Last Updated: October 13, 2025
Right. Let's talk about the elephant in the room.
You're spending £50K a month on ads. Meta says you're crushing it—4.2x ROAS (return on ad spend), baby! Google Ads? Even better—5.1x. TikTok's chiming in with a respectable 3.8x. By their math, you should be swimming in cash.
Except your bank account tells a different story. Your actual blended ROAS? Maybe 2.3x if you're lucky. And you're sitting there wondering if you've somehow broken mathematics.
You haven't. Welcome to the attribution crisis—where every platform takes credit for the same sale, and you're left holding the bag.
Here's what we're going to fix today.
Marketing attribution is figuring out which of your marketing touchpoints actually convinced someone to buy your stuff.
Not which ones happened to be nearby when they bought. Not which ones correlate with purchases. Which ones caused the purchase.
Think of it this way: A customer sees your TikTok ad. Clicks through. Leaves. Two days later, sees a Meta retargeting ad. Clicks again. Abandons cart. Then Googles your brand name, clicks your ad, and finally buys.
Quick question: Which channel gets credit?
If you answered "all of them, apparently," you're starting to understand the problem.
The average e-commerce customer touches 7-13 points before buying. That's seven opportunities for platforms to claim credit. And claim they do.
Without proper attribution, you're making decisions in the dark:
It's like playing darts blindfolded while three people shout contradictory directions. Entertaining, perhaps. Profitable? Not so much.
Here's the uncomfortable bit.
Meta says your ROAS is 4.2x. Google says 5.1x. TikTok claims 3.8x. Add those up and you should be printing money. But your actual blended ROAS? 2.1x.
This isn't a conspiracy. It's just how attribution works when everyone uses last-click attribution—meaning whoever got the last click before purchase takes full credit.
The problem? Multiple platforms can claim credit for the same bloody sale.
In 2021, Apple decided to make things interesting with App Tracking Transparency (ATT). Now platforms can't track iOS users without explicit permission. And guess what? Only 15-25% of people opt in.
The result? Your Meta dashboard is missing 40-60% of conversions. Google's missing 20-30%. And you're making budget decisions based on data with more holes than Swiss cheese.
Brilliant.
Multi-touch attribution (MTA) does what it says on the tin: it gives credit to multiple touchpoints in the customer journey, not just the last one.
Instead of one channel hogging 100% of the credit, attribution gets distributed based on each touchpoint's actual contribution. Revolutionary, I know.
What it does: Gives 100% credit to the first touchpoint that introduced the customer to your brand.
Best for: Measuring awareness channels (TikTok, YouTube, influencers)
The catch: Completely ignores everything that happened after. Like judging a film by the opening credits.
What it does: Gives 100% credit to the last touchpoint before purchase.
Best for: Quick decisions, simple funnels
The catch: This is what platform dashboards use—and why they over-report ROAS. It's the participation trophy of attribution models.
What it does: Distributes credit equally across all touchpoints.
Best for: Understanding the full customer journey
The catch: Treats all touchpoints as equally important. Spoiler: they're not.
What it does: Gives more credit to touchpoints closer to the conversion.
Best for: Longer sales cycles where recent interactions matter more
The catch: Still undervalues the awareness touchpoints that started the whole thing
What it does: Gives 40% credit to first touch, 40% to last touch, 20% distributed among the middle.
Best for: Balancing awareness and conversion optimization
The catch: Arbitrary weighting that might not match your actual funnel. But it's a decent starting point.
What it does: Uses machine learning to assign credit based on actual conversion patterns in your data.
Best for: Brands with serious data (1,000+ conversions/month)
The catch: Requires volume and technical chops to implement properly
The king of e-commerce metrics. How much revenue you generate for every pound spent on ads.
Formula: ROAS = Revenue from Ads ÷ Ad Spend
Example: Spend £10,000 on Meta Ads, generate £42,000 in attributed revenue. ROAS = 4.2x.
What's good?
What it costs to acquire one new customer across all your marketing.
Formula: CAC = Total Marketing Spend ÷ Number of New Customers
Rule of thumb: CAC should be less than 30% of Customer Lifetime Value. Otherwise you're in the business of losing money.
The total revenue a customer generates over their entire relationship with your brand.
Formula: LTV = Average Order Value × Purchase Frequency × Customer Lifespan
Example: Average order £80, customers buy 3x/year for 2 years = £480 LTV
Blended ROAS: Total revenue ÷ Total ad spend (with accurate attribution) across all channels. This is the truth.
Channel ROAS: Revenue attributed to a specific channel ÷ Spend on that channel. Useful for optimization, but will always over-report.
Trust blended. Question channel.
