Exclusive Interview with Microsoft Live Search’s Justin Osmer
This week, I talk with Justin Osmer from Microsoft, a Senior Product Manager with the Online Services Group, which includes both MSN and Windows Live, including Windows Live Search. Obviously, the last few months have been busy ones at Microsoft, with the roll out of Windows Live and the transition from MSN Search. We talked […]
Obviously, the last few months have been busy ones at Microsoft, with the roll out of Windows Live and the transition from MSN Search. We talked about that, Microsoft’s rebranding strategy for it’s search product and the reasons behind it, as well as where they plan to take Search in the future.
We also touched on some of the findings in the eye tracking study, but because the Microsoft version of Search is notably different than the one the study was done on, we didn’t dwell too long on the specifics. However, it was interesting to see how many of the user issues identified in the study were addressed in the design changes we saw with Windows Live.
Finally, we talked about personalization of search results and the impact on search marketing. This interview took place before Google’s recent announcement, so I didn’t have the opportunity to frame any questions specific to it, but Justin’s comments indicate that Microsoft might be the next to follow in Google’s footsteps.
Gord: What were the major changes in the transition from MSN Search to Windows Live?
Justin: I’d be happy to walk you through the various obvious changes that have occurred as well as some of the smaller things, but to address the relevancy question most immediately as I know that’s the most important thing to you. For us we really do approach it as a full-page relevance so we go far beyond the what the top 10 blue links are and really think about the entire page and how we can make the whole page more relevant to the user. And certainly eye tracking is a great way to test that and that’s certainly a tool we also employed on our end when we’re doing testing and development. We also do a lot of focus group work and bring folks into the lab all the time to look at stuff and get feedback on it, so we’re constantly, actively collecting feedback and making improvements to the service.
I would certainly like to walk through the transition from MSN Search to Live search because that ties in very nicely with what we’ll be talking about. But before we go there I want to talk about the concept of full-page relevancy because I find that quite interesting. I just launched a search on Live Search for Las Vegas hotels. Now in looking at the search results I definitely see a number of top sponsored results. I also see over at the top of the right rail that you’ve pushed the ads down and you have related searches in there. Can you talk a little bit more about that?
That’s a great example. Related searches is one of the areas where we realize that when people are typing in their query there may be some refinement to that query that would get them even more precise answers, even more relevant results that they’re looking for. So offering a list of related searches for people helped make the search experience more relevant to them and help them decide for themselves if maybe they wanted to rephrase their query or know more about related queries. It’s right there for you.
And another thing is very similar to that is spell correction. It seems quite simple but it’s actually one of those things that people can very quickly get frustrated about an abandoned their searches are having a hard time spelling the word correctly or getting the wrong information so by having a related search or recommended search based on spell checkers it’s another thing that we’ve done to really improve that process. So those are two examples of things we’ve done.
Another thing that is related to the full-page relevance is really most obvious if you click on images and when you get image search results back you get a full page of just the images themselves. So the search results page is very clean up by hovering over the image to get a little pop-up to give you all the metadata you would need so you can very quickly figure out if the dimensions to that are too big or too small and if it’s the image you need.
So there’s an easy way for you to very quickly scan the page and get a sense of what you need. We’ve also introduced something called Smart Scroll. It’s a continuous scroll that gets rid of pagination so you can just keep scrolling down through the images. What we learned with image search in particular is that users like to see a lot of images and like to see them very quickly. By introducing this into user interface it becomes all that much more relevant to you, you’re not having to click next page, next page, next page and you can see everything right here.
But also up in the upper right-hand corner we’ve introduced some tools to make it that much more relevant. If you want to see the picture is a little bit larger, there is a slider bar you can slide to the right so the picture’s a little larger. If you want to see a ton of them, just postage stamp size, you slide it all the way to the left. And so you can, very quickly, scroll through these things. Then there’s a pull down menu where you can see all the images, or you can preselect images based on their size so if you knew you wanted to get a new desktop or wallpaper for yourself you can select my desktop size and it will only bring back search results that match the same size as your desktop or you can bring back small images, medium images, large images. It allows you to filter very quickly, very easily and makes the search experience all that much more relevant to you without having to do a whole lot.
