The State of Touch-Based Attribution in B2B: What We Learned When We Dug Into Our Own Data

At Primer, we’ve faced significant challenges in accurately measuring our marketing ROI due to the limitations of touch-based attribution. Here's a breakdown of the key issues we've encountered:

1. Cookie-Based Tracking: Incomplete and Unreliable

Most analytics tools rely on cookies to track users across sessions, but we've found this approach to be increasingly ineffective:

• Browser Blocking: Browsers like Firefox, Brave, and Safari automatically block third-party cookies. This prevents our analytics tools from tracking users who visit our site using these browsers. Note: More and more cookie based analytics tools recommend using a custom proxy to make your cookie look like it is first party. (Koala does this. So does Posthog).
• User Rejection: Many users simply reject cookies, making it impossible to track their activity.
• Cross-Device Challenges: When users switch devices or browsers, our systems often fail to recognize them as the same person, resulting in multiple, fragmented user identities. Note: Even Google, the leader in the space, has huge challenges with cross-device tracking. The user has to be signed into Google on the browser and searching from the signed in browser for the Google assigned User ID to work. If the user has cookies blocked Google signals won’t work to resolve identity.
• Cookie Deletion: If a user deletes their cookies, we lose the ability to track their sessions, further fragmenting our data.
• VPNs and Proxies: Users who employ VPNs or proxy servers complicate tracking even more, as these tools can make one user appear as multiple unique visitors.
• Safari's Intelligent Tracking Prevention (ITP): Safari's ITP limits the lifespan of cookies to seven days, which severely restricts our ability to track longer sales cycles typical in B2B.

These challenges mean that a significant portion of our traffic is either untracked or misclassified, leading to gaps in our data.

Some analytics tools are turning to device fingerprinting instead of cookies to resolve sessions into single users. (Hockeystack & Mesh do this). They’re positioning themselves as “cookieless”. Fingerprinting is still legal in both EU and US though there is legislation under consideration in the EU. Device fingerprinting is very unlikely to solve the cross-device challenge even if it gets around the browser’s privacy blockers.

2. Misclassified Traffic: A Hidden Problem

Our analysis also revealed that misclassified traffic was a significant issue:

• Direct Traffic Misclassification: Users who accept cookies after the first page view may appear as "Direct" traffic, even if they originated from a different source.
• Ad Blockers: Ad blockers can disable our analytics tools, resulting in traffic being categorized as "unknown."
• Dark Traffic from Social Media: A substantial amount of social traffic is being misclassified as "Direct," making it difficult to measure the true impact of our social campaigns.

Apple's iOS 17 update has exacerbated these issues by stripping UTM codes from links in Apple Mail, Messages, and private Safari browsing sessions, further complicating our ability to track and attribute traffic accurately.

3. The Reality: We’re Missing Half the Picture

In B2B, the path to conversion is long and complex, often involving multiple touchpoints. According to Hockeystack, a B2B touch-based attribution tool, it can take up to 54 touchpoints after the first website visit to become an MQL. That’s 54 touchpoints that can be resolved into a single user. Who knows how many more touchpoints are not being resolved?

The limitations of touch-based attribution mean that we’re likely missing significant portions of this journey. The implications:

Faulty ROAS Calculations:  The data gaps we encounter mean that ROAS calculations and budget allocations based on multi-touch or first/last touch attribution should be taken with caution. These metrics depend heavily on accurate identity resolution, which, as we can see, is fraught with challenges. The only scenario where multi-touch attribution (MTA) could be accurate is a near-impossible one: a user who accepts cookies, browses on a single device and browser (not Safari, Firefox, or Brave), and doesn’t use a VPN or proxy.

Mobile’s Influence Goes Unacknowledged: Since most B2B conversions happen on desktop, it’s logical to assume that mobile traffic isn't fully connected to these conversions. For example, if 30% of our traffic is mobile, then it’s very likely a significant portion of your desktop conversions are influenced by mobile, but you don’t recognize it.

Channel Attribution Bias: Given that most B2B conversions occur on desktop, channels driving desktop traffic—like LinkedIn and Google Search—tend to get more attribution credit. It’s no surprise that these channels dominate B2B ad spend:

• LinkedIn: 43% of traffic is desktop.
• Google Search: 34% of traffic is desktop.
• Programmatic Display: 30% of ads are allocated to desktop.

This creates a skewed picture of marketing effectiveness. While desktop-heavy channels appear to perform better, this might be due to the limitations of our attribution models rather than the channels themselves.

Reconciling Data Discrepancies: The discrepancy between what our touch-based attribution tools report, what UTM tracking shows, and our own intuition often leaves us questioning our data. We’re not just missing pieces of the puzzle; in many cases, we’re looking at an entirely different puzzle altogether.

4. Addressing the Mobile vs. Desktop Discrepancy

Most of Primer’s conversions happen on desktop, but we noticed that ~50% of our traffic comes from mobile devices. And our mobile conversion rate was terrible. We had not optimized our experience for mobile conversions and almost all our conversions were from desktop. As  mobile traffic is not accurately linked to desktop conversions we couldn’t even assess the influence of these mobile ads on desktop conversions.

Assessing Mobile’s Influence on Desktop Conversions

We wanted to assess the impact mobile traffic had on desktop conversions. Pausing mobile ads resulted in a significant drop in desktop conversions, indicating that mobile plays a crucial role in the overall conversion journey—even if it’s not always credited properly by touch-based attribution.

Conclusion: The Path Forward

Touch-based attribution presents significant challenges, particularly in B2B where the user journey is complex and often spans multiple devices. At Primer, we are focusing on improving our targeting, refining our mobile strategy, and continuing to explore alternative attribution methods that can provide a more complete picture of our marketing performance. We’re big fans of Paramark and what they’re doing in MMM. And we’re starting work on some new features to help fill in the gaps in this awfully incomplete attribution picture.