Analytics & Data-Driven Decisions
Analytics & Data-Driven Decisions
Data-driven is one of the most abused phrases in marketing. Brands that report weekly dashboards of traffic and engagement and call it data-driven are reporting, not analyzing. True data-driven marketing means having a decision framework where measurements change behavior — specifically, where insight from data changes what you invest in, stop doing, or start testing.
The Measurement Framework Hierarchy
Level 1 — Operational metrics: Daily/weekly numbers that tell you if something is broken. Traffic, ad spend, conversion rate. If these spike or drop, something changed. These are monitoring metrics, not strategy metrics.
Level 2 — Performance metrics: Weekly/monthly numbers that tell you how efficiently you are achieving goals. CAC by channel, LTV by cohort, revenue per email subscriber. These guide resource allocation.
Level 3 — Strategic metrics: Quarterly/annual numbers that tell you if the business model is working. CAC:LTV ratio, payback period, net revenue retention. These inform positioning, pricing, and market strategy.
Most marketing teams are excellent at Level 1 and poor at Level 2 and 3. This means they can tell you if campaigns are performing but not whether the business is on a healthy trajectory.
Building the Right Measurement Stack
The analytics stack most marketing teams need:
- Web analytics (GA4, Plausible): Traffic, acquisition source, user flow, goal completions
- Product analytics (Mixpanel, Amplitude): User behavior within the product, feature adoption, activation funnel
- Revenue analytics (Stripe, Baremetrics, ChartMogul): MRR, churn, LTV, cohort revenue
- Ad platform data (Meta, Google): Channel-level performance, but deduplication required before trusting it
- Customer data platform (Segment, RudderStack): The connective tissue that joins the above with a unified customer ID
Without a CDP or consistent customer ID across tools, you cannot build the customer journey view that makes cohort analysis possible.
Cohort Analysis: The Most Underused Marketing Tool
A cohort analysis groups customers by when they were acquired and tracks their retention over time. The insight it reveals is invisible in aggregate data.
Example: Your overall MRR is growing. But cohort analysis shows customers acquired in Q3 churn at 2× the rate of customers acquired in Q1. Investigation reveals that Q3 campaigns ran heavy discounts that attracted price-sensitive customers with lower LTV. Without the cohort view, you would have called Q3 a success.
How to run a basic retention cohort:
- Group customers by acquisition month
- Track what percentage of each cohort is still active (paying, logging in, purchasing) at 30/60/90/180 days
- Compare retention curves across cohorts to identify patterns
- Investigate what was different about acquisition source, messaging, or product experience for cohorts that retain better
Setting the Right North Star Metric
The NSM selection process:
- What single action, if taken repeatedly by a customer, is most predictive of their long-term value?
- Can this metric be measured consistently across the entire customer base?
- Does every team understand how their work influences this number?
- Does improving this number always mean the business is getting better (not just gaming a metric)?
Common NSM mistakes: choosing revenue as the NSM (it's a lagging indicator that doesn't tell teams what to do day-to-day), choosing engagement metrics that don't correlate with revenue (time on site without conversion rate is vanity), or choosing a metric so complex it requires explanation every time it's mentioned.
The Analytics Audit
Before building new dashboards, audit what you have:
- What decisions changed last quarter because of data?
- Which metrics are tracked but never acted upon?
- Which decisions are being made based on intuition that could be validated with data?
The most valuable analytics investment is eliminating reports no one acts on and replacing them with questions the business needs answered. Reports are not decisions. Answers are.