How to make your Intranet like Amazon.com

Sites like Amazon use a range of techniques to show you personalised pages, content and recommendations. It’s possible to personalise intranets too, presenting different people in your organisation with the content that suits them best. What better way to save your staff time, while keeping them updated with the information they need most?

Personalisation for breakfast

Personalised content has come a long way since US food giant General Mills offered customised products on their website. For a staggering $20, their site allowed you to create your own customized box of cereal from a vast array of choices and recommendations (do you want your marshmallow pieces star-shaped or smiley face-shaped?).

This used to be what personalisation was all about. Fortunately, content creators have learnt a lot since then.

Amazon – our trusty role model

Moving on to more positive things, we can turn to the trusty Amazon for relief. They’ve been doing personalisation for a long time and have turned it into one of the main sales engines of their business.

We’ve all experienced it. You search for a CD title and in the search results you will also see something like “customers with similar searches also purchased these titles…”. They then encourage you to listen to 30 second samples of each song on the recommended CDs and before you know it, you’re at the checkout having spent more than you planned.

Is it possible to apply the Amazon model to your intranet?

Nowadays, many organizations allow users to login just once at the start of the day and get access to all the systems in the organization. In other words there’s no longer the need to remember lots of different passwords.

Couple this with the huge growth of content and systems, and the possibilities offered by Amazon-type personalisation become both attractive and attainable.

There are three types of personalisation

The personalisation space is full of jargon propagated by the vendors and consultants. To make matters worse, jargon is often used inconsistently: one vendor might call a spade a spoon whereas another vendor will call a spoon a fork. Before setting out, you need to have a clear understanding of what personalisation does and how it works. You can also take a look at our “Pros and Cons table” [below].  

Personalisation Type One – Tell me what you want and I will show you what I’ve got

As a web user you often come across sites that try to win your loyalty by encouraging you to spend time expressing your tastes, choices and preferences. For example, on My Yahoo! you might check a box to record that you are interested in politics and finance.

You might also customise the page to show the specific stocks and shares that you own. In other words you explicitly tell the site what your preferences are. This is the simplest type of personalisation and I am going to call this “customisation”.

Personalisation Type Two – I know you’ll love this. It’s right up your street!

The second type of personalization I will call “individualisation”. If you visit a site several times, over time that site can build up an understanding of your individual tastes and preferences based on your behaviour. It can then use this understanding to suggest to you content or products that you would find interesting.

For example, if you are always reading news about Formula 1, the website will automatically present Formula 1 related content to you. These recommendations are implicit – your behaviour on the site implies that you are interested in Formula 1.

The advantage of this is that you don’t have to spend any time telling the site what you like. The difficulty is that the site needs time to build up a profile of each user. In turn, it is necessary to define complex rules that match content with profile.

Personalisation Type Three – users who bought X also bought Y

Amazon likes to compare your tastes with those of other groups. Because of the mass of purchasing data available, Amazon is able to spot patterns and use these to suggest what you might like to buy. I’ll call this type of personalisation “Collaborative Filtering”.

This approach works well if there is a high volume of traffic and transactions but loses its effectiveness with smaller visitor groups and with more specialised content. You can see therefore that it might be less effective for intranets.

However, this could be overcome with intranets by blending collaborative filtering techniques with customisation: it’s possible to utilize the tastes of the minority who spend time customising their intranet page and then apply these tastes to users who don’t.

Build accurate profiles about your users

You will no doubt have spotted some challenges that apply to all three of these personalisation types. It’s important to have some fundamental building blocks in place before personalisation is possible.

Firstly, you will need to have basic profile information about your users so that you can segment your audience into manageable chunks. For intranets, this could include department, division, rank, geography etc.

But you could add further information depending on the type of personalisation you are implementing. For example you can give each user a “propensity score” which reflects the likelihood he or she will be interested in, say, the financial information on your site, or the marketing material. This score allows you to compare users and group them together into segments. The concept is fairly straightforward but the maths gets quite complex (although no match for a computer).

Tag your content or be damned

A second important challenge is content tagging. Content needs to be given relevant labels such as “finance” or “marketing” etc.

Taxonomy is a type of labelling that is gaining importance on today’s intranets. Taxonomies are tree-like information structures that can be used for classifying content. For example, you could have a branch called “Supply chain”. Within that branch you could have further smaller branches called “Procurement”, “Manufacturing”, “Transportation” and so on.

Cool rules hold it all together

The third and final challenge with personalisation is rules.

When all of your content is beautifully tagged and your users have rich profiles that define them, you need to create logical rules that match content with profile. These rules are built using knowledge of your users gained through discussions with them and also studying traffic data about them.

It’s perhaps the most complex part of the whole personalisation exercise but sophisticated software is available to help build collaborative filtering rules automatically.

Are you really ready for it?

Personalisation is a complex area but when correctly applied the payback can be big – especially on large intranets where the content mass is overwhelming users. There are no secrets behind successful implementations, just some sound common sense and a few ground rules:

• Do lots of research and planning – talk to your users!
• treat it like an ongoing task and not a project
• get the profile and labelling fundamentals right first
• start with simple personalisation techniques and build on these as you learn.

 

Finally, perhaps one of the most important questions to ask yourself before getting started: do you need it? In an article, usability guru Jakob Nielsen claims that “Web personalisation is much over-rated and mainly used as a poor excuse for not designing a navigable website” – do make sure you’ve got that piece right before setting out.