Metrics Explorer Best Practices: 7 Tips 🔭

Looking for pro tips on using Metrics Explorer? Here's some ideas from our Media Strategy team.

Always use at least 30 days of data to ensure statistical significance. The statistical model powering Metrics Explorer is just that - a model. The more data you feed it, the more effective it will be. If you use less than 30 days of data, you're likely to over-credit correlations that maybe look weaker over longer time frames. We also know that marketing has a lagging effect on revenue, visits, and other performance. Too short of a time frame can result in confusing or misleading results.

Remember, correlation doesn't imply causation. What this means is that while you might see strong correlations between your metrics, this isn’t a guarantee that future behavior of those metrics will be the same. This isn’t a limitation of Metrics Explorer - this is a limitation of all statistical analysis. Tools like the Pearson Correlation Coefficient are designed to give us an idea about what has and what could happen. It is impossible to predict the future - we can only make extremely educated guesses. The wisest marketers use a combination of statistics, historical data, and industry knowledge when interpreting results or forecasting the future. Metrics Explorer can be a critical part of that measurement framework.

Use Metrics Explorer to explore channels that feel misunderstood (or under-measured.) TikTok, YouTube Ads, CTV - these are all channels that struggle with typical Meta-style direct response measurement. Use Metrics Explorer to evaluate these higher-funnel, impression-based channels. As you spend more on those channels, how does that correlate with metrics like first-time revenue? What about revenue on your lower-funnel channels like Google and Meta?

Prioritize strong positives and strong negatives when interpreting results. It’s difficult to get strong positive or negative relationships with the PCC model. Strong positive and negative relationships are the most trustworthy and effective for measuring what actually happened. Moderate or weak relationships are more likely to have been influenced by unseen variables - use your own judgment when evaluating those results.

Use Metrics Explorer to evaluate what you think you already “know.” This allows you to create a strong foundation of learnings to iterate against. Certain relationships may seem obvious (like ad spend’s impact on transactions) but with Metrics Explorer you can prove those relationships with data, instead of anecdotal experience. Whether you’re presenting to your boss or to your clients, establishing trust in your data is paramount. By demonstrating with data what your team “already knows” you can validate Northbeam’s data in the eyes of all your stakeholders.

Remember that Metrics Explorer demonstrates what HAS happened, not what WILL happen. Past performance is no guarantee of future results. However, it doesn’t mean you can’t make an educated guess based on past variables. If all else holds true, and you don’t change anything else in your strategy, you can have confidence that the relationships in Metrics Explorer can be assumed for future results. On that note…

Use Metrics Explorer alongside a calendar to tighten up your analysis. Maintain a “changelog” where you write down all big changes you make, experiments you launch, or other variables like product drops or major promo periods. You want to isolate periods of time where the two metrics you’re comparing are mostly untouched by other variables. Measuring a correlation between spend and revenue would be affected by a product drop in the middle of that date range, for example. Working alongside a calendar of important dates can help you identify why relationships are - or aren’t - appearing in the data.

👍

Looking for more hands-on help? Try our Metrics Explorer Quick Start Guide. 🚀