I think for Image Search all those things are great from usability point of view. Getting back to the Web search side of things it seems to me that one of the betas I looked at last spring had the scrolling feature you described and I don’t see that in the current interface. Is this something that was dropped for the final version of Live Search?
We are certainly still considering using it. We found in our testing was that it worked really well in images but people weren’t necessarily using it in core search. It’s a little easier to keep track of where you are when you’re looking at images but if you’ve got text or copy on the page it sometimes get a little jarring to keep scrolling and scrolling, then go back where you were. We are still going to consider using it and that’s the whole reason we did the beta to begin with. We were testing a lot of that stuff before we launched in September and then pick the stuff we knew would work out of the box and are continuing to tool with the other things.
Knowing what we were seeing with MSN Search and what we’re seeing now with Live Search I’ll just list a few things that we had noticed and then let you walk through some of the other changes in how MSN evolved into the interface we see now. One of the things we noticed was query bolding in the title which wasn’t present in MSN when we did the study. It was incorporated into MSN later and we found that was a fairly significant from a user’s perspective. Being able to pick up scent on the page wherever the query showed up, being able to see the actual words in the query being bolded.
Also the contrast with the fonts between what was bolded and what wasn’t. The original MSN used a fairly strong font face so the contrast between the bolded and the unbolded wasn’t quite as strong as what you’d see on Google.
The white space between the organic listings and the right rail seems to be better proportioned now. The listings used to run over very close to the right rail so you didn’t have the same idea of white space and open accessibility for the listings to help them be scanned.
The division between top sponsored and the rest of the page seems to be a little more subtle now. There’s not quite such a physical boundary to the eye. And then of course the introduction of the related searches and a little more prominent placement on the vertical search results that fall in between top sponsored and organic.
All those things that you list are great and were absolutely intentional in the redesign when we introduced Live Search. They were things that we found tested very well for folks. The bolding and the font changes that we made were, in particular, two areas that were subtle yet impactful in helping people get to that answer they were looking for more easily. Just introducing the overall redesign, with the overall white space and the other items, fit very well into the overall Windows Live brand experience. We’re trying to be a little bit more stark, a little bit more intentional in what you see and then allowing you to find comfort in that extra space around the content.
I’ve done the same search on Google and on Live Search.
The look of the two is a lot closer right now than it was when we did the study. For example, the font. You’re using different fonts but it looks a lot closer. The one big difference between the two I think is your inclusion of related searches. It seems that as you’ve made the changes you’ve moved a lot closer to a Google type of format. Has Google defined what that search format looks like right now? Is it just the fact that it’s been tested and seen by so many people that this is what we expect from a search results page?
You raise a very good point there. To be quite honest, you can only do so much with a query box and search results, right? There’s only so many different ways that you can skin a cat. We’ve put a lot of effort into our UI design, done a lot of testing, and have tried to introduce some personalization tools and some relevancy tools to help make it that much more different, that much more unique, and obviously that much more relevant to you. But certainly Google did a great job out of the gate and has continued to refine their offering. Thankfully, they’ve innovated and have brought search up to the forefront. Search is now the place where most people start their online experience. They deserve a lot of credit for developing those things in their early days.
I think what we’re seeing now is that, while there are some similarities in the way that it looks overall, if you were to stand back a few feet and squint your eyes it might be tough to tell the difference between the top five engines, but what’s going to be the differentiator going forward is going to be more of this personalization opportunity; the way that you are allowed to interact with the search results and fine-tune them in an intuitive way. Try to make it so as a user, you don’t have to do a lot of work to get the results that you want. That’s where we are headed and I think Image Search is a great example of that. Some of the work that we’ve been doing in mapping and the book and academic search are also examples of that. So we’re focused on what’s next and we feel that we’ve got the basics down as far as relevancy and a good search experience for people. Now we need to fine-tune and hone in on those unique differentiators going forward that are going to get people to adopt of service and use it even more.
OK, to paraphrase that: Right now you’re offering “just as good as” search experience and really the next stage is to look at differentiating yourself from the competition?
