The New Era of Omnichannel Growth: Why Measurement Must Evolve Beyond the Device

In an era where the customer journey is no longer a linear path but a fragmented web spanning mobile apps, connected TVs, web browsers, and gaming consoles, traditional marketing measurement is facing an existential crisis. For years, marketers relied on device-level tracking to attribute growth. Today, that approach is not just outdated—it is actively misleading.

Gil Bouhnick, Director of Product at AppsFlyer, is at the forefront of this shift. As brands navigate a landscape defined by signal loss, evolving privacy regulations, and the rapid integration of Artificial Intelligence, the fundamental philosophy of how we measure "growth" is being rewritten.

The Core Fundamentals: If They Move, You Must Follow

The primary challenge modern marketers face is the disconnect between fragmented data sources. A user might discover a brand on a smartphone, research it on a laptop, and eventually make a purchase via a console or smart TV. If the measurement system treats these as three separate individuals, the data becomes noisy, expensive, and ultimately useless.

Bouhnick argues that regardless of the specific software or platform used, three fundamentals remain constant:

1. Unified Identity Resolution: Measurement must follow the customer, not the device. If your measurement stack cannot stitch together the path of a single user across platforms, you are effectively flying blind.
2. Context-Aware Data: Data without context is just noise. Marketers must understand the intent behind a touchpoint—was the user just browsing, or were they ready to convert?
3. Cross-Platform Visibility: Decisions should only be made once the full spectrum of the journey is visible. Investing in a channel that shows high conversions without seeing the preceding touchpoints often leads to the "cannibalization" of organic traffic.

The Attribution Trap: Why Your Dashboard Might Be Lying to You

One of the most provocative points Bouhnick makes is the danger of "vanity metrics" in attribution. When a dashboard shows a positive Return on Ad Spend (ROAS) but the business itself isn’t growing, the problem isn’t the marketing—it’s the attribution model.

Three Diagnostic Questions for Marketing Teams

Before reallocating budgets or pivoting strategy, Bouhnick suggests that teams ask three critical questions to determine if their "wins" are legitimate or merely illusions:

• Are we re-acquiring existing customers? This is the most prevalent blind spot in digital marketing. If a campaign brings a web-based user to a mobile app, it often registers as a "new acquisition" on the mobile dashboard. In reality, the company has simply paid to acquire the same person twice. This inflates CAC (Customer Acquisition Cost) and masks true growth.
• Are we finding demand, or just catching it? Some campaigns are highly efficient at targeting users who were going to convert regardless of the ad. This is "harvesting" demand rather than "driving" it. To verify the difference, marketers should utilize incrementality testing. If a holdout group converts at the same rate as the target group, the campaign is not driving growth; it is simply capitalizing on existing intent.
• Are our campaigns competing with themselves? In a fragmented digital ecosystem, narrow targeting often results in multiple campaigns bidding for the same audience segment. If five different campaigns claim credit for the same conversion, the business is effectively paying five times for a single customer.

Navigating the Post-Privacy Landscape

The narrative that privacy laws have "killed" marketing measurement is, according to Bouhnick, a misconception. "Perfect visibility never really existed," he notes. The industry has always dealt with noise and edge cases. What has changed is the transition from deterministic, cookie-based tracking to a model that blends deterministic data with probabilistic modeling.

Four years after Apple’s App Tracking Transparency (ATT) update, the industry has reached a "new normal." With roughly 50% of iOS users consenting to tracking, there is a solid, high-quality deterministic base. Modern measurement tools now use this 50% to build sophisticated, AI-driven probabilistic models that fill in the gaps.

The New Cadence of Decision-Making

Teams must also recalibrate their expectations regarding speed. In the old world, marketers expected real-time feedback. Today, privacy-centric measurement requires a more patient approach. Attribution may take 24–48 hours, incrementality tests require weeks, and Marketing Mix Modeling (MMM) requires months. Teams that attempt to react to hourly fluctuations in noisy data are likely to make erratic, poor decisions. True growth, Bouhnick suggests, comes from planning around these cadences rather than fighting them.

Advice for Founders: Ownership and Compliance

For early-stage startups operating in regulated environments, the temptation is to rely heavily on third-party tools to handle data and compliance. Bouhnick warns against this.

"Own your first-party data, and build your system to be compliant from day one," he advises. Every third-party vendor should be treated as an input, not as a system of record. When companies build their data architecture without GDPR or CCPA in mind, they often incur massive technical debt. He recalls instances where companies had to manually delete rows across six different databases to comply with data subject access requests—a costly, manual process that could have been automated if privacy were treated as a design requirement from the start.

Why Custom Attribution Models Often Fail

While it is tempting for engineering-heavy teams to build custom attribution models, Bouhnick strongly advises against it. Building a bespoke measurement engine is a massive, ongoing engineering burden that distracts from the core product.

Most custom models eventually become "black boxes" that break as soon as the business changes. When the original engineers move on, the remaining team is left with a set of complex, undocumented formulas that no one understands. Unless a business has a truly unique customer journey—such as high-frequency uninstalls or complex online-to-offline transactions—it is almost always better to leverage a specialized measurement partner and use that platform’s raw data to run experiments.

The Future: AI Needs Context to Succeed

Perhaps the most significant shift in the marketing landscape is the role of Artificial Intelligence. As AI takes over day-to-day optimization, the human role is evolving from manual tweaking to "context curation."

The critical mistake many marketers make is believing that measuring a device is the same as measuring a human. AI models are only as good as the context they are fed. If an AI is tasked with optimizing retention but only sees one device, it may mistakenly interpret a user logging in from a desktop as a "new" user and attempt to re-acquire them, wasting budget and potentially degrading the user experience.

"The marketer’s job is shifting," Bouhnick explains. "It is less about reading numbers and more about ensuring the machine has the full context: acquisition, engagement, retention, and lifetime value, stitched across every platform the customer touches."

Implications for the Modern Marketing Stack

The implications of this shift are profound:

1. Silo Elimination: Marketing teams must bridge the gap between web, mobile, and offline data to provide a unified source of truth for AI models.
2. Strategic Oversight: Marketers must shift their focus to the "inputs" of the AI—ensuring the data is clean, the events are tracked correctly, and the cross-platform identity is resolved.
3. Focus on Incrementality: As AI manages the "how" of bidding and targeting, human marketers must focus on the "why"—using incrementality testing to prove that the AI is actually driving business value rather than just correlating with it.

Conclusion: The Path Forward

As the digital landscape becomes more complex, the path to growth lies in clarity, not complexity. By rejecting the myth that device tracking is customer tracking, embracing the new cadence of privacy-compliant measurement, and feeding AI with holistic, cross-platform data, marketers can finally move past the fragmented, high-CAC era of the last decade.

The future of measurement is not about having more data, but about having the right data, stitched together in a way that respects user privacy while delivering the actionable insights necessary to scale. As Gil Bouhnick and the team at AppsFlyer emphasize, when the measurement foundation is solid, the business can stop guessing and start growing.