{{ searchResult.published_at | date:'d MMMM yyyy' }}

Loading ...
Loading ...

Enter a search term such as “mobile analytics” or browse our content using the filters above.


That’s not only a poor Scrabble score but we also couldn’t find any results matching “”.
Check your spelling or try broadening your search.


Sorry about this, there is a problem with our search at the moment.
Please try again later.

Many businesses have dipped their toes into the world of online testing, and sometimes expectations of high returns from simple tests can outweigh reality.

The truth is that there are genuine and continuous benefits to be had, but you should think outside the box to drive real change.

High volume sites like Google and Amazon are known to run extensive content tests. In fact the former was rumoured to have tested 41 different shades of blue for its search links to establish which shade we found the most clickable.

It has such enormous traffic that this test may have proven fruitful, but most sites should invest their time in more customer-centred areas.

1. Test your buttons, and move on

The most common AB tests seem to involve buttons. If someone tells you they doubled sales by changing the colour of the buttons, ask them for the sample size of the test and the full journey stats for those visitors. It's very unlikely to be true. 

That said, you can and should drive improvements in page conversion with button redesigns, and ultimately improve sales.

So change the colour, size, and content of your button and measuring uplifts in click through - simple. These tests are pretty easy to setup, and can (and do) deliver quick benefits.

That said, if you've run a button test recently I'd bet that the following attributes won in the majority of cases:

  • Larger button design.
  • More prominent placement on the page.
  • Higher contrasting colours.
  • Increased urgency within text.

Be realistic when you plan to run a test like this. There is a limit to how much a button can influence someone to click on it. Sometimes when you analyse results fully, you'll find that you increased clickthrough but that those additional customers subsequently dropped out on the next page. Ensure that any improvements are genuine.

2. Run simple but bold AB and multivariate tests

It's possible that any web change will have an impact on your customer KPIs, but generally it's the more daring changes which drive the interesting and usable results. The following aspects are all relatively easy to test, and are likely to have some interesting impact on visitor behaviour and perception:


Don't swap a photo of call centre agent Sally for one of Margeret, it’s not going to teach you anything meaningful in the long term. Go for a theme and test corporate imagery versus non-corporate, brand imagery against product imagery, or pricing focus over service focus.


The best results can sometimes be found by using urgent or persuasive text around your call-to-action points.

After that think about; language styles (all the way from informal up to corporate language), length of text, key messages (product versus brand and then service) and how prominent you make those key messages.

What you call your products can drive behaviour changes too, using words like ‘recommended’, ‘most popular’ and ‘essential’ often increase uptake (don’t try to fool people, only say ‘popular’ when it really is).

Contact options

Altering the contact options you offer, or even showing/hiding some altogether can have surprising results.

I’ve seen some tests indicate that showing phone numbers on every page of a sales flow increases pure online conversion, perhaps because the customer is reassured that they could call if they have a problem.


This can be powerful but the changes need to be well planned. Re-ordering page sections, making key sections larger and removing complex looking elements can work well particularly if the user is making a purchase decision based on complex information on that page.

The following suggestions are more technically advanced, but have produced some of my most successful performance improvements.

  • Changing journey length: one page checkout vs three page checkout with a progress bar.
  • Defaulting: meaning you pre-set values in fields rather than presenting the customer with a blank form. Think about shipping options, extra cover and popular payment options.
  • Offering incentives or product options to some customers to understand their true impact over a control group.

3. Move on to behavioural targeting

The best way to think about behavioural targeting is this: if you ran a highstreet store, you wouldn’t say the same thing to someone walking in for the first time as you would to someone who just bought your product. Their needs and expectations of you are totally different.

If someone is new to the site, sell them something. Talk about the benefits of your products and how much their life will improve with your product or service.

If someone spent half an hour on your site reading about plumbing parts, they’re probably doing some plumbing. Make sure you show relevant content on their next visit and don’t blast them with messages about your awesome new range of fencing.

If someone already bought and came back to your site, show them complimentary products (in most industries they’re unlikely to want exactly the same thing again), or encourage them to log-in and use your online service functionality.

Similarly if they arrive from search keywords which contains words like ‘repair’ or ‘complaint’, should your landing page really talk about how fantastic your new range is? Help them out by immediately offering content to resolve their most likely query.

Unsurprisingly, this can boost conversion down key journeys and reduces the amount of time people spend looking for what they want.

4. The big one: personalisation

Visitors engage more when presented with things that align with their interests, their sociodemographic segment or their current requirements.

Generally, pushing content which aligns to one of these three things will create a more engaging user experience, and often this flows through to your key metrics: sales, service, retention etc.

Personalisation campaigns might be a technically demanding journey, but can deliver some of the most impressive uplifts in KPIs.

Amazon appears to invest in this area. The products you see when you visit and in their emails are statistically matched to what you’ve bought, added, and browsed before.

If your site has log-in functionality then this is easier to achieve because you can identify visitors quickly, but it’s not essential.

Some easy quick wins include:

  • Show the relevant currencies and only the products/delivery options applicable to that region. It doesn’t make me want to buy when I see that you offer free shipping in other countries but not to mine.
  • Adhering to local languages (English and US English are not the same!)
  • Matching content to the time and day, particularly weekends and public holidays.

More technically complex, but highly rewarding methods:

  • Show targeted cross sell promotions based on the visitor’s spending history with you. People who buy mp3 players might be more likely to buy earphones than another mp3 player - the clues are in your old sales records.
  • Show copy based on their customer segment. Teenagers don’t talk the same way as pensioners, so talk to them in their tone (test to find the right one).
  • With service based products, work out a customer's estimated value based on their account history, and tailor your offering to ensure you convert those high value customers without giving away too much to low value prospects.
  • Match relevant messaging to location. If the visitor appears to be in Peterborough and wants breakdown cover, tell them how many breakdowns you sorted in Peterborough last year, and your average response time in the area. Stick up the name and photo of the local branch manager - make it feel personal.
In conclusion, it’s impossible to know which tests will deliver value for any business. The only way to find out is to run small, bold tests and react to the results.

I’ve not talked about technology here, but you don’t necessarily need expensive testing software. Some of the above can be set up by a savvy front-end developer and measured using your standard web analytics tools. 

Dan Huddart

Published 24 January, 2012 by Dan Huddart

Dan Huddart is Web Analytics Lead at RSA Group and a contributor to Econsultancy. 

1 more post from this author

Comments (0)

Save or Cancel

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Daily Pulse newsletter. Each weekday, you ll receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.