Justin: Absolutely. We’ve nailed the fundamentals. We feel pretty confident, at this point, that we’ve done that. Now what we need to do is innovate on top of that, add those differentiators, and make our experience stands out and be unique from the competition. That’s where our focus is now. It just took us two or three years to get to where we are now, which if you look at search history is relatively impressive. Building our own algorithmic engine, building our own ad platform, getting all this stuff up, we’ve made some tweaks to it over time. But now we feel that we’re in a very good place for the side-by-side comparison. Now what we need to do is take it to that next level.
You’re absolutely right. When you consider all the work that has to be done to basically launch a search engine from scratch two to three years is not an unreasonable time frame, but it seems like the public and the people in the industry expected more from Microsoft. We expected it to happen sooner. Was that just an unfair expectation?
We’re pretty proud of how far we’ve come in such a short time. This is a difficult, difficult task. It’s a difficult software challenge to build an algorithmic engine that can return the type of results that you need in order to be in this space these days. Being able to nail those fundamentals: getting relevant results, doing it quickly and all those sort of things. Sure, there are critics out there and there always will be. I think people always want things, especially in this day and age, faster and faster and faster. They expect things to take less time. We have been heads down on it since day one and so we’re pretty proud of how far we’ve come.
We also fully acknowledge that we still have a ways to go and we also acknowledge the fact that while we are in third place right now, it’s certainly not our intention to just be there forever. We want to move up in share, we want to move up in volume and get to second place and eventually, first place. But we know that that’s going to take years, and we’re in it for the long haul, we’re in it for that investment, and I think what you’re going to see from us in this calendar year and the years to come, is more and more of that differentiated offering.
Gord: The gaining of that market share comes from the battle to be fought, right? That differentiation that happens in the future. As you roll out more features to help users interact with the search results that’s where you start chipping away at that market share?
I think also if you start looking at the different verticals or different content areas as well. We’ve got a lot of credit with leading in maps right now with our 3-D product, Birdseye imagery and those other things we’re rolling into our local search product and our search maps. That’s a strong category. Image search, same thing. We’ve got a lot of credit and a lot of traffic driving users to that category, so I think over time, if we continue to focus on those subject areas and really nail those, it will raise all boats and we’ll be able to get to where we want to be in the future.
It strikes me that when you look at things like images or maps it presents a safer test bed. When you look at the bread-and-butter web search, that’s the main monetization channel right now, that’s where the sponsored ads are appearing. I would suspect you have to tread rather carefully making any significant changes on that page just in case you impact the monetization opportunities. Whereas maps and images, because they don’t have that same monetization channel attached to it, is a place where you can be a little more daring with playing around with the interface and see what works, and then eventually roll back into web search. Would that be a fair comment?
Yes, absolutely. You’re exactly right. In core search you need to tread very carefully. Not only for the monetization purpose, but you don’t want to jar users who one day will show up and see something completely different that they don’t recognize. So that’s why we had this transition from MSN Search to Live Search roll out over a few weeks time, when we were introducing the new UI to folks. We’re going to do some unique things in these different categories and then what works will get ported into core search.
Right now under our scope bar you have Web search, news, maps, images, classifieds and then under more is where you find the betas of all the different things that we’re working on. That also is obviously a great test bed for us to test all the different things out and get some feedback from users to see what’s really resonating with folks.
Ironically, is it a advantageous position for Microsoft in some ways, when we talk about core search, to not be the market leader who’s held under the microscope to quite the same extent that Google is? Maybe it gives you a little bit more latitude to play with the user interface without tripping off alarm bells everywhere.
There is certainly some truth to that, as well. Being in third place affords us a number of opportunities. We can do a little bit more with our service and not upset as many people. We can also do some unique marketing things that other folks aren’t doing. So it really is a good opportunity for us to really test some things out.
Which brings me to another question I’ve wondered about. It seems like when we talk about interfaces for search, the really bold stuff is being done by smaller third party players, ones like Quintura and Collarity, Blinkx. As you get closer and closer to the mainstream that has the majority of the market share, the less diversity you see in interfaces and the more they are standardized. How does Microsoft view those third parties? Obviously you must be watching them fairly closely to see what resonates with the public as far as bold innovations with the user interface.
