Conversion optimization is part art (listening to your gut) and part science (listening to the numbers) – it’s about making informed decisions and scientifically A/B testing your hypothesis to make your landing pages perform better. It’s also a lot of fun if you really invest the time that it deserves. Hitting the refresh button on an A/B test stats page is incredibly addictive – a bit like a game, where the prize is money for your business.
If you want to play, you need to know the rules. To get you started, I’ve put together a quick list of things you should and shouldn’t be doing. Follow these guidelines and you’ll soon be optimizing your pages and beating your competitors.
Winning conversion optimization strategies
Gather feedback from the frontlinesWho knows the most about your customers? Perhaps the marketing department does. Who knows the most about their problems? Customer support. You can glean invaluable insight by talking to the people who spend all day on the phone with your customers. This can then be fed back into your new testing ideas. Follow this rule Ask your customer support team to create a prioritized list of the pages that cause the most problems (so you can choose which page to tackle first), and the issues that are occurring (to give you ideas for your test hypothesis).
Gather feedback at the point of conversionThis is similar to the first idea, but instead of asking an internal team, you ask your customers directly via tools like live chat or in-page surveys to uncover areas of friction at the point of conversion. Follow this rule Embed a tool like Olark (live chat) or Qualaroo (survey) on your landing page. For live char ask questions like “What prevented you from [insert description of conversion goal] today?” or “Can I help with any questions you have?”. Use surveys to ask multiple choice questions to establish a prioritized list of issues/concerns. For more information on using this strategy to inform your conversion optimization strategy, read this case study that used Qualaroo to survey customers (and gain some surprising insights). For more user feedback tools read “The Top 10 User Feedback Tools for Improving Conversion“.
Develop a hypothesis before starting your testA hypothesis is a reasoned statement of what you are going to test, why you are testing it, and what you are looking to achieve. It’s the foundation of how to alter/design your test variant page and is developed either from data learned from the two ideas above, or from a planning session (brainstorm) with members of your team.
An example hypothesis “The page fails to clearly communicate the value of the offer, and prospects struggle to understand how they will benefit from accepting it. By clarifying the value proposition more prospects will understand the value of the offer and perform the desired action on the landing page.” Read this post on developing a conversion optimization test hypothesis.Follow this rule Involve people from different departments in your planning sessions to gain perspectives you might not have considered. There are many things to test on a landing page, but you need to plan not only what to test, but what types of change you’l make (position, messaging, size, contrast, simplification etc.)
Run valid A/B testsA valid A/B test ensures that your ideas are given proper consideration. Without a valid test, you can come to incorrect conclusions that only create problems down the road. Imagine having a great idea, running a test, seeing a big lift in conversions – only to find out that had you let it run a little longer, you would’ve in fact seen it fail – instead you’ve already implemented the change and are losing conversions. Follow this rule Make sure your test adheres to the following criteria:
- Wait for at least 100 unique visitors to each page in your test
- Run the test for at least a week, preferably two, to ensure you cover daily variations in behavior (weekdays are often different to weekends)
- Don’t stop the test until it reaches at least 95%+ statistical significance
Simplify your landing pagesSimpler is pretty much always better. As you learned on Monday (if you read my post on CTA placement), the 5-point punch can be used to make your landing page short and simple by incorporating the right elements. When talking about simplicity, relying on a small subset of informational elements can help you focus your design and testing strategies. For a recap, here’s the 5-point punch diagram once more: Conversely, if you require a long page, it showed that having less information at the top of the page can increase the likelihood of people scrolling down to read the rest. Follow this rule Follow the wise words of Steve Krug: Delete 50% of the copy on the page, then remove half of what’s left. In other words, don’t be verbose. Analyze every sentence to see what can be simplified or removed.
