Interview with Thomas Roberts, co-founder of mobile search The Texperts

thomas roberts

The keen-eyed amongst you may have noticed a certain penchant here at Distilled for blogging on the subject of mobile search. In this interview, we were lucky enough to speak to Thomas Roberts, co-founder of The Texperts, an award winning text-based Internet answering service that leads the field in human powered mobile search.

For those of you who think you've seen the name 'Texperts' somewhere on this blog before, you'd be right: it was one of the options listed in Will's recent post on the possible future of mobile search.

We caught up with Thomas earlier today to have a chat about mobile search, geo-location and his mum.

So Thomas, tell us a bit about the basics...

> [Thomas]: Well, we've been going for about four years now and the service, in its simplest form, answers questions asked via text message. However, it gets a lot more complicated once you look at the actual delivery of this. There are a lot of systems in place to make the process more efficient and cost-effective; in service terms, we have an expertise in finding answers.

> There are about 250 Texperts. The majority of these work from home in the UK, but we have several employees in South Africa, Canada and the US. Demand for these jobs is high. The Tex-factor Challenge is part of our recruitment process and only 2% of those who take the test pass and can then be considered for work.

Thomas tells us that the Texperts make themselves experts by using the best possible resources out there. For example, they have negotiated a relationship with Guinness World Records which gives them full access to the whole database. Normal users only get a fraction of this and the books themselves only contain about 5% of the total records. Obviously, there's some pretty random stuff on there, but then they get asked some random questions!

This kind of categorised searching is apparently really important; Thomas estimates that only about 10% of texperts' searches go through google. Each question is slotted into one of about 100 categories and each of these is kitted out with a range of vertical search engines, which are then presented to the Texpert as the best place to start finding an answer.

Do you ever use, for example, wikipedia as a source?

> Our current thinking on wikipedia is that it's a great place to start searching for the right place to search, if you see what I mean.... Obviously, because it's user generated, we don't use it as a reliable source. If a situation arises where the only information we can find is unreliable, then we'll say so in the answer.

This is a rule used across the board with unreliable sources. If a question asks, for example, 'Is it true that Paris Hilton has a new boyfriend?', then the Texperts might reply with something like 'We can't be sure, but according to a few unverified sources online, Paris was getting cosy with a certain Benji Madden'. Thomas emphasises how transparent the whole search process is. Users can log-in online at any point and see exactly where their previous answers came from.

Ok, so you can compete on an information level, but how do you shape up against the search engines?

> Our service does not compete (or attempt to compete) with the experience of searching at your desktop computer. Primarily, it's a service for people who are busy doing something else at the same time as needing the information, like trying to get somewhere or socialising with friends in the pub. Speed-wise, if you were sitting in front of google and asked for a definition or something then, yes, you're going to beat the Texperts. But, in my experience, when I'm out and about, I'll nearly always get a quicker service from the Texperts than I would from a mobile search.

> Similarly, phones that have a bearable level of multimedia browsing suit another purpose. If you're rushing to the airport and need to find the right terminal, you're probably not going to have the time to search on your phone, however fancy it is. You'd use your iphone if you were sitting in a waiting room with time to spare.

Thomas tells us about Google's SMS service (only available in the US), which can answer simple queries like 'pizza in new york'. This sort of question would probably deliver some addresses and phone numbers, much like the first results in a desktop search might. However, the difference between this kind of response and texpert answers is that the Google SMS doesn't offer any kind of qualitative element. If the Texperts are asked about pizzas in New York, they will provide information on the best pizza that they can find- the best value, best quality, best location.

And I suppose, thrown into this, is the problem (or rather the advantage for you guys) that a lot of people just don't know how to search?

> Yeah, absolutely. My mum is a classic example of this. She's perfectly capable of using the Internet and does so regularly. But if she wants to find a good hotel in Paris then she will simply type 'Paris hotel' into google and, as a result, have nothing very reliable to go on. In contrast, asking the Texperts this kind of question will deliver a response in about five minutes and would give a really useful answer.

We got a bit side-tracked here.... Generally speaking, people are bad at searching aren't they? We decided that, for a lot of people with a query, even if they're sitting in front of a ready-to-go search page on the computer, they would be a lot better simply using a service like the Texperts rather than trying to find the information themselves. They are the experts, after all.

To go back to mobile search, how do you intend to protect your company and your revenue stream as your users become more and more accustomed to having free mobile internet at their fingertips?