For us it’s great because we think any new innovation in the marketplace is good for consumers and it’s good for the industry as a whole. We welcome a lot of the smaller players and obviously pay attention to what they’re doing and see what resonates with the public and with users in general. I think one of the things we’ve learned too is that a lot of the technology you see on smaller sites are very niche, meaning that a very small percentage of users, the power users, if you will, who are on the web almost 8 hours a day, constantly doing web stuff, find that stuff extremely valuable and useful. But the mainstream consumer who may be logged in for an hour a day or even less, don’t need some of the things the smaller sites are offering. We’re going to balance the cool, new, whizbang features that we, of course, as a technology company think are very cool too, versus what the market demands and what do people really need. What we’ve found is we need to deliver folks the ability to get their answer as quickly and as easily as possible. To do that we need to introduce some of these tools that we’ve been chatting about.
Now take it down a few notches to the next level, or to the future, some of these new technologies could certainly catch on. I think you’re starting to see that with RSS and some of the other things are starting to become a little bit more mainstream. People are starting to understand a little bit more about what they are and what they do. You need to consider the things but if you look at the visitor coming through your service, it’s usually a very small percentage that is adopting those leading, bleeding edge technologies. The early adopters.
One thing that I’ve always noted is that if it takes more than a few second to launch your search and get your results back, you’re not going to get a lot of people using it. And a lot of these new tools require you to apply filters, build queries dynamically. You do a lot of things that just take too long for the average user. If you could build that power transparently in the background, so people don’t have to do the heavy lifting, that’s great, but in a lot of cases their technology isn’t robust enough, or developed enough, to make it transparent in the background. It still requires the user to do a lot of work at the front end.
Exactly. What we see, in most cases, is the less we ask of people, the better. Until we can get to a point where the service knows enough about your intentions to answer some of those questions for you, that’s when some of that stuff will start to catch on.
I guess with that service getting to know more about you, Microsoft in some ways now offers the most advanced opportunities to target through the network, through demographic targeting, which obviously ties into knowing more about the user. As you get to know more and more about the user both an aggregate and an individual basis, that gives Microsoft the opportunity to start tailoring results as well.
Absolutely, results become more relevant, advertising becomes more relevant, and it’s one of those things that we obviously want to be very cautious with. You know, the whole privacy and confidentiality aspect of that is something that we take very seriously. But things that in aggregate we can start to learn about users will help us, and then, down the line it certainly could be one of those scenarios where if you log in and agree to have certain demographic information available to us, then you get that more personalized experience versus someone else who’s just showing up to the site and may not want to include the fact that they’re a male in their thirties, looking to buy a car, for example. I think that it’s going to be really interesting, and frankly, kind of exciting, to see where this can go in the next months and years.
This idea of logging in and personalization of results, obviously all the three major players have different flavors of that. How long, from the Microsoft perspective, from now does that become a significant factor for the search marketer out there to understand? How far away are we from the day when you just can’t assume you rank number one or number two or number three, and I’m talking more on the organic side, for everyone? You may rank number one for one person and you may rank number 55 for someone else.
Now, that’s a great question. That would be a great panel discussion at one of these conferences actually. I would peg that out probably at least a year but it’s certainly something that people need to start thinking about.
Right now, there’s a bit of arms race with marketers trying to get themselves ranked higher but at the end of the day, if you are marketing, we’ll take that car example again,an auto sales site, then hopefully the demographic information would match up and you would still be ranking high if you’re doing the right things on your end to do that. I think it’s more of a subtle shift in your mind set about how you want to get at your target audience and it’s not as much about a shot gun approach as it may be today. It’s a bit more fine tuned.
But it’s a great point that you raise, because it could very easily get to the point where you’re number one or two on your (result page) and on mine, you’re not even on the first page of results, so it will be very interesting to see.
Why the move in branding from MSN Search to Live Search? What was the thinking behind that?
