How to Personalize the Web


Written by Martin Stiksel, co-founder of

I recently got to thinking about how I engage with the web, and how much more compelling the experience would be if the web engaged more actively with me.

The way it works with so many of the sites, networks and applications that Om and Seesmic founder Loic Le Meur make mention of is that you adapt yourself to what they offer. In that sense, it’s initially all about search — finding the YouTube clips you like, for example, or the RSS feeds you want to collate. Next it’s about having a conversation, using the likes of Gmail, IM or Twitter. Then you add those things together – having found the stuff you like, you share it with those people you connect with.

But how do you find anything worth sharing, and what if your friends don’t use the same applications? In that case you’re Twittering about nothing to no one. To make it interesting you have to adapt yourself to the system, to actively engage, something many of us have neither the time nor the inclination to do. No point in sharing for sharing’s sake. How many Facebook application invites do you ignore because, well, you can’t really be bothered anymore?

But what if all this stuff adapted to you, rather than the other way around?

How would that work? It would rely on everyone having a profile, or personal ID, that they could take with them when they explore the web. users get one that reflects music taste just by listening to music. We work out what each user’s personal music preferences are based on their listening habits (we call it ‘scrobbling’) and use it to recommend new music they might like.

Now imagine taking that music ID and applying it everywhere you go on the web that’s music-related. Developers on, our third-party application showcase, are working to make this happen. One app listed there sorts your Facebook friends according to musical taste using your profile, for example. Another finds eBay auctions suited to your music preferences.

When this kind of portability grows beyond the developer level, you should be able to make the web adapt to you instead of the other way around. What if you could use your profile or personal ID from social network A to customize web sites B through Z? What if, say, every online newspaper automatically filtered stories according to your profile? Which sites would you like to take with you on your travels through cyberspace?



It would be great if all of this integration could come ‘plugged in’ to openID.

I use openID ever single chance I can get because it doesn’t require me to have a profile at tons of different places.

I see the web as a loose nit place where you have fifty million accounts at fifty million different places. Wouldn’t it be great to have one signin and one account to maintain that you could take with you everywhere?

That’s the point of openID and if a ‘profile’ of sorts could be attached to openID and you’d bring it with you…i think it would take the world by storm.


One of the best articles i have read. has really good innovation..


Er, isn’t this what Facebook is looking to do with all our demographic data anyway?

Chris Osborne

@Mike In my mind this type of thing is in addition to what already is offered by the destination site (from a discovery/navigation perspective), e.g. a dynamically created component in the page, or some kind of overlay, or through the browser toolbar/extension. So it creates an additional form of navigation/discovery, as opposed to a replacement. And from my perspective this isn’t about targeted advertising – it’s about exposing to the user the stuff on that site that’s directly relevant to them and so of most interest.

@Martin How do you envisage it being implemented?

Ian Duncan

Seems to me you can kind of fudge this already. Facebook, for me at leasts, acts as a place all my friends come together. From that I can then add applications for my Digg profile and Twitter updates can be fed straight into my facebook status.

The weakness of facebook is that it’s quite imperial in some respects. Surely much better to use Flickr and YouTube for photo and video sharing but at the moment that’s not really practical.


So, this idea – technically is just an expansion of cookies; except the cookie now knows of reference a lot more about you – not just ID but likes and dislikes, and customization options. There are obvious privacy concerns which would need to be addressed; as this is sort of the holy grail of targeted advertisement – a wide ranging description of likes dislikes and interests.

For security/data protection applications, this is already being worked; policy based access controls are not so different, with data being available based on a users attributes (security clearance, location, date/time, etc.).

What really concerns me is the lack of random discovery; if everything is selected to fit your stated likes, how do you find new things? Algorithmic selection can only go so far before feeling limited and constrained; this is one reason why so many people still enjoy editorially selected and developed content. Half of the New York Times or Slate articles I wind up enjoying are not the ones I would decide to read entirely based on existing/defined preferences.

Chris Osborne

Really interesting piece, Martin.

Definitely think that this type of on-the-fly personalisation based on user-data/profile portability will tranform the experience for users, and the proposition that sites/services have to offer (in terms of massively increased relevancy and therefore interaction potential). And it’s obviously interesting in relation to all forms of discovery, whether social (i.e. people), content, or products/servives (as you say).

Guess the main challenge (beyond mutual adoption) is in the site-side plugin/data interaction for the dynamic personalisation to take place (whether done just through data standardisation or also in conjunction with smart search technology). Great to see that is pioneering this from a music perspective through your developer platform. Microformats obviously represent an enabler in some areas (well, in respect to the data standardisation bit at least), and Yahoo’s recent announcement of support for microformats in their shopping proposition (millions of sets of products info being microformatted) is interesting, and potentially a building block for this stuff in that space.

We’re very into this as a theme for, which is a shopping utility which has various social angles (characteristically similar to delicious, but specifically for products, so providing a universal list of the things that you want to get from any shop/site on the web). So the application of this approach would be that when a user visits an online store their boxedup profile/lists travel with them and they can automatically see either specific things they want that are available from that site (and what the price is), or if there are things that are related to the things they want/the types of things that they’re into. So your delicious example is directly relevant, but in the context of shopping rather than news. Obviously it’s a non-trivial challenge, but it’s the type of thing that’s needed on the web to make the shopping experience more coherent and intelligent (and shopping is obviously a significant, and fast growing, online activity). We’re a very early stage start-up, and are keen to get people onboard who have this type of shared vision (in terms of backing and guidance), so if this is a space that you’re interested in then we’d love to talk to you! (Sorry for the plug, but hopefully I’ve also contributed to the debate here)

Anyway, thanks for the interesting piece. Think this type of dynamic/on-the-fly (but passive) personalisation will be a central theme in the development of the web – it definitely needs to be in order to reduce the barriers to engagement for users on new sites/services; it’s openness with very tangible payback potential for participants.

PS Om – I love the type of discussion and debate you get here

David H. Deans

Adaptive multimedia and lifestreaming applications seems to be coming together. The sharing of a lifestream, with friends and family, will include all manner of digital media assets

Loic Le Meur

Thanks for the mention! However Seesmic does not have any “i” would be good if you could correct. Thank you.

Comments are closed.