Building a Conversion Optimization Strategy From Scratch

With ad budgets shrinking through the downturn, a lot of businesses have quietly discovered the same thing this year: it’s cheaper to convert more of the traffic you already have than to buy your way to more of it. That’s the whole case for conversion optimization, and building a real strategy for it doesn’t require a big team or a big budget, just a disciplined process. Here’s how to start.

Step 1: Find where you’re losing people

Before testing anything, look at your funnel and find the specific step where the biggest share of visitors drops off, whether that’s a product page, a shipping cost reveal or a signup form. Analytics tools free and paid alike will show you this if you’re tracking goals properly. Fix the largest leak first; a five percent improvement on your worst-performing step usually beats a five percent improvement anywhere else.

Step 2: Form a real hypothesis, not a guess

“Let’s try a different color button” isn’t a strategy, it’s a shot in the dark. A useful hypothesis names the problem and the reason: visitors are abandoning the checkout page because the total shipping cost isn’t shown until the final step, so showing it earlier should reduce drop-off. That specificity is what makes a test worth running and what makes the result mean something afterward.

Step 3: Run one test at a time, and run it long enough

Google’s free Website Optimizer tool has made split testing accessible to businesses that couldn’t afford dedicated testing software, which is a real shift this year. But a tool doesn’t fix a rushed test. Don’t call a result early because it looks promising after two days; give a test enough traffic and time to reach a result you’d trust with real money, and resist the urge to change anything else mid-test.

Step 4: Separate what worked from why it worked

When a test wins, don’t just roll out the winning version and move on. Write down what you think caused the lift, because that reasoning is what lets you apply the lesson elsewhere on the site instead of treating each win as a one-off. A pattern of small, understood wins compounds; a pile of unexplained ones doesn’t.

Step 5: Build testing into the calendar, not the to-do list

A conversion program that only runs when someone remembers to run it will stall the first busy month. Put a recurring slot on the calendar, even a modest one, to review the funnel, pick the next test and check on the last one. Consistency matters more here than any individual test’s size.

What this looks like for a smaller business

You don’t need enterprise traffic volumes to benefit, though you do need patience proportional to your traffic; a low-volume site will take longer to reach a trustworthy result on any given test. Start with your highest-traffic page and your most obvious drop-off point, and treat the first few tests as much about learning your own customers as about lifting a number. The businesses getting real value out of this in 2009 are the ones treating it as an ongoing discipline rather than a single project to check off before returning to buying more traffic.

---

Keep exploring: browse more Marketing Analytics guides, see everything we published in 2009 or check out the Digital Business Marketing Awards.