Skip links

Influencing Human Behaviour in CRO Testing

All CRO tests are centred in influencing human behaviour. And whilst it can be easy to overlook this when the actions you take are on a screen away from real people, it should never be forgotten.

How do we influence human behaviour?

To drive an alteration in human behaviour, we have two directions to take:

  1. Reflect visitors’ existing behaviour to make their actions easier
  2. Change visitors’ existing behaviour to gain a more beneficial outcome

Of course both methods can deliver positive results. But understanding the differences between them will enable you to better plan for, and predict, the likely impacts of your tests.

Reflecting visitor behaviour

Tests that reflect existing behaviour are generally focused on speed & ease. Historical data suggests that a significant percentage of visitors are already undertaking these actions & therefore by making them faster & easier to achieve, you can increase the total percentage of visitors who do so successfully.

Here are three examples of such tests & their backgrounds:

  1. 60% of visitors proactively login once they are in your checkout process – therefore test move the login step into the main flow of the checkout
  2. 75% of visitors proactively select an annual term for one of your subscription products – therefore test setting this product’s term to Annual by default
  3. Visitors selecting from their Past Orders are twice as likely to convert than those who don’t – therefore test showing the Past Orders page to all visitors who login

Tests of this nature generally represent fairly “safe” bets. You know that a decent proportion of visitors do these things already and/or there is a clear conversion advantage from those who do take those actions, and therefore your confidence in the outcome is high. The strength of these ideas exists in the historical data you have to back them up, but therefore also relies on having that data available!

The downside of these tests is that they often represent a lower potential return. With a sizeable proportion of visitors taking these actions already, the scope to improve the overall audience’s conversion rates is lower. As such, most test concepts that rely on reflecting existing behaviour provide low-risk, but low-yield in terms of global site performance improvements.

Changing visitor behaviour

Tests that look to alter current visitor behaviour are focused on pushing visitors to do things that the vast majority of them are not currently doing. As a result of that aim, it usually means that historical data does not exist to prove the value of that new behavioural action. In the most basic sense, the data you have only exists to show what they are not doing, and therefore there is only a reasoned hypothesis rooted in human behavioural norms that backs up the validity of the test.

Here are three examples of such tests & their backgrounds:

  1. 80% of visitors currently buy the bottom-tier package of your subscription product – test increasing the price of that package to increase the comparative perceived value of the next tier up
  2. 1% of visitors to your blog proactively subscribe via the Subscribe CTAs you provide – test prompting visitors to subscribe via an overlay once they reach a certain scroll depth on your posts
  3. 0% of visitors currently purchase the brand-new add-on product you have created as it isn’t live on the site yet – test adding a new standalone upsell page into the checkout process to push that new product

Tests of this nature are inherently riskier than those that reflect existing behaviour as there is far less historical data to support the likely impact. This does not of course mean that they are not data-led, but merely that the available data is telling you what is not occurring right now and that there is a business value in getting visitors to complete those actions.

The strength of these ideas therefore is in the opportunity size. Because these ideas are based on what the vast majority of visitors are not doing, the opportunity size is significantly larger than in tests that reflect existing behaviour. Rather than testing a change to get 75% of visitors to do something vs. the current 70%, you are more likely to be testing a change that could deliver 5% of visitors to complete an action vs. the current 1%. As such, most test concepts that rely on changing visitor behaviour provide higher-risk but higher potential yield in terms of global site performance improvements.

So what then?

Always the question a CROer must ask!

Being cognisant of the different types of behavioural influencing helps you to better balance your CRO programme. Depending on the status of your programme, you can select the right sort of test to deliver on your short- & long-term objectives.

For example, a programme may come under pressure if it is perceived to have a low success rate, and therefore the budget for it may come under scrutiny. If this is the case, you may benefit from prioritising tests that reflect existing visitor behaviour as they are significantly more likely to increase that success rate to help restore wider confidence in your programme.

Conversely, a programme with a very high success rate can come under pressure if it is perceived to not be generating a significant ROI, and therefore budget may come into question from that angle. In this case, you may find that prioritising tests that seek to change existing visitor behaviour could help to deliver that “big-bang” result that brings everyone back onside.

Join the Discussion