« 3 posts on 3 topics | Home | Widemile’s Big Launch »
Find success in every test: Looking beyond conversion rates
By Billy | March 17, 2008
Every test teaches something. Almost every campaign test has the goal of raising conversion rates, but really the power of testing is in answering questions. It is just a happy coincidence that it raises your conversions at the same time.
Here are a few of the questions that can be answered with testing:
- Is giving away a free widget worth while? Would a cheaper widget perform better?
- Does promoting a few key products work better than promoting them all equally?
- How much should this product be priced?
- Does e-mail traffic respond to the same things as SEO and advertising traffic?
- What benefits are consumers interested in?
- Will a discount offer make up in conversions, what is lost in revenue per sale?
- Do banner ads reduce conversions?
- Is an extra learn more page more effective than a longer page? Is extra information even necessary?
Answering these questions is pretty easy, but you have to think about them beforehand. Design your test so that it asks questions and offers answers. Figure out what questions you want to ask and based on those, give a few separate and distinct answers in the form of variations of your page.
The easy ones are the ROI questions, e.g. giving away widgets, discounts, banner ads, and pricing. Just include them on or off in the test and/or with different amounts. At the end, do a ROI analysis comparing the conversion rates of each variation you test.
If you are wondering if one group of traffic responds differently to a campaign than another, segmentation is what you need. Separately track those segments, but don’t forget to design your test with answers too. Make your variations appeal to different audiences and your segments will be pulled towards the one they like the most with their conversions as proof.
If you want to learn what your customers are looking for, do some market research through testing. Test different types of benefits and see which get the visitors to convert, e.g. technical (this camera contains 1 gb of memory) versus lifestyle (this camera stores 100’s of wedding and birthday pictures).
You can test these ideas in an A/B split test or a multivariate test, but a multivariate test is much quicker and will allow you to test multiple questions and answers simultaneously.
Testing gives you a lot of answers and the better you design those questions into your tests, the more sense those answers will make of your data. Conversion rates are always important, but focusing solely on them won’t get you great results time after time. Think about testing in terms of learning more about your audience and you will find continual improvement in your campaigns.
Related posts
Topics: Methodology, Testing Techniques |