We’ve recently started doing multi-variate landing page testing via Google’s Website Optimizer. With our first full project under our belts – we’ve been absolutely blown away. The results are one thing, but the ease of running the test has us totally stoked as well.
Up until this tool was available, you had two options. Scream, yell & fight with your Business Prevention Unit (read: IT Department) to run a test, or outsource to a firm with it’s own platform and pay a LOT of money. Neither of these options are particularly realistic for most small & mid-size businesses. In a lot of cases, it’s hard to even get clients to invest in landing pages for their PPC campaigns despite the fact that on average our clients with dedicated PPC landing pages have conversion rates in the double-digits, while those who don’t tend to have conversion rates in the single digits.
Multi-Variate PPC Landing Page Test – Background:
We had been working with this client’s PPC campaigns for well over a year, and to be honest we had such good conversion rates (15%-20%), that I was skeptical as to whether we were going to be able to improve on those numbers. For this particular test – we focused on one campaign which had been consistently logging a conversion rate right around 15%.
The PPC Campaign:
This client’s bread & butter is offering an online real estate license training program with a 100% guarantee that students who complete the program will pass the real estate test. The nature of the business is such that the school can only offer training in states where it has been accredited. So, our campaigns are geo-targeted to the states where they can offer training.
Designing the Test:
We didn’t have the option of testing radically different landing page designs, or even different offers at this point. So, we stuck with the existing page layout and offer and focused on testing key page elements. The offer is an information packet for prospective students, and the landing page is a straight-ahead explanation of the value proposition (online classes, guarantee). The goal of our experiment was to make the page more appealing through use of graphics and testing different language on the page.
We were excited that even when using the existing page layout, and a form that’s too long (IMHO), and the same basic copy – we have seen dramatic improvements. You can see a screen shot from the reporting center below, which gives some estimated ranges for conversion rates. A breakdown of the results is as follows:
- Conversion rate lift: 43%
- Change in cost per conversion: 38% decrease (and dropping)
- Limit variations. You may have lots of good ideas to test, but try to keep your focus. Our initial ideas factored out to well over 200 landing page variations. With the size, budget & average number of clicks for this particular campaign, it would have taken about 5 years to get a statistically significant result. You can always move on to phase II ideas after you get your initial results.
- Increase your click budget to speed things up. More clicks in a shorter amount of time get you the results much faster so you can move on to the next test.
- Be different. The differences in your variations for each page element should NOT be subtle. Otherwise, you’ll get subtle results.
- Run a follow-up test. Test your original vs your winning combination for a time to confirm your results.
- Being a search marketer and not a designer, I was surprised how much impact images had on the result even when the content was the same.
The results for this one campaign have energized the team. We’re rolling out follow-up A/B experiments across all of their campaigns to test the winning page combination vs the current combination. We’re also planning on running tests for:
- Radically different page designs
- New offers
- Form variations (fewer fields)
We’d love to hear your suggestions for follow-up experiments, or your experiences with landing page tests!