{{ 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.

No_results

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.

Logo_distressed

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

A colleague of mine, Arjan (@arjanharing), is a great table tennis player. And when I say great, I mean great: I have witnessed a table tennis tournament in our offices during a BBQ with over a 100 people, and nobody – repeat nobody – was able to return his service. 

Now, I must admit that his service is definitely his main strength in the game. I mean, he is a good player, but the exceptional service makes him a player who is hard to compete with even for the well trained and the profs.  

But recently, “they”, the people who run the table tennis rules, decided to change the rules of the game. 

The change might seem minor to you and me (assuming here that you are, like me, not a trained table tennis player): they changed the way in which you are allowed to hold your hand when throwing the ball into the air to serve.

When I heard about it—and tried it—I could not really tell the difference. However, to Arjan, a trained expert, the little change of rules made him loose his extraordinary skills. A drop from exceptional to mediocre, by a little change of the rules. 

Now, why would you care about changes in the rules of table tennis? Well, that’s because the emotions that overwhelmed Arjan when notified of the change share lots of similarities with the emotions that are overwhelming internet marketers all around the world: “They” are changing the rules of the game.

Selling stuff online, getting people to use your services, and creating profitable business about two years ago started to looked like a manageable task.

Get yourself some good products (or build a great service), attract people to it by using your SEO and SEA skills, and finally optimise the layout of your webpages using A/B or multivariate tests.

Do all of it well, and at one point in time you will have a service or e-commerce store that runs itself. You have your money making machine that will attract customers and sell them things indefinitely.

Not true. And most online marketers are starting to understand that this is not true. Not anymore: they changed the rules of the game.

So what are those changes I am talking about?

Well, I think there are three big things happening in online marketing and selling that everyone should understand, three manjor changes to the rules.

First of all, your understanding of that A/B test, or even the multivariate tests that you were doing to optimise your game needs to change.

While many have been engulfed by design => test => decide cycles this is by no means the most optimal way of using the different design ideas that you have. The A/B testing cycle totally ignores the trade-off between testing (to learn new things), and exploiting (using the knowledge that you have). See also this discussion by Myna’s director Noel Welsh.

Our objective should not be: “Which is better? A, or B?” but rather: “Given that I have A and B, how do I use them wisely to contribute to my bottom line?

This first shift changes how we think about our products and service since the aim becomes optimising over all the design choices that we have instead of trying to find a “single best one”. The latter of which is a fallacy to begin with.

AB test. Printscreens taken from Threadless.com
Fig1: “Which is better? A, or B?” will soon be an obsolete question. (Example taken from Threadless.com)

Second, we are stepping up our game in how we use our “big data”. Whether it actually gets big or not is not the question here, the question is what we do with the data.

We used to log everything our customers and users did, and then produce a great report for somebody to look at and go “Wow, that is interesting”. But intriguing an online marketer inside your company was never the main point of collecting that data. We should actually act on the data as it comes in.

If one of your products sells better, make it more prominent dynamically. If one type of sales pitch resonates with (a part of) your customers (e.g. “This is a bestseller!” see also the TechCrunch article Mass Persuasion, One User at a Time) make sure you use it more right there and then. Use the data to act and serve the outside world, instead of to amaze and baffle your inside team.

Using data this way comes with a note. While in your A/B tests you could test anything you like: button colors, rounded corners, margins, etc. etc., and in your “big data” endeavors anything could be plotted in hind-sight, you have to be selective in what you act upon if you are really having the data inform the design and interaction of your product or service.

You will have to choose to focus on things that actually affect your users, which might not be the rounded corners.

You have to start understanding things like persuasion and habit formation, and the core drivers of behavior. A good start would be the famous Cialdini persuasion principles for example. Those are the ones you want to log, and act upon.

Cialdini Principles applied to Threadless.com 

Fig2: Instead of testing rounded corners “focus on things that actually affect your users” like the Cialdini principles of Scarcity and Social Proof. (Example taken from Threadless.com)

Finally, the game is changing from a game of averages to a game of individuals. That AB test was addressing groups of people. Perhaps segments, but at all times groups of people.

However, just like table tennis services are unique, even over the top players, so is the behaviour of your users (see also the science behind this Heterogeneity in the Effects of Online Persuasion).

And, you are able to track it, log it, and, see above, act on it. So why tailor the design of your e-commerce store or service to the masses, and not to your users? If you understand the variable that influence decision making, such as the key tactics of persuasion, you can monitor how effective they are for individual users, and make sure to select the correct one for that specific user.

Just like in table tennis, to the outsider the new rules do not seem to affect the game that much. If you don’t have a specialist serve that is affected by the game change than why would you care?

On the other hand, if you are not affected, that means you were—like me—at most a mediocre player. While that is OK for table tennis, it’s not what you want to be when running your online business. You want to be affected by the changes in the game. 

All too often I talk to entrepreneurs or marketing managers of large online vendors about the changes above and they seem like minor changes. Nothing that affects their game.  And we talk again, often weeks or months later, and suddenly reality has hit:

By losing out on minor change after minor change, they find themselves in a totally different league. And, it’s not the league you want to be in.

Maurits Kaptein

Published 27 November, 2012 by Maurits Kaptein

Maurits Kaptein is Chief Science Officer at Science Rockstars and a contributor to Econsultancy. 

4 more posts from this author

Comments (6)

Comment
No-profile-pic
Save or Cancel
Avatar-blank-50x50

John

Hey....

I remember you couldn't return my serve either! ;)

Nice Article. Agree that personalisation works better than the average.

Amazon are a great user of this, but also factor other "environmental" events etc as well. e.g. If it's raining, and user a comes to site. use this banner.

Seems to work for them...

