In the race to optimise digital performance, it’s tempting to lean heavily on in-app/platform data (that’s the reporting data from inside Google/Meta/Linkedin etc themselves).
After all, their dashboards are all neatly packaged and presented, offering a full-stack of important sounding metrics — conversions, ROAS, CPA, impressions etc.
But here’s the problem: most of this data is modeled, not observed.
That means you’re not always seeing what actually happened — you’re seeing what the platform thinks probably happened.
And if you’re making major marketing decisions based solely on this modeled data, especially when you’re reacting to every dip or spike, you might be holding your business back more than helping it grow.
Modeled Data Isn’t Reality — It’s an Estimation
Due to growing privacy regulations (like GDPR, CCPA, and Apple’s iOS privacy restrictions), it’s harder than ever for ad platforms to track users directly across devices and apps. To fill in these gaps, they use modelled data — essentially algorithms that predict user behaviour when direct tracking isn’t available.
For example, Google Ads may use machine learning to estimate how many conversions happened after a click. And Meta might show you “estimated results” based on probabilistic attribution rather than confirmed actions.
While these models can be sophisticated, they’re still just that: models. They involve assumptions, margin of error, and black-box algorithms that can vary significantly from the actual impact of your campaigns.
And that’s before take into account the issue of cross-platform attribution – where each channel will claim full attribution, rather than acknowledging it perhaps only had ‘contribution’ to the eventual sale.
The Illusion of Precision
Whilst these platform dashboards look precise — down to decimal points and hourly breakdowns – that sense of precision is often an illusion.
Daily or weekly fluctuations in reported conversions, ROAS, or cost-per-click often reflect changes in the model’s assumptions, not actual changes in customer behaviour or campaign performance.
If you’re adjusting budgets, creative, or bidding strategies every time one of these metrics moves a few points, you might be reacting to noise — not signal. That can create instability in your campaigns, confuse your team, and ultimately undermine long-term results.
Analysis Paralysis: When Data Becomes a Distraction
When businesses become fixated on platform data, they risk slipping into analysis paralysis — constantly dissecting metrics without gaining real insight.
We’ve seen this scenario play out countless times:
-
Teams spend hours each week debating minor shifts in CPA or click-through rate.
-
Strategies are overhauled based on a 2-day dip in conversions.
-
Budgets are reallocated mid-week based on partial data.
This reactive approach not only wastes time but often leads to suboptimal decision-making. You end up prioritising what the platform says is working over what’s actually moving the needle for your business. And a lot of the time end up derailing what is actually driving the results as the platforms themselves may not be able to accurately assess the impact of longer product sales cycles and considered purchases (e.g. product awareness starts with the husband/wife on one device – then eventually converts 3 weeks later on their partner’s device on a separate IP address.)
Shift the Focus: From Platform Metrics to Business Outcomes
We argue that – instead of obsessing over platform dashboards and poor performance metrics like ROAS (which is so very easily manipulated btw) – companies should refocus on core business KPIs such as:
-
New Customer Acquisition Cost (nCAC)
-
Lifetime value (LTV)
-
Media Efficiency Ratio (MER)
-
Effective Cost per New Visitor (ECNV)
These are the numbers that matter if you’re a growth-focused business. They don’t live in a single ad platform — they live in your CRM, your finance tools, and your profit and loss account.
So should always look at your top line business metrics as a base line for analysis when you push into one paid media channel more over another during a period of time. Same goes for campaign types.
And time is a crucial factor too – don’t throw the towel in after 2 weeks if you’re testing a hypothesis. Marketing effectiveness takes time to play out as new customers enter the awareness to conversion funnel. This matters even more if you have higher ticket value products that require group decision making.
A Smarter Measurement Strategy
To truly understand how your marketing is performing, build a more balanced measurement approach:
Triangulate data sources: Compare platform-reported results with backend sales data, or other 3rd party analytics tools.
Focus on trends over time: Bi-weekly or monthly performance reviews reduce the risk of reacting to noise.
Use testing frameworks: Tools like incrementality tests and geo holdouts can give you cleaner insights into what’s truly driving performance.
Consider marketing mix modeling (MMM): If you have the budget, MMM can help you understand the broader contribution of each channel across your media mix.
Final Thoughts
Yes, data is a critical part of modern marketing — but more data doesn’t always mean better decisions. When that data is modelled, opaque, and incomplete (as it often is on paid media platforms), it should be treated with skepticism.
If your team is stuck in a loop of dashboard watching and daily budget tweaks, it’s time to zoom out.
Stop chasing platform-reported metrics. Start focusing on the numbers that reflect your actual business performance.
This is how you get the edge over those who are obsessing over the minutua.