John Woods is the CEO of Synature, a UK firm developing ‘attitudinal matching’ solutions for etailers and portals.

Like a cleverer version of Amazon-style book suggestions, its technology offers a social search tool for internet shoppers to find products that similarly-minded people recommend. Companies can also use it to segment their customer bases and target users with personalised content and advertising.

We spoke to John about a new partnership Synature has formed with MyTravel to provide holiday ideas to its customers, and to ask him a bit more about the technology.


In a nutshell, what’s attitudinal matching all about?

Basically, it’s using like-mindedness to search for things online. We came up with the idea back in 2001, but what sounded quite futuristic back then seems much more down to earth these days.

What’s changed in the meantime is the social internet boom. Sites like and Digg are already using other people’s attitudes and opinions to allow users to find things, and what we are doing is to make that quantitative – to operate in the same way as behavioural targeting and other similar techniques.

The key to that is being able to measure like-mindedness and reduce that to a set of numbers that can be used for marketing or some other purpose.


Who has adopted the technique so far and what have the results been like?

We’ve done a project with a company called LunarStorm, which is a social networking site. We asked how people feel about day-to-day subjects such as politics, TV shows and celebrities.

We’ve recently gone live with MyTravel, which is using our qubox system, branded as a ‘Holiday Matchmaker’, to help people find holiday ideas. It’s great for us because it’s a mainstream consumer internet application, and a holiday is a product that has quite a high emotional element to it. You buy a holiday as an experience to make you feel happy, so it’s a natural fit for attitudinal matching.

To make this into a social search tool, we used quite a large sample from the MyTravel customer base, put them through the attitudinal profiling tests and asked them some questions about the holidays they would recommend to like-minded people. So what you end up with as a customer are some surprisingly useful ideas about holidays. 


As the test you ask users to take lasts a few minutes, how difficult is it to persuade people to do so?

It’s actually quite easy. We haven’t given any financial incentive or prize, but said that we were building this thing and asked for people’s opinions. We did say we would send participants some holiday ideas – that was their only incentive.

But it turned out to be a sufficiently appealing proposition for us to get some good numbers. One of the things about social search in general is that people are quite altruistic and want to work together to help each other. Also, because we are not offering a financial incentive to people, we can be very confident about the quality of the results.


How are you measuring the accuracy of the recommendations? Have you been soliciting feedback from users or doing research?

I can’t be very scientific about it at the moment, because we haven’t yet had an engagement that’s susceptible to that. Really, we should do a drug trial-type thing where you give some people real recommendations and others random recommendations, but we haven’t yet done that.

What we do have is qualitative feedback from users of LunarStorm, and they say that it works. From our internal research, we also know that when it is used for targeted marketing purposes, it can increase the success of campaigns.


What are the costs and difficulties that sites need to consider before adopting attitudinal matching?

There is a certain level of cost involved up-front, and that is because the way we build the qubox is research-driven. We don’t come into a project with any pre-conceptions, so we have to do a bit of research. But it’s not massively expensive – we do days of work in that process, rather than months.

In terms of the software itself, it’s an ASP model and it’s designed to be easy to implement. Technically, there isn’t any complex integration.


How is this technique any more effective than the product recommendations you get on


, based on what other customers are reading or buying?

If you take the recommendations Amazon produces, they are oriented around the product dimension.

They work by recommending that you buy Book A because you have bought Book B and someone else has purchased Book A and Book B. It’s a very clever technique and Amazon makes a lot of money out of it, but you might read Book A and Book B for very different reasons than somebody else.

Our technique would ask how you feel about books, use that to identify other people that feel the same way about books that you do, and recommend books to you that those like-minded people have recommended. You get into the underlying reasons why people make choices.

We don’t think it will replace those collaborative filtering techniques that Amazon and co. are using – it’s complementary. But it works in areas those techniques struggle in.


Are there any areas or sectors in which you feel it will work better than others?

It’s early days for us in terms of experience, so we can’t say definitively. But applications where the primary objective is to match up with other people, such as social networking and social search, are a very obvious area for us.

In terms of e-commerce, products with a high emotional content are good ones. Financial services don’t sound very emotional but why users choose one mortgage provider or bank is often about how they react to their brand. Automotive is another area – you don’t just care what horsepower your car has, you care about the brand as well.


What other applications do you see the technology being used for?

We’re still discovering some of the applications. I used to work in the web analytics industry, which is a relatively mature industry, and we have the opposite challenge here of it being totally new. The ability to target users based on like-mindedness has never existed before, so we have to discover new applications for it.

The product suggestion market already exists and people are using other tools like collaborative filtering, so we can go after that. We can make recommendations for products that existing technologies can’t work on at all – such as what would be your dream holiday; things that are hypothetical.

We think that social networking is another big area, and the real commercial pay-off may come from the ability to do better targeting of advertising. So a social networking running the technology to match people would get the benefits of increased usage and stickiness, but would also be able to do targeted advertising.

We’re also interested in strategic relationships with portals – they could make more money if they had better profiling of their users.

So in general, it’s a personalisation tool. We’re not expecting the data itself to be a sufficient reason for using it – although the data is quite interesting. The real commercial pay-off is when you use it for personalisation.


Are you talking to any behavioural targeting providers about allowing sites to easily combine both sets of data? Or do you see behavioural targeting as competitive?

We are talking to one or two people that are using behavioural targeting and looking forward to working with a customer which has both techniques in operation, as they are extremely complementary.

With behavioural targeting, you don’t know why someone is doing something – you just know what they have been doing. With our attitudinal approach, you know what’s behind their behaviour. So if you link the two, you would get a great opportunity to target them with the right content or advertising.

We’ve also set the system up to be easily integrated with other targeting techniques. But at the moment, no-one is doing both things together.


Are you thinking about providing trusted reviews from like-minded people alongside recommendations?

We’re not actually doing this yet, but we are interested in it. And that goes for other forms of user generated content as well – it becomes another way to search for forum postings and blogs.


In the future, could there be ways to gain attitudinal data on customers automatically, without them having to volunteer that information?

Yes, combining with behavioural targeting, for example, is one way that can happen.

If you are operating behavioural targeting and attitudinal matching in combination, there will be a group of people that you have attitudinal information on. You could see how that relates to their behavioural information and potentially use it to refine your behavioural targeting strategy, even for the people you don’t have the attitudinal information on.

But we’re focused in a different way. We would prefer to make the case for an end-user to volunteer the information sufficiently compelling.


And that would also help you avoid Big Brother-type concerns?

Yes. All these techniques like behavioural targeting that just observe or infer; they can be totally automated, but a lot of the time observing the way people behave on the internet is not a good way of understanding why they are acting in that way. There is no better way of finding out why people are doing what they are doing than asking them.

We’re in a slightly different mindset – we want to make this as compelling and enjoyable to participate in as possible.


Finally, where are you at as a company at the moment?

The last twelve months or so have really been about validation for us. We’ve done a few external implementations, and we’re just at the point where the technology has been used in anger. We’ve demonstrated it can be used easily, and that there are no technical barriers to a full commercial launch.

We are funding by a company called Angle Technology, which is an early stage investor. The investment from it will last for an indefinite period at the current level of activity, but we want to scale up.

A lot of the activities happening the social internet space are US-driven, and we would ideally like to have a serious investor come on board and help us pursue the global opportunity. We are in discussions with a few people.


Also see our

Online Retail User Experience Benchmarks

 reports for more information on how to use personalisation tools to build loyalty and drive sales, and our

2007 Ad Networks Buyers Guide

for the latest developments in behavioural targeting.