Hope all is well Sharif!

Regards.

Canon - John.

almost 4 years ago

Chris Ellis

Chris Ellis, Group Digital Marketing Manager at Belron International LtdEnterprise

Not sure how the table tennis example illustrates the point. In the case of table tennis "they" really did change the rules. In the case of online marketing we have discovered more complex ways of optimising yield. Given that online marketing is often only just cost effective in competitive areas then yes, you need to keep up or risk losing money on your marketing spend.
Are you possibly missing the point about the aggregation of data? - Amazon may adapt their offering to each individual, but only because they know the results of many many tests. You need enough data points to make a decision at any scale. "you can monitor how effective they are for individual users, and make sure to select the correct one for that specific use". Can you really? Or can you do what Google and Amazon do and base personalisation decisions on vast amounts of collective data? E.g. the big A/B test shows that social proof improves conversion, therefore you show social proof relevant to an individual. But you will never have enough data to know that social proof makes John Smith more likely to convert.

almost 4 years ago

Avatar-blank-50x50

Karissa Davis

I agree with the above comment. You have to have such a large sample to implement this, that many small businesses will not have that information. The best thing that they can do is watch how the major online players change their sites and to learn why they are doing it. This will allow you to move with the times, but not cost you the time and money that it costs to do all of the testing.

almost 4 years ago

Avatar-blank-50x50

Maurits Kaptein

Hi Chris and Karissa,

Thanks for the comment! Yes, aggregate data is off course useful, and often the only place to start. However, given large heterogeneity in (e.g.) price sensitivity, cognitive style (See Hauser 2008) or responses to persuasion (Kaptein & Eckles 2012, Hirsh 2012) it seems worthwhile to focus on individuals if such data is available.

By using hierarchical models we can readily combine what we know at the aggregate level with datapoints at the individual level and thus borrow strength from aggregate responses to inform estimates about individuals. I am not saying that the individual estimates will ever be perfect, but let's consider your example (using social proof): We see that about 1/3 of people responds negatively to social proof if you obtain repeated measures experimentally. Thus, for quite a number of your visitors not using the social proof argument - despite its large average treatment effect (ATE) - will be beneficial. Once you have observed a "negative" response to social proof 2 or 3 times (e.g. no click, no purchase, etc.) you could opt for experimenting with different strategies for that user.

I am not at all motivating that we should not look at ATE's or draw comparisons between people. However, for many psychological constructs heterogeneity does exist, and we should be flexible in dealing with it. Often we actually do obtain 5+ observations of individuals (over multiple pages, emails, etc.) and hence we can try to adapt to individual tendencies. Again, not a perfect estimate or adaptation, just better than without considering the individuals history.

almost 4 years ago

Maurits Kaptein

Maurits Kaptein, Chief Science Officer at Science Rockstars

Hi Chris and Karissa,

Thanks for the comment! Yes, aggregate data is off course useful, and often the only place to start. However, given large heterogeneity in (e.g.) price sensitivity, cognitive style (See Hauser 2008) or responses to persuasion (Kaptein & Eckles 2012, Hirsh 2012) it seems worthwhile to focus on individuals if such data is available.

By using hierarchical models we can readily combine what we know at the aggregate level with datapoints at the individual level and thus borrow strength from aggregate responses to inform estimates about individuals. I am not saying that the individual estimates will ever be perfect, but let's consider your example (using social proof): We see that about 1/3 of people responds negatively to social proof if you obtain repeated measures experimentally. Thus, for quite a number of your visitors not using the social proof argument - despite its large average treatment effect (ATE) - will be beneficial. Once you have observed a "negative" response to social proof 2 or 3 times (e.g. no click, no purchase, etc.) you could opt for experimenting with different strategies for that user.
Hi Chris and Karissa,

Thanks for the comment! Yes, aggregate data is off course useful, and often the only place to start. However, given large heterogeneity in (e.g.) price sensitivity, cognitive style (See Hauser 2008) or responses to persuasion (Kaptein & Eckles 2012, Hirsh 2012) it seems worthwhile to focus on individuals if such data is available.

By using hierarchical models we can readily combine what we know at the aggregate level with datapoints at the individual level and thus borrow strength from aggregate responses to inform estimates about individuals. I am not saying that the individual estimates will ever be perfect, but let's consider your example (using social proof): We see that about 1/3 of people responds negatively to social proof if you obtain repeated measures experimentally. Thus, for quite a number of your visitors not using the social proof argument - despite its large average treatment effect (ATE) - will be beneficial. Once you have observed a "negative" response to social proof 2 or 3 times (e.g. no click, no purchase, etc.) you could opt for experimenting with different strategies for that user.

I am not at all motivating that we should not look at ATE's or draw comparisons between people. However, for many psychological constructs heterogeneity does exist, and we should be flexible in dealing with it. Often we actually do obtain 5+ observations of individuals (over multiple pages, emails, etc.) and hence we can try to adapt to individual tendencies. Again, not a perfect estimate or adaptation, just better than without considering the individuals history.
I am not at all motivating that we should not look at ATE's or draw comparisons between people. However, for many psychological constructs heterogeneity does exist, and we should be flexible in dealing with it. Often we actually do obtain 5+ observations of individuals (over multiple pages, emails, etc.) and hence we can try to adapt to individual tendencies. Again, not a perfect estimate or adaptation, just better than without considering the individuals history.

almost 4 years ago

Avatar-blank-50x50

Jennifer lopezus

Unique content and information make the site difference and hard to compete.But how, what type of content can i posted to target specific users.Average user better than high traffic because of due to low traffic sometimes can fall page rank.

over 3 years ago

Comment
No-profile-pic
Save or Cancel
Daily_pulse_signup_wide

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.