mason.os
~/blog

Debugging Affiliate Tracking Discrepancies: A Systematic Approach

When your internal tracking shows $112,300 in revenue but your affiliate network reports only $81,993, you have a problem worth solving.

7 min read#engineering#ecommerce

When your internal tracking shows $112,300 in revenue but your affiliate network reports only $81,993, you have a problem worth solving. In my experience debugging affiliate tracking for e-commerce clients across Commission Junction, Awin, Impact Radius, and Rakuten, I’ve developed a systematic methodology that consistently uncovers root causes that would otherwise cost thousands in monthly commission.

The Anatomy of a Tracking Discrepancy

Affiliate tracking discrepancies typically fall into three patterns:

Pattern 1: Revenue Amount Mismatches Orders exist in both systems, but the dollar amounts don’t match. You’ll see transactions like USI: $491.84 vs Affiliate: $746.50 — a consistent 51–52% difference that screams currency conversion problem.

Pattern 2: Missing from Affiliate Your internal system captured the sale, but the affiliate network shows $0.00. The pixel fired on your end, but the affiliate never received the conversion. This is usually attribution or cookie issues.

Pattern 3: Missing from Internal The affiliate network has conversions you don’t. Your pixel didn’t fire, or the order data wasn’t captured. This points to confirmation page detection failures or race conditions.

Understanding which pattern you’re dealing with immediately narrows your investigation. A 30% revenue discrepancy with matching order counts is a completely different problem than a 30% sales discrepancy.

The Diagnostic Workflow

Step 1: Quantify the Problem

Before touching any code, establish your baseline. I use a simple threshold: if the discrepancy exceeds 25% and represents more than $400/month in commission, it warrants investigation. Below that, the debugging time may exceed the recovered value.

Pull your comparison reports covering at least 4–6 weeks. Set the date unit to weekly to distinguish between ongoing issues versus recent regressions. Sometimes affiliate networks delay reporting until orders ship — the current week often looks worse than it actually is.

Step 2: Run Diagnostic Console Scripts

The browser console is your primary debugging tool. Before making any code changes, understand the current system state.

Here’s a diagnostic script I run on confirmation pages:

(function() {
    console.log("=== AFFILIATE TRACKING DIAGNOSTIC ===\n");
    
    // 1. Page Detection
    console.log("1. PAGE DETECTION:");
    console.log("   URL:", location.href);
    console.log("   Is confirmation page:", /confirm|thank|complete|success/i.test(location.href));
    
    // 2. Order Variable Status
    console.log("\n2. ORDER VARIABLES:");
    console.log("   USI_orderID:", typeof USI_orderID !== 'undefined' ? USI_orderID : '(undefined)');
    console.log("   USI_orderAmt:", typeof USI_orderAmt !== 'undefined' ? USI_orderAmt : '(undefined)');
    console.log("   USI_currency:", typeof USI_currency !== 'undefined' ? USI_currency : '(undefined)');
    
    // 3. Cookie Status
    console.log("\n3. COOKIE STATUS:");
    if (typeof usi_cookies !== 'undefined') {
        console.log("   usi_click_id:", usi_cookies.get("usi_click_id"));
        console.log("   usi_email_id:", usi_cookies.get("usi_email_id"));
    }
    
    // 4. Affiliate Network Cookies
    console.log("\n4. AFFILIATE COOKIES:");
    // Awin
    var awinCookie = document.cookie.match(/_aw_m_\d+=[^;]+/);
    console.log("   Awin (_aw_m_*):", awinCookie ? awinCookie[0] : 'Not found');
    // CJ
    console.log("   CJ (cje):", document.cookie.match(/cje=[^;]+/) ? 'Found' : 'Not found');
    
    // 5. Pixel Timing Check
    console.log("\n5. PIXEL CONFIGURATION:");
    if (typeof usi_pixel !== 'undefined') {
        console.log("   Reporting currency:", usi_pixel.reporting_currency);
        console.log("   Site ID:", usi_pixel.site_id);
    }
    
    // 6. Data Scraping Test
    console.log("\n6. SCRAPING FUNCTION OUTPUT:");
    if (typeof usi_pixel_commons !== 'undefined') {
        console.log("   grab_order_id():", usi_pixel_commons.grab_aff_order_id ? 
            usi_pixel_commons.grab_aff_order_id() : 'Function not available');
        console.log("   grab_order_amt():", usi_pixel_commons.grab_aff_order_amt ? 
            usi_pixel_commons.grab_aff_order_amt() : 'Function not available');
    }
})();

This script answers the critical questions: Is the page being detected correctly? Are order variables populated? Are affiliate cookies present? What values would the scraping functions return?

Step 3: Trace the Data Flow

Affiliate tracking involves a chain of data handoffs. Failure at any point breaks attribution. The chain typically looks like:

  1. Click tracking → User arrives via affiliate link, cookie gets set
  2. Session persistence → Cookie survives through checkout flow
  3. Confirmation detection → Pixel fires on the right page
  4. Order scraping → Correct order ID and amount are captured
  5. Pixel transmission → Data reaches both internal and affiliate systems
  6. Currency normalization → Values align between systems

I systematically validate each step. When testing pixel firing, I simulate the full flow:

// Mock a pixel fire to test the data chain
USI_orderID = "TEST123456";
USI_orderAmt = "100.00";
USI_currency = "GBP";
// Set affiliate cookie if testing attribution
document.cookie = "_aw_m_17946=test_click_id; path=/";
// Load and fire the pixel
var script = document.createElement('script');
script.src = '//www.upsellit.com/active/client_pixel.jsp?orderID=' + USI_orderID + 
    '&orderAmt=' + USI_orderAmt + '&currency=' + USI_currency;
document.head.appendChild(script);

Then check the Network tab. If you see the pixel request go out with empty orderID and orderAmt parameters despite setting them above, you've found a scraping function that's overwriting your values.

