Most marketing decisions are made on instinct, habit, or 'what we've always done.' Data changes that — but only if you're tracking the right things and asking the right questions. Here's how to get there.
Most marketing decisions are made on instinct, precedent, or “what we’ve always done.” The social media posting schedule doesn’t change because nobody’s measured whether it works. The channel mix stays the same because changing it requires a conversation nobody wants to have. The ad creative runs until someone gets bored with it.
This isn’t unique to small businesses. It happens at companies with dedicated marketing teams and seven-figure budgets. The instinct problem is human. Data is how you override it, but only if you’ve built the foundation that makes the data meaningful.
Before any analysis is worth running, three things need to be in place. Clear goals that define what success looks like in specific, measurable terms. Tracked conversions so you know when a desired action happens and what preceded it. And consistent attribution, meaning you know where your traffic and leads are actually coming from.
Without these, you’re not analyzing marketing. You’re analyzing noise. I’ve seen businesses spend months optimizing campaigns based on data that was misconfigured from the start. The output looked analytical. The decisions it drove were no better than guessing.
The first question: which channels bring the best customers? Not the most customers, the best ones. Best usually means lowest acquisition cost combined with highest lifetime value. A channel that sends 50 high-value customers beats one that sends 500 one-time buyers, but you can’t see that without tracking both acquisition source and downstream customer behavior.
The second question: what messaging actually converts? This is harder to answer than it sounds. Most businesses run one version of their messaging and assume it’s working if revenue is growing. Testing alternatives requires volume and discipline. But even without formal testing, looking at which emails get replied to, which landing page variants hold attention, and which ads get clicked tells you something about what’s resonating.
The third question: where are you losing people in the funnel? Every marketing funnel has a biggest leak. That’s the step where the highest percentage of potential customers stop. Finding it and fixing it is almost always more impactful than finding a new channel.
Attribution is the question of which touchpoints get credit when a customer converts. Last-click attribution gives all the credit to the final thing a customer interacted with before buying. It’s easy to measure and consistently wrong, because it ignores everything that built awareness and intent before the last click.
Multi-touch attribution distributes credit across multiple touchpoints. It’s closer to reality but requires clean tracking across every channel and meaningful volume to generate reliable signals.
The honest answer is that perfect attribution doesn’t exist. The goal isn’t the most technically correct model. It’s the model that helps you make better decisions about where to put the next dollar. For most businesses, that means tracking first-touch and last-touch attribution side by side and using the tension between them to understand which channels build awareness versus which ones close.
A/B testing is only statistically valid at sufficient volume. A landing page that receives 200 visits per month cannot produce a reliable test result in any reasonable timeframe. Running a test on low-traffic pages produces conclusions that look precise and aren’t.
When you do have the volume, test in order of impact. Headlines and calls to action change behavior more than design changes do, so test those first. A headline test on a high-traffic page can produce a meaningful result in two weeks. A button color test on a low-traffic page will take months and tell you nothing.
Instead of asking how marketing performed this month, ask how customers acquired in a specific month are performing six months later. This is cohort analysis, and it changes which channels look good.
A channel that drives cheap first purchases but produces customers who never return looks great in last-click attribution and terrible in cohort analysis. A channel that drives expensive first purchases but produces customers with high retention looks expensive and looks excellent in cohort analysis. Without the cohort view, you optimize for acquisition cost and miss the lifetime value signal entirely.
Map every step from a prospect first encountering your business to completing a purchase. Measure drop-off at each step. Then fix the step with the highest drop-off rate first.
This sounds obvious. Almost nobody does it systematically. The businesses that do usually find that their biggest leak isn’t where they expected. A checkout page, not the homepage. A follow-up email sequence, not the ad creative. A pricing page that confuses rather than converts. Finding the actual constraint and fixing it beats adding a new traffic channel every time.
It’s not complicated. There’s a weekly review of the numbers that actually matter. Hypotheses get written down before campaigns launch, not invented afterward to explain the results. Learnings get documented after each campaign ends so the next one doesn’t start from scratch. And decisions are made by comparing data to a pre-set expectation, not by comparing this month to last month and calling whatever direction it moved a win or a loss.
Building this habit takes about 90 days to stick. Once it does, the instinct problem gets much smaller.
Marketing strategy grounded in data is something we work on directly with clients. If that’s what you need, get in touch.
