Cut vanity metrics, measure action

A team can post a chart showing 1 million impressions and still have no idea whether one extra dollar of spend produced one extra customer. Google’s own analytics docs separate traffic reports from attribution reports for exactly that reason: reach tells you who saw something, while attribution tries to estimate what actually moved a conversion. (support.google.com) That is the split between a vanity metric and an operating metric. Vanity metrics are raw counts like page views or followers, while operating metrics are numbers tied to a business action like conversion rate, customer acquisition cost, or customer lifetime value. (circuit.ooo, v2mom.io) Customer acquisition cost is the average amount spent to win one paying customer. The American Marketing Association says teams use that number against customer lifetime value to decide whether an acquisition strategy is economically sustainable, which is a much harder test than asking whether a campaign “got attention.” (ama.org) Customer lifetime value is the revenue or margin a customer is expected to generate over time, not on day one. If a channel brings in cheap customers who cancel after one month, the low acquisition cost can still hide a bad business. (ama.org, clevertap.com) Conversion rate fixes another common reporting mistake: counting total clicks without asking how many clicks turned into the next step. A landing page with 10,000 visits and a 1 percent signup rate is usually less useful than a page with 2,000 visits and a 6 percent signup rate, because the second page creates more real users from less traffic. (circuit.ooo, leanpivot.ai) Cohort analysis is how analysts stop averages from lying to them. Google defines a cohort as a group of users who share a characteristic such as acquisition date, then tracks how that group behaves over time instead of mixing January customers with April customers in one blended average. (support.google.com) That matters because the average customer lifetime value can look healthy while newer cohorts are quietly getting worse. If customers acquired in March retain for 6 months and customers acquired in June retain for 2 months, the blended average hides the decline until the budget has already been spent. (support.google.com, growth-onomics.com) Attribution modeling is the other half of the job. Google says data-driven attribution assigns conversion credit based on how much each ad interaction changes the estimated probability of a key event, which is a more realistic approach than giving 100 percent of the credit to the last click before purchase. (support.google.com) This shows up in everyday analyst work faster than it sounds. An A/B test is just a controlled way to ask whether one change raised a conversion rate, and a dashboard is just a repeated way to show whether acquisition cost, retention, and lifetime value are improving or slipping by channel, cohort, or campaign. (support.google.com, upscend.com) Modern analytics platforms have already moved in this direction. Google removed several older rule-based attribution models from its key event comparison report in November 2023, leaving teams to compare models with more emphasis on current conversion-path analysis instead of simplistic credit assignment. (support.google.com) The practical shift is small but brutal: replace “How big was the number?” with “What changed after we spent money?” A weekly report built around customer acquisition cost, conversion rate, retention, and cohort lifetime value gives a manager four levers to pull, while a weekly report built around impressions and likes mostly gives them a reason to feel busy. (circuit.ooo, ama.org, v2mom.io)