Common Root Causes and Fixes

Currency Mismatches

This is the most common cause of revenue discrepancies. The math usually gives it away immediately.

I had a client showing ~52% discrepancy. Every matching order showed patterns like USI: $46.53 vs Affiliate: $70.52. The ratio? 0.66 — almost exactly the AUD to USD conversion rate at the time.

The pixel was configured with:

usi_pixel.reporting_currency = "USD";

But the affiliate network tracked in AUD. The fix:

usi_pixel.reporting_currency = "AUD";

Always verify your pixel’s reporting currency matches what the affiliate network expects.

Pixel Timing Issues

The default 2000ms delay in many pixel implementations causes significant conversion loss:

setTimeout(usi_pixel.main, 2000);

Two seconds is an eternity on a confirmation page. Users close tabs, navigate away, or their browser starts loading the next page. I’ve seen reducing this to 500ms recover 10–15% of previously missing sales.

setTimeout(usi_pixel.main, 500);

The tradeoff is that faster timing means less time for order data to populate on the page. Test thoroughly — 500ms is usually sufficient for most platforms, but some SPAs need more time.

Value Overwriting

One particularly frustrating pattern: you pass order data via URL parameters, but the scraping functions overwrite it with empty values.

The problematic pattern:

usi_pixel.main = function() {
    USI_orderID = usi_pixel_commons.grab_order_id();  // Returns "" if not on confirmation page
    USI_orderAmt = usi_pixel_commons.grab_order_amt(); // Overwrites valid data with ""
    // ... fires pixel with empty values
};

The fix adds conditional logic:

usi_pixel.main = function() {
    if (typeof USI_orderID === "undefined" || USI_orderID === "" || USI_orderID === "InsertOrderID") {
        USI_orderID = usi_pixel_commons.grab_order_id();
    }
    if (typeof USI_orderAmt === "undefined" || USI_orderAmt === "" || USI_orderAmt === "InsertOrderSubtotal") {
        USI_orderAmt = usi_pixel_commons.grab_order_amt();
    }
    // ... fires pixel preserving valid data
};

This preserves explicitly passed values while still falling back to scraping when needed.

Missing Affiliate Cookie Restoration

For Awin specifically, if users arrive via affiliate links but the proper cookies aren’t being set, conversions won’t attribute correctly.

Many implementations miss the fix_awin call:

// This should be called BEFORE monitor_affiliate_pixel
usi_aff.fix_awin("MERCHANT_ID");
usi_aff.monitor_affiliate_pixel(function() {
    if (typeof USI_orderID == "undefined") {
        usi_commons.load_script("//www.upsellit.com/active/client_pixel.jsp");
    }
});

The fix_awin function captures the awc parameter from the URL and sets the cookies that Awin's conversion pixel looks for.

Platform-Specific Gotchas

Shopify Web Pixel Sandbox Shopify’s checkout runs in a sandboxed environment with limited DOM access. Standard scraping selectors fail. You need to use Shopify’s checkout data objects or Customer Events API.

BigCommerce Cart API BigCommerce requires async API calls to retrieve cart data. Your scraping functions need proper callback handling:

fetch('/api/bigcommerce/cart')
    .then(response => response.json())
    .then(data => {
        // Process cart data
    });

Next.js/React SPAs The confirmation page may render after your pixel script loads. Add DOM observers or use platform-specific data layer events instead of direct scraping.

Validation and Deployment

After implementing fixes, validate before deploying:

  1. Console test the fix — Run your diagnostic script again and confirm the expected values populate
  2. Mock a full transaction — Use test mode or a real test order to verify end-to-end flow
  3. Check Network tab — Confirm the pixel request contains correct orderID and orderAmt
  4. Verify affiliate dashboard — Ensure test conversions appear with correct attribution

Post-deployment, monitor the Snapshot Reports for 1–2 weeks. You should see the discrepancy percentage drop. If you fixed a timing issue, you’ll often see immediate improvement. Currency fixes show instant correction on matching orders.

The Debugging Mindset

The methodology I’ve outlined follows a consistent pattern: quantify → diagnose → trace → fix → validate. Each step builds on the previous one, preventing wasted effort on symptoms instead of causes.

The key insight is that affiliate tracking discrepancies are almost never mysterious. They’re systematic failures with identifiable root causes. Currency mismatches follow predictable math. Timing issues correlate with page load behavior. Missing cookies trace back to initialization order.

Console-first debugging — running diagnostic scripts before making code changes — saves enormous time. You’re not guessing at fixes; you’re confirming hypotheses with data.

The fixes themselves are usually simple. A reporting currency change is one line. A timing optimization is changing a number. The value is in knowing which line to change and why it matters.