Testing new hypotheses and exploring data comes with a special tax, hindering the testing culture and feedback loops in digital marketing

Performance marketing thrives on innovation, testing, and iteration. Yet, exploring hypotheses, data and new questions comes with a price. This “testing tax” hampers agility and limits the potential to uncover insights in an increasingly competitive digital landscape. Let’s explore why this happens and how teams can overcome it.

The testing tax

Digital marketing evolves faster than ever. AI Campaigns like Google’s PMax and Meta ASC+ are automating more processes, reducing intra-channel optimization opportunities. Traditional levers like bids and budgets are giving way to creative assets, keywords, offers, and landing pages as the primary drivers of success.

Meanwhile, rising CPCs and growing scrutiny place marketers under constant pressure. Testing and iterating quickly remain essential, but achieving this requires actionable insights—fast. Unfortunately, the distance between questions and insights is often too long.

Challenge: the distance between questions and insights

Despite access to BI tools like Power BI, Tableau, and Looker, many enterprise teams face a critical gap: the data needed for hypothesis testing isn’t readily available. BI tools are great to understand what's happening on the high level (management). Ad platforms are enough to answer channel-specific and pointed questions. However, exploring new questions and hypothesis requires special data preparation and modeling. Why?

1. Abstraction into meaningful groups: analyzing individual assets, ad groups or even campaigns it's often not very helpful. Actionable insights typically require grouping data in a meaningful way - such as keyword groups, creative concepts, campaign types, etc

2. Cross-channel insights: driving insights across ad platforms helps to . This requires harmonizing and consolidating raw data from multiple ad platforms.

This creates a middle-layer gap between platform data and BI dashboards, leaving many questions unanswered. And optimization opportunities missed.

Specific examples and use cases

Let’s examine how this challenge plays out in common scenarios:

1. Paid Search: keyword grouping

• Goal: Compare keyword performance across markets (e.g., brand vs. generic vs. competitor).

• Challenges:

  • Reporting on BI tools only provide campaign-level data and become too slow for granular data. Ad platforms don't allow data manipulation.

  • This analysis requires joining keyword-level data across multiple ad platforms and grouping it based on specific rules

2. Paid Social and Display: Creative Analytics

• Goal: Identify winning creative formulas (concepts, formats, elements) through creative analytics

• Challenges:

  • Reporting on BI tools doesn't provide asset-level information and can't enable qualitative insights (i.e., ad previews)

  • Platform-specific analysis lack the holistic view and backend data

  • This analysis requires harmonizing asset-level data from multiple ad platforms, join backend data (attribution, CLV) and enable ad previews next to all KPIs

3. Cross-channel campaign performance

• Goal: Evaluate the impact of a specific promotion across channels.

• Challenges:

  • Platform-specific analysis lack the holistic view and backend data

  • This analysis requires consolidating and harmonizing data from multiple platforms and enabling proper filters

  
  

The common pitfalls

Faced with these challenges, teams often try one of these approaches:

1. Hiring marketing analysts - thinking "if we hire more marketing data analysts, they'll build the dashboards we need"

2. Workarounds and point solutions - thinking "each digital marketing team can find their own solution" (e.g., Google Studio for paid search, Smartly for paid social)

On the surface, both approaches sound logic and valid. In fact, every company went down this path at some point. Here's why these approaches don't work:

1. Constantly changing needs - Marketing moves fast. The dashboards your team builds today will be outdated in just a few months. You're always playing catch-up.

2. Slow speed to insight - When it takes weeks to get a new dashboard or analyze data, you can't make quick decisions. By the time you understand what happened, the moment has passed. Data teams often become bottlenecks, leading to significant friction between both teams. In addition, all this reporting work takes away from other priorities, namely measurement (MMM, MTA, Incrementality).

3. Silos across teams - Each team using different tools creates barriers and silos. No one shares best practices, and systems don't work well together. In addition, these workarounds tend to break at some point (lack scalability and flexibility).

These challenges lead to the following symptoms:

• Reactive marketers doing minimal testing (just monitoring campaigns and answering management questsions);

• Marketers becoming data janitors, spending hours preparing data for one-off analyses and hypothesis testing (this is why marketers keep going back to spreadsheets);

• Data teams stuck on Jira tickets, dedicating most of their time to reactive, ad-hoc work, having little time for key measurement priorities - MMM, MTA and Incrementality.

Without a scalable, agile and dynamic solution, marketers test fewer hypotheses and spend more time reacting than strategizing. This leads to missed optimization opportunities and wasted ad spend.

Solution: The Analytics Middle Layer

To escape this cycle, performance marketing teams need a middle-layer solution that bridges the gap between data and actionable insights. To go from questions to insights in minutes, without any dependencies on data/BI teams. This is the only way to democratize access to data and insights.

This Analytics Middle Layer needs to meet the following requirements:

• Centralized granular data across channels and backend data (attribution, CLV, contribution margin, etc).

• Automated data harmonization to enable cross-channel comparisons.

• Easy grouping for keywords, creatives, ad groups, etc using naming conventions (no-code);

• Marketer-friendly exploration with filters, metrics, and dimensions.

• Built-in governance to ensure accuracy and trust.

• Independence from data teams to avoid bottlenecks.

Closing the gap

By eliminating the testing tax and shrinking the distance between questions and insights, performance marketing teams are able to test much more and faster. Ultimately moving faster than competition and uncovering more optimization opportunities.

Working with leading marketing teams at companies like Uber, Delivery Hero and HelloFresh, we’ve seen firsthand the results:

• 20-30% boost in operational efficiency.

• 2-3x increase in testing volume.

• 4-8x faster time to insight.

With the right analytics platform, marketers can stop being data janitors and focus on what they do best: driving results through creative, data-informed strategies.