Optimize the post-conversion experienceI’ve said it before, but confirmation pages are still very underutilized. They are visited every time you get a conversion, so you should be paying close attention to what you say, and more importantly ask people to do on them. Examples of what you can use your confirmation page for:
- Ask someone to follow you on Twitter, Facebook
- Give something else away for free as a bonus
- Suggest something similar they might like to buy (for ecommerce pages)
- Ask for an email subscription
Create great content to improve lead gen conversionsContent is not King anymore. It’s more of an Emperor. Inbound marketing has taken over in smart companies as the way to drive quality traffic and generate relevant leads. So how is it used to increase lead gen conversions? Essentially, it’s about producing super helpful and educational material that can’t be found elsewhere (or is better than what others are saying/doing). The key to conversions is balancing the content acquisition friction (the size of the form), with the perceived quality of your content. Follow this rule To increase conversions on your content-driven lead gen landing pages, add a preview of a key part of the content. People love the try-before-you-buy concept, and providing a preview (like Amazon’s “Look Inside”) instills trust that you are not hiding shoddy work behind a form wall.
Discuss your successessTesting is hard. Not everyone has the skills or ideas to make it a success, and not every good idea ends up being successful. But a key to producing success is to be transparent. Sharing successes gets the whole company excited and gives more weight to, and belief in, your marketing optimization team, and the concept of optimization in general. Follow this rule Another key concept of transparency is that you should also share you failures. This can be a really useful strategy in finding the next idea that will create success. This works twofold: firstly, by talking/sharing you may get a lightbulb moment about what went wrong, and secondly, you may get an idea from someone else who can see where you may have gone wrong.
Track your results to create case studiesWhen you finish a test, you MUST keep a record of the results. Too often a new test starts and you can lose data or even the rationale behind your test. This is something you need to do before, during and after the test. Your documentation should include the hypothesis, test results (screenshots and written stats) and the lessons learned (have a post-mortem meeting). You can then use this information to create in-house case studies to show stakeholders and management as required – or use as content to share with your customers so they can learn fro your experiences. Follow this rule Develop a template for recording all of this information to make it easier to gather and keep track of for the future.
Remind yourself that every page can be betterThis doesn’t need much explanation. Just keep telling yourself that the optimization and testing process is never really over. Yes, there is always another page to optimize, just don’t forget about pages you’ve tested. A break may be good to allow you to come up with new ideas, but remember to revisit important pages to test them with new ideas as you continue to learn.
Losing conversion optimization strategies
(Don’t) Listen to HiPPO’sSounds silly, but HiPPOs are real – even outside Africa. If you didn’t know, it stands for the Highest Paid Person’s Opinion, and is one of the biggest problems that conversion optimization aims to solve through A/B testing. If your boss says it should be one way, and you’re not in a position to argue, you can test it to find out what your customers respond to more favorably. This does two things: first it removes doubt from the argument, second it can free you up to go with your own ideas in the future as the HiPPOs will learn that they’re not always right.
(Don’t) Change more than one thing at a timeWhen running an A/B test, there are two approaches. One is to do a big band change, where you change many page elements, layout, design, messaging etc., all at once. This can be very effective in producing a big win – or a big fail. The problem is that even if you succeed, you won’t know which element had the most impact. By testing one thing at a time, you can run a series of structured and strict tests that educate you over time, with definitive insights from every test. Choose your poison, but I recommend just doing one thing at a time.
(Don’t) Let your visitors leakWhat this refers to is giving your visitors too many options. Each link that isn’t based on your conversion goal should be renamed to be a “leak” as it can allow people to leave your page or desired flow. In short, don’t put any more than one link on your landing page (this is most often a button).
(Don’t) Be greedy on your lead gen formsThis goes back to producing great content. Asking for too much, or too personal information on your lead gen forms is a conversion optimization killer. If you only need an email (most often this is all you need), then only ask for an email. If you do need more, then ensure that what you are offering in return is relevant. for instance, it’s okay to ask for a phone number if you are getting someone to register for a consultation, but not for an ebook.
(Don’t) Stop testing… everNuff said.