Most e-commerce purchases follow this pattern:
Channels: TikTok, Instagram, YouTube, influencers, display ads
Goal: Introduce your brand to people who've never heard of you
Attribution challenge: These touchpoints rarely get credit in last-click models, even though they're doing the heavy lifting
Channels: Retargeting ads, email, SMS, organic social
Goal: Build trust, educate about your products
Attribution challenge: Multiple touchpoints contribute, hard to isolate individual impact
Channels: Google Search, direct traffic, email, cart abandonment
Goal: Close the sale
Attribution challenge: These channels get all the credit in last-click models, even though earlier touchpoints did most of the work
It's like crediting the striker who taps in from two yards while ignoring the midfielder who made the 40-yard through ball. Technically accurate. Completely misleading.
The old way: Install a tracking pixel on your website. When someone visits, the pixel drops a cookie to track their behavior.
Pros: Easy to implement
Cons: Broken by iOS 14, cookie deprecation, ad blockers, and general modernity
The new way: Your server sends conversion data directly to ad platforms via API, bypassing browsers entirely.
Pros: Not affected by iOS 14, more accurate, better privacy
Cons: Requires actual technical implementation
Meta's server-side solution. Sends conversion data from your server to Meta, bypassing browser limitations.
Impact: Brands using CAPI see 10-30% more attributed conversions compared to pixel-only tracking. Not optional anymore.
Tags added to URLs to track where traffic comes from.
Example: yourstore.com/?utm_source=tiktok&utm_medium=paid&utm_campaign=spring_sale
Pro tip: Use consistent UTM naming conventions. Your future self will thank you.
For most e-commerce brands, start with position-based (U-shaped) attribution. It balances awareness and conversion channels reasonably well.
Once you hit 1,000+ conversions per month, consider data-driven attribution.
Attribution only works if you can see the full journey. Connect:
Review attribution data weekly to identify:
Meta, Google, and TikTok all over-report conversions. Always compare to your actual revenue and blended ROAS. If the numbers don't match, trust your bank account.
"Dark social" is traffic from private channels (WhatsApp, Messenger, email) that shows up as "direct" in analytics. This can be 20-40% of your traffic. It exists. You just can't see it.
Paid ads drive brand searches. Someone sees your TikTok ad, searches your brand on Google, and buys. Google gets credit. TikTok gets nothing. This is why you can't just look at last-click.
Pick one model and stick with it for at least 3 months. Constantly changing makes it impossible to compare performance over time. Consistency beats perfection.
The gold standard: incrementality testing—measuring what would have happened without your marketing.
Example: Test group converts at 3%, control at 2%. The 1% difference is your incremental impact. Everything else would have happened anyway.
This is the only way to know if your ads are actually working or just taking credit for organic demand.
The next evolution in attribution is causal inference—using statistical methods to determine actual cause-and-effect relationships between marketing actions and outcomes.
Instead of just tracking correlations ("this ad was clicked before purchase"), causal inference asks: "Would this purchase have happened without the ad?"
This combines:
The result? Attribution that actually reflects reality, not just correlation. Which is rather the point.
Right. Enough theory. Here's your action plan:
The gap between what platforms report and reality is where your opportunity lies. Close that gap, and you'll know exactly where to invest.
Which means you can scale profitably instead of guessing. Rather useful, that.
Analytics tells you what happened (traffic, conversions, revenue). Attribution tells you why it happened (which marketing touchpoints drove those outcomes). Analytics is the scoreboard. Attribution is the play-by-play.
Basic tools: £50-200/month. Enterprise solutions: £1,000-5,000+/month. Cost depends on revenue, data volume, and features. But the cost of NOT having proper attribution? Usually 10-20x higher.
Yes, but with limitations. Google Analytics 4 has built-in attribution reports. Shopify has basic attribution. You can manually analyze UTM data. But dedicated tools provide significantly more accurate multi-touch attribution. Worth the investment once you're spending £10K+/month on ads.
Basic setup (tracking pixels, UTM parameters): 1-2 weeks. Full multi-touch attribution with server-side tracking: 4-8 weeks. Data-driven attribution: 3-6 months (requires sufficient data volume).
At least 100 conversions per month for basic multi-touch attribution. For data-driven attribution, aim for 1,000+ conversions per month. Below that, stick with simpler models.
Use unique discount codes or UTM-tagged links for each influencer. Track both direct conversions and assisted conversions (customers who clicked an influencer link but converted later through another channel). Most influencer impact is assisted, not direct.
First-party (your own data) is more accurate and privacy-compliant. Third-party (platform data) is easier but less reliable post-iOS 14. Use both, but trust first-party more. Always.
For brands with retail stores, use unique promo codes from online ads, customer surveys at checkout, or CRM matching (connecting online ad exposure to offline purchases via email/phone). It's messier, but doable.
Struggling with attribution discrepancies? If you're spending €100K+ per month on ads and can't tell which channels are actually driving sales, you're not alone. Learn how leading Shopify beauty and fashion brands are solving attribution challenges to scale profitably.
Ready to see which marketing channels actually drive revenue? Causality Engine uses causal inference to show you exactly which touchpoints matter—not just which ones happened to be nearby when someone bought.