The thinking behind Windows Live in general is that it’s a suite of services that are available across the Microsoft platform of products. So MSN has the ability to give you your email, get to search and so on, but those things are Windows Live branded experiences and services. MSN still has search today, but it’s powered by Windows Live Search. Same as if you go to Microsoft.com today, it’s powered by Windows Live Search. So, it’s somewhat intentional in setting up in Windows Live a collection of things to allow people to get all the information they care about in one spot. With one log in you get your contact list, a melding of your IM contacts and your mail contacts, you’ve got your Windows Live personalized page that you can set up to get your own RSS feeds, you’ve got your search results there, you can build in your search macros, you’ve got all sorts of things that you can do to make it even that much more uniquely yours.
MSN continues to be a very strong brand and a place where people love to go get content, read content, review content and get immersed in that content. Windows Live is more about you cherry picking what’s important to you, pulling that down and having that in one spot. And then the services that underlie Windows Live will then power the experiences on these other properties.
So it’s really just separating search as a function from the identity that MSN had and, in some ways, recognizing search as a fundamental core activity that deserves equal billing with some of the other core activities that are wrapped into the Live offering?
That’s a great way to look at it. It’s a core product and a core service for us, and you’re exactly right. We felt that it needed to be brought up in the hierarchy and acknowledged that it is something that could be plugged into all these different things across the network.
In user studies, we constantly see that although people may be choosing sponsored, they always want at least the option of looking at the organic listings. Talk a little bit about your approach to the balance of organic versus sponsored on the page. How important they each are, relative to the user?
It certainly is specific to the query, as I’m sure you know. Many times, the sponsored links are even more relevant than the organic results. In another example, looking for a digital camera, you may just want to go buy it right now, so the results in the sponsored links may be exactly what you were looking for, exactly what you need and you’re just a few clicks away from buying that camera. There’s always a balance there. We don’t want to page to look like a subway stop full of ads and you can’t really find your way through it and figure out what you want or what you need. But we also want to balance that with the need to monetize it and have it have some differentiation, so that’s what you’re seeing now with the results at the top of the page. By highlighting, as you mentioned earlier, with the boldfacing of the search terms, calling that out in the sponsored links as well as the organic links to again help you see where your query is falling. We will always have a place for sponsored links on the page, obviously, along the top and then along that right rail. The number of sponsored links along the top will depend on the people who have bought that query.
How is it determined when those sponsored links show up and when they don’t? Is there a click through threshold where if users don’t click on it for certain queries it automatically gets turned off or is it monitored on a keyword by keyword basis?
It’s actually a combination of a whole number of things that the adCenter guys have come up with. The most common is certainly the keywords, and so that will dictate how many are on the top, if any, and how many are along the right rail. And certainly click throughs have an impact on how long an ad may stay on the page, where it will rank in the results. But there’s a whole number of factors that go into play mathematically behind the scenes with the adCenter team.
You actually brought up the example of digital cameras, which seems to have become the default query any time people test for consumer intent. And one of the points that I’ve always noticed in that is if you search for digital cameras, yes, there are a percentage that want to buy right now, but the majority of people using that query will actually be doing research.
If I search on Live Search right now for digital camera I get three sponsored ads that are all trying to sell me a camera right now by offering me free shipping, by offering a 70% year end discount, by offering great deals. I don’t see in those top ads anything to help me decide which the right digital camera is for me. When do we get to the point where a search engine is able to disambiguate a research versus a purchase query and serve up top sponsored ads that are even more relevant because they’re aligned to my intent?
That exact question gets to the point that I raised earlier when I said that the more we know about your intent going in, whether it’s based on search history or other factors, such as your demographics that we know, that will help dictate that. One of the things that we can start to do too, and you noted that earlier, is pull in results from the other categories, so if you do a query for digital camera or digital cameras, you’ll get most popular products for digital cameras right below the sponsored link, and that is pulling from our product search vertical. And so you get Canon cameras with a star rating and the price right there. You automatically can see right there, “well gosh, Canon must be a good bet and the PowerShot seems to be the most popular” so you can click through and see those results. You can also click through on the most popular products and it will take you through into the products vertical. So, there are some things that we can do now to fine tune that and make that experience more helpful to you. But I think a lot of it will depend on where the future takes us with the technology as far as knowing more about user intent.
Thank you for your time Justin.
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.