This article previously appeared on marketingfacts.nl (by Bas van de Haterd)

Last Thursday there was another Online Dialogue Thursday. An afternoon where the experts of Online Dialogue share their knowledge with the market and, among other things, look at websites of visitors from the different personas and give improvement suggestions. There is always a case, and this time one of the internet marketers of ABN AMRO, Jorden Lentze, talked about conversion optimization at ABN AMRO. He was not shy to share facts and figures from practice which made it a very inspiring case.

At ABN AMRO, they test everything. Some great quotes from Jorden included: best practices mainly work on others and “you have to avoid the highest paid person opinion counts syndrome”. Not the person with the highest salary should determine what the site looks like, the visitors determine that based on their behavior.

Testing also keeps your feet on the ground. Only 2/3 of really good ideas turn out to work. So not 2/3 of the ideas, but the ideas that have been really well thought out and the whole team is really totally convinced it's great turn out to work. Another interesting fact is that all untested pages improve on average by 40%. So that's average, some pages can improve by as much as 100% or 200% by testing different elements.

Results

Some unusual results shared with us included that different days of the week different colors pay off better. A case of SNS also came up (not by ABN AMRO but from the audience) where it had been shown that a green button worked better on the weekend, a red one on Monday, Tuesday and Wednesday and on Thursday and Friday it didn't matter. ABN AMRO has also seen similar results, also with, for example, the text you put in a button. At ABN AMRO, they noticed that a different text in the button performed better on weekends than during the week, as did different colors. The problem here is that your system has to support this, because most CMS systems do not have the possibility to show different texts or colors on different days. This is being worked on now, though.

Another extraordinary result was that different levels of education show different reading behavior. People also did eye-tracking research in usability tests, and there were clear differences where and how much higher educated people read compared to those with secondary education. They are now looking into whether and how this can be incorporated into a website.

A conversion myth is that less distraction makes for better conversion. ABN AMRO tested, against all corporate identity rules, to omit the ever-present login option from Internet banking. That led to fewer requests at the application form. Apparently it gave confidence, although one cannot test the reasons.

Not putting yourself at the top of a comparison row also led to much better conversion. In a comparison of different interest rates on loans, they tested with five or six competitors. Among the five, ABN AMRO was at the top; among the six, one party (unknown name, but subsidiary of major bank) was still above with a lower interest rate. The conversion on the page where one was not on top as the best was 50% higher.

Online - offline

Finally, Jorden wanted to share a nice case with us that showed that it is not always as easy to measure as many would like to believe. For this he used the figures from a Google adwords campaign for car insurance from ABN AMRO. Here they tested a variant that had a phone number and one that had none. That phone number was unique to that campaign, so one could say with certainty that the callers had come in through the Google adwords (unique landing page, unique phone number). It turned out that the number of people who called and requested a quote offline was 2.5 times the number of people who requested the quote online. So in total, the campaign had 250% more return on investment (quotes was the goal of the campaign) than would have been measured before if they had not deployed the unique phone number. Estimates are often made about this, but this concrete measurement makes the numbers a lot more tangible.