User feedback to improve Conversions

Often when working on CRO projects, it’s tempting to jump straight in and set up tests based on your own ideas and suspicions. But, for me, a more effective and measured approach is to spend time finding out what users think of a particular feature, page or product, and address any issues that they come up with, or place further emphasis on the things that they like. They are more representative of the people using your site, and often consultants or client teams are too close to a site to be able to look at it objectively.

I’ve put together a few quick, easy, and free methods of gathering feedback that can give your CRO projects a more solid foundation rather than basing your tests on assumptions.

Start with Analytics

But before you start collecting qualitative data from users, you should work out which parts of your site aren’t working well. User engagement metrics give you an indication of what users like about your site, and more importantly what they don’t like. A good report to start with is page bounce rates vs. site average. To view this report in Google Analytics:
  • Go to Content > Site Content > Pages
  • Click the Comparison button at the top right of the table of data
  • In the right-most column, select Bounce Rate from the drop down at the top.

As always, be careful with how you interpret bounce rates. A high bounce rate on a page where there is a desired next step, such as add to cart on a product page, is obviously a bad thing. On the other hand. a high bounce rate on an information page or resource might just mean that the user got everything they needed from the page and left, in which case a high rate could be interpreted as good. But in most cases, you should be able to pick out some pages that are genuinely underperforming.

This is just one example - think about other reports that you might be able to use such as traffic sources or your conversion funnel, and which other metrics might be useful to you, such as conversion rates or time on site in order to find weak pages or features.

Now that you know what isn’t working on your site, user research can help you to identify why.

Set up a survey

The most obvious way to get feedback. 4Q Survey is a great free tool that was developed to get the most useful feedback in an efficient manner. The free version allows you to collect up to 100 respondents per month which should be enough for most sites, but if you want more headroom you can set up a rudimentary imitation using Google Docs’ forms. I’ve set up an example here, which you can make a copy of (File > Make a Copy) and use yourself.

This survey is set up to get feedback about the whole site, so you may wish to tailor it to get information about specific issues you’ve identified above. Obviously you’ll need to find some participants, so leverage your email lists, social media accounts, or blog readers to drum up a sufficient number of responses. Google Docs allows you to embed the form on your blog or in emails.

Getting feedback from people who are already engaged with you through one of these channels can result in much more relevant feedback than paid feedback services, such as Feedback Army, whose random respondents are sourced from Mechanical Turk. Your followers = your potential customers = your target audience.

Here’s a great example from the Ordanance Survey blog, asking for feedback on the blog itself: http://blog.ordnancesurvey.co.uk/2011/01/tell-us-what-you-think/

Twitter

In most cases, a large proportion of a brand’s Twitter followers will be it’s customers and website users, so if you have a significant following it’s an ideal place to either solicit survey repsonses or ask for feedback directly.

If you’re asking for direct feedback, it can be specifically about a page such as “Just launched this page on our site - would love to hear some feedback” or it can be more specific to a product or service, for example “Why haven’t you signed up to our email newsletter? Would love to hear some feedback” or “Looking for feedback on our free trial - what did you think?”

 

Asking for feedback directly can result in plenty of responses because of the low commitment required by participants, although the downside is that it won’t be particularly in depth. You might also want to consider whether you want to stimulate potentially negative conversations on Twitter about your site or product. If this seems likely, the opaque survey approach would be better.

With Twitter, consider using a unique hashtag to make responses easier to collate for analysis. You can use Google Docs to archive the tweets automatically so that you have them in one place for analysis later.

Feedback tabs

Most readers of this blog will be familiar with feedback tabs such as User Voice. They offer a means of gathering feedback on an ongoing basis rather than the single hits listed above. But I’m not sure how many people consider using this as a source of information when planning CRO tests. What do people love about your site? What do they really hate? How can you address these issues?

Remember that users are only really likely to use this feature if they’re either pissed off about something that happened on your site, or they really love your brand, so it’s not a fair reflection of the average user. It might be best to take some of the feedback with a pinch of salt, and perhaps combine it with some other research methods.

Ask your support or sales team

Talk to the people in your organisation that have the most interaction with your clients or customers. They are the ones who are most likely to be aware of the common objections, complaints, and general feedback they pick up from spending time talking to users, so conducting interviews with them can lead to valuable information.

Survey Incentivisation

Typically I’ve found that you often don’t need to incentivise feedback if your brand is strong. Followers and newsletter subscribers already like you and are likely to help you out, especially when the outlay is minimal for them like a short survey. Moreover, many customers love to give their opinions.

However, if you don’t tend to get a lot of responses or need to guarantee them quickly, there are a few things that you can offer in order to attract participants such as the chance to win 1 of 5 Amazon vouchers (gift card for US), or an extension on a free trial of your product.

I attended a usability testing seminar recently where the subject of incentivisation came up, and the majority of attendees said that they had found that the chance to win an iPad, even if there’s only one on offer, typically worked best for them. Obviously this depends on your budget, and how many responses you need, but keep it in mind if you’re really struggling!

Analysis

Once you’ve solicited enough feedback, you more than likely be left with a whole load of text that you need to turn into something more useful and actionable. The most common approach is Thematic Analysis, which essentially boils down to reading through your responses carefully and annotating them to identify commonly occurring or strong themes. Read more about it here and here.

This should leave you with some great user insights to use as the foundations of your CRO work. You might be left with usability or information architecture problems to fix, or there might be clear concerns and objections from potential customers that you can address on your landing pages. The point is, at least the work is based on something rather than nothing. There are no excuses for running speculative tests or experiments based on your own personal whims!

I hope this post goes some way to demonstrate that this approach can be quick, easy to set up, and cheap. If you have any further ideas or thoughts, please let me know in the comments or catch me on Twitter.

Get blog posts via email

0 Comments

  1. There are currently no comments

Leave a Reply

Your email address will not be published. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>