> Interestingly, a lot of our best users are the 'early adopters'- ie. the sort of people who would buy an iphone and use mobile browsing. We figure this happens because they're the kind of crowd who are willing to pay for something that works well. In our eyes (and theirs presumably), the cost for the relevancy and timeliness of the information is justified.

> We actually see phones like the iphone as an opportunity for us to give an even richer answer. For example, it might be possible to send an image of that hotel in Paris along with the reply.

Have you considered the possibility of running a free ad-supported model to compete with free wireless internet mindset?

> The possibilities presented by ad revenue are something we're keen to understand, and we intend to keep a close eye on its progress elsewhere in the field. Having said that, objectivity is really important to us. As soon as the user feels they're being sold something in their answer then trust is lost.

We take this conversation a step further and discuss how, despite this trust issue, an ad that's really relevant and comes in response to an action by the user (rather than out of the blue), can be well received.

In Thomas' words: > Pull rather than push marketing is far more important in mobile search. People don't want to be bothered with needless information when they haven't asked for it but are far more likely to be ok with it if they've already asked for relevant information.

For example, if someone was to text in asking for the nearest Borders, the Texperts might text back saying 'your closest Borders is here but Waterstones is closer and if you take this message in you'll get 10% off'. We agree that that sort of thing might conceivably have a future as it offers an opportunity for some highly targeted advertising.

In the same vein, would you ever consider paid listings? In other words, could a hotel pay to be recommended by you?

> We don't accept paid listings. If a hotel wants to be recommended by us then it has to be the best possible solution to our user's query. If it has good reviews and is good value for money etc, then it might make it on to one of the vertical search engines that we regularly use.

In your opinion, would it be worth these hotels aiming for a spot on those search engines. Can you give us any idea of search volume?

> Obviously, I can't give you exact numbers, but we're in the hundreds of thousands per month now.

So yes, it probably is worth making sure you've got a few positive and prominent reviews out there. We go on to discuss how, for every one of these questions that concerns a location, an optional link to a map is sent with the answer. It's surprising how popular this service is: according to Thomas, it results in an uptake 10 times the industry standard for paid extras. He reasons this is partly because the link takes you straight to the right map, rather than anything more complicated.

One of the other things we're really interested in (as well as mobile search) is geo-location and the issues surrounding it.

> Yeah, it interests us too. When we were starting up the business, we did actually think about buying the technology (that's already available) that would allow us to know a mobile's location, but it was quite an expensive process and there were of course the privacy issues surrounding it...

So do you not get a lot of questions asking for information about the user's location, with them sort of assuming that you'll know where they are?

> Actually, no, not really. We had the same concerns when we were starting out but it's really not an issue. It's interesting how the majority of questions on that theme ask about a location that's going to be visited in the future. It's much more rare to be asked stuff about where the user is right now.

Thomas explains that the person texting in is usually very understanding when asked to confirm his location. This means that geo-location is not a high priority for the Texperts as, when necessary, it can simply be asked for. This surprises us at Distilled; we were expecting the same sort of issues that the search engines are dealing with at the moment to come up.

Let's talk a bit about the future. Where do you see your biggest threats coming from? Algorithms or competing companies, or something else we haven't thought of?!

> There's no way algorithms can compete with us. Rather, we see them helping us to push the cost of the service down and make it more efficient. We already use a few clever algorithms to answer some of the most straightforward questions.

> In term of other threats, it's a tough call. The company is still relatively small and this area of business is very much starting to have its day. This means it's a really exciting space but that also, potentially, there's a lot of money that could compete with us.

Just to wrap up... Obviously, we're really interested in your 'online' presence. How do you integrate your 'offline' mobile searching and your online brand?

> As you may know, we had a big re-brand over the summer so we did, and still do, quite a lot of advertising offline. But our website is still important. Our users can sign up to the website, buy credits and, as mentioned, check their question and answer history. We've got a few plans on how we want to develop this. One of them is to provide the opportunity for users to feedback on the answers they receive. They can already do this via text, but people are far less likely to do this with their mobile than they are while at a computer.

Thomas went on to explain some further perks of the website.

> Our blogs are really popular. Because we have such a huge volume of information passing through our systems, both from the users and their niches, and from the various vertical searches our Texperts use, we really know what the zeitgeist is, we know what people are interested in right now. ______________________________________________________________

Just to end by saying thanks to Thomas. We were on the phone for nearly an hour and I hope you'll find the topics covered as interesting as we did!

Get blog posts via email