {
        if (i === 0) return 0.4;
        if (i === journey.length - 1) return 0.4;
        return 0.2 / (journey.length - 2);
      });
      return journey.map((tp, i) => ({
        touchpoint: tp,
        credit: credits[i] * conversionValue,
      }));
    }
  }
}

The Attribution Dashboard

Channel Performance (Position-Based Attribution):

| Channel        | Last-Click Rev | Multi-Touch Rev | Difference |
|----------------|---------------|-----------------|------------|
| Google Brand   | $300K (60%)   | $150K (30%)     | -50%       |
| Facebook Ads   | $25K (5%)     | $125K (25%)     | +400%      |
| Email          | $100K (20%)   | $80K (16%)      | -20%       |
| Google Non-Brand| $50K (10%)   | $75K (15%)      | +50%       |
| Direct         | $25K (5%)     | $70K (14%)      | +180%      |

Insight: Facebook is 5x more valuable than last-click suggests.
Cutting Facebook spend would reduce Google Brand conversions
because Facebook drives the awareness that creates brand searches.

Privacy-Compliant Attribution

With third-party cookies dying, attribution needs to adapt:

First-party data strategies:
  → Server-side tracking (events sent from your server, not browser)
  → First-party cookies (your domain, longer lifespan)
  → Authenticated user matching (email-based cross-device)
  → Media Mix Modeling (statistical, no user-level tracking needed)

Implementation:
  → Move tracking server-side (Meta CAPI, Google Enhanced Conversions)
  → Encourage account creation earlier in the funnel
  → Use post-purchase surveys: "How did you hear about us?"
  → Supplement with MMM for channel-level budget allocation

Multi-touch attribution isn't just a reporting change — it's a budget allocation change. When you see the real contribution of each channel, you spend differently. And when you spend differently based on better data, you grow faster.