Here is a scenario that plays out in marketing meetings everywhere, every month.

The Google Ads dashboard says it drove 120 conversions last month. The Meta Ads dashboard says it drove 95. GA4 says total conversions were 140. The actual number of sales recorded in the CRM is 87. Every single figure is different, and everyone in the room has a different explanation for why.

Welcome to the attribution problem — one of the most persistent, most misunderstood, and most commercially significant challenges in paid media.

Attribution is the process of assigning credit to the marketing touchpoints that contributed to a conversion. It sounds straightforward. In practice, it's anything but — because the customer journey rarely follows a straight line, because every platform has an incentive to claim as much credit as possible, and because the tracking infrastructure underpinning it all is increasingly unreliable.

Why the numbers never add up

The core reason attribution data is confusing is that most platforms measure conversions independently, using their own tracking, their own attribution windows, and their own models — with no coordination between them.

Consider a simple customer journey: someone sees a Meta ad on Monday, clicks a Google Search ad on Wednesday, and completes a purchase on Thursday. Under most default settings, Meta will claim that conversion (because the person saw the ad within the attribution window), Google will claim that conversion (because it was the last click), and GA4 will record one conversion attributed to Paid Search. The business made one sale. Three platforms are claiming varying degrees of credit for it.

This double and triple counting is not a bug or an act of bad faith — it's the natural result of each platform measuring independently. But it means that if you add up the conversions reported across all your platforms, you will almost always arrive at a number significantly higher than the actual conversions your business recorded.

The main attribution models — and what each one misses

Last-click attribution

The simplest and historically most common model. 100% of the credit goes to the last touchpoint before conversion — typically a brand search or a direct visit. The problem is that last-click attribution systematically undervalues everything that happened earlier in the journey: the Meta ad that introduced the brand, the blog post that built consideration. Channels that do the awareness and consideration work look useless under last-click, which leads businesses to cut them and then wonder why brand searches start to decline.

First-click attribution

The mirror image of last-click: all credit goes to the first touchpoint. This overstates the value of discovery channels and ignores everything that actually closed the sale. Useful as a lens for understanding how people first find you, but not a sensible basis for overall budget allocation.

Linear attribution

Credit is distributed equally across all touchpoints in the journey. More balanced than last-click or first-click, but it assumes every touchpoint contributed equally — which is rarely true.

Time decay attribution

More credit is given to touchpoints closer to the conversion. This makes intuitive sense for short buying cycles. For longer buying cycles, it can undervalue the channels that built initial awareness.

Data-driven attribution

Google's preferred model, and now the default in Google Ads and GA4. Rather than applying a fixed rule, it uses machine learning to analyse your actual conversion data and estimate how much each touchpoint contributed. It's theoretically the most sophisticated option — but it requires significant conversion volume to work reliably (at least a few hundred conversions per month), and it's still operating within Google's ecosystem.

View-through attribution

Perhaps the most controversial model. A view-through conversion is recorded when someone saw your ad (but didn't click it) and later converted within a defined window — often 24 hours or longer. The problem is that this window overlaps heavily with conversions that would have happened anyway. View-through conversions should be treated with extreme scepticism and reported separately from click-based conversions rather than included in headline performance numbers.

The platform incentive problem

It's worth being direct about something that often goes unsaid: every paid media platform has a financial incentive to show you the largest possible conversion numbers. More conversions attributed to their platform means you spend more on their platform. Their attribution models are built with this commercial reality operating in the background.

This doesn't mean the numbers are fabricated. But it does mean that default attribution windows are often generous, view-through conversions are included in headline figures without being clearly labelled, and the models are calibrated in ways that tend to favour the platform reporting them.

Meta's default attribution window has historically been 7-day click and 1-day view — meaning a conversion is attributed to Meta if someone clicked an ad within the last seven days or merely saw one within the last 24 hours. That's a wide net. Always check which attribution window and model is being used when you review platform performance data.

Why tracking has become less reliable

Even setting aside the attribution model question, the underlying tracking data that attribution models rely on has become significantly less reliable over the past few years.

Apple's iOS 14 update in 2021 introduced App Tracking Transparency, requiring apps to ask users for permission to track them across other apps and websites. The majority of users opted out, which created substantial gaps in Meta's ability to attribute conversions from iPhone users — in some accounts, reported conversions dropped by 30–50% overnight, not because performance had fallen but because the measurement had.

Browser-level changes have compounded this. Safari blocks third-party cookies entirely. Firefox has similar restrictions. The net effect is that browser-based tracking, which underpins most digital attribution, is working with increasingly incomplete data.

This is why server-side tracking — sending conversion data directly from your server to the ad platform, bypassing browser-based limitations — is becoming an increasingly important part of a robust tracking setup.

A more useful way to think about attribution

Given all of the above, what's the right approach? The honest answer is that perfect attribution doesn't exist — and chasing it can lead you to spend more time analysing imperfect data than making good decisions based on it. A more practical framework combines several lenses:

MER — Marketing Efficiency Ratio

Total revenue divided by total marketing spend, across all channels. This is a blended, top-down view that bypasses the attribution problem entirely — it doesn't matter which platform claims what credit, because you're looking at the overall relationship between what you spend and what you generate. It's a blunt metric, but it's honest in a way that individual channel attribution rarely is.

Incrementality testing

Rather than asking 'which channel gets credit for this conversion?', incrementality testing asks 'would this conversion have happened without this ad?' This is done by running controlled experiments — turning off a channel or reducing spend in a specific region and measuring the impact on overall conversion volume. It's more operationally complex than reading a dashboard, but it produces genuinely reliable insight into which channels are actually driving incremental business.

CRM-based tracking

Your CRM or order management system records actual sales — not platform-reported conversions. Comparing CRM data against platform data regularly is a useful sanity check: if Google Ads is claiming 150 conversions and your CRM shows 90 sales from online sources, that gap is telling you something important about how much to trust the platform numbers.

GA4 as an independent reference point

GA4 sits on your website rather than within any individual ad platform, which makes it a more neutral source of attribution data than the platform dashboards themselves. Using it as a cross-reference against platform-reported numbers gives you a more rounded picture than relying on any single source.

Practical steps to take now

  • Standardise your attribution windows — check what window each platform is using and align them as far as possible
  • Separate view-through conversions from click-based conversions in your reporting — never include view-through figures in headline performance numbers without explicitly labelling them
  • Track MER monthly as a top-level health metric, alongside individual channel CPAs
  • Implement server-side tracking (Conversions API for Meta, enhanced conversions for Google) to recover the signal being lost to browser restrictions
  • Regularly cross-reference platform conversions against CRM or actual sales data
  • When making significant budget decisions, consider running a simple incrementality test rather than relying solely on attribution data

Make peace with imperfect data — but demand better than noise

Attribution will never be perfect. The customer journey is too complex, the tracking infrastructure too imperfect, and the platform incentives too misaligned for any single model to give you the complete truth. Accepting that is not a counsel of despair — it's the starting point for making better decisions.

The businesses that navigate attribution well aren't the ones with the most sophisticated models. They're the ones that use multiple lenses, maintain healthy scepticism about platform-reported numbers, cross-reference against real business data, and make budget decisions based on the weight of evidence rather than any single figure.

If you'd like help putting any of this into practice for your own campaigns, get in touch or book a free discovery call.

Related Reading