• Register
• Log in

SEO Theory - SEO Theory and Analysis Blog

Algorithm analysis, Web community relationship analysis, SEO practices and techniques, industry news, etc.

• Home
• About SEO Theory
• SEO Glossary

Exploiting the holes in search market share data

Posted by Michael Martinez on April 22, 2008 in Competitive Analysis

I have been monitoring and measuring search market share since public data has been released by the various services. Danny Sullivan has been doing it for many years more than me. However, as I have noted in past discussions, today’s metrics don’t accurately report real search market share. Search market share metrics are important for optimization for a variety of reasons, but most importantly because they need to drive your optimization priorities. So it doesn’t help that we’re being given inaccurate data to work with.

Danny Sullivan’s latest trend recap underscores some of the problems with search market share metrics but he doesn’t address one of the most significant deficiencies in the numbers. He doesn’t take into consideration how many people actually use search engines, or for what they use the search engines, or how long they use the search engines. He doesn’t really have that much data to work with, so he is careful to warn people not to read too much into the search market metrics.

As optimizers we have three sources of information on search market share:

1. Primary independent reporting services like Compete, comScore, Hitwise, and Nielsen
2. Alternative independent reporting services like Alexa, Quantcast, and others (and there is some overlap between their reporting and the first four)
3. Our own analytics data
4. What other people disclose of their data

Wouldn’t it be nice if you could independently review a multitude of Web sites’ search referral data in aggregate form? That kind of data has been made available for years but unfortunately it’s never robust enough to be trustworthy. Still, it’s better than nothing (sometimes). We’ll probably never have perfect knowledge of search engine referrals unless the industry falls under regulation (which doesn’t have to be as bad as it sounds because regulation doesn’t have to cover every aspect of performance).

Setting government requirements for reporting search engine statistics could affect many things well beyond what choices we make as optimizers. Given how much money has been invested in search engine stocks, for example, a more accurate measurement of real search market share would help investors determine which search services have the most revenue growth potential (so they can plug earnings into spreadsheets and make projections based on trends in use and revenue).

If search engines were to semantically categorize their query traffic in terms of interests (sports, entertainment, technology, news, etc.) they could help us codify their demographics better. Right now about the best we can conclude is that Ask feels it appeals to housewives and academics, Google feels it appeals to technical people, Yahoo! probably appeals to casual business people, and Microsoft probably appeals mostly to people interested in entertainment.

If you knew for certain that search demographics fell out in that pattern, you could plan your Web optimization accordingly. Hence, entertainment sites would want to target Microsoft and Yahoo! first, Google and Ask second. News Web sites would want to target Yahoo! and Google first, Microsoft and Ask second. Financial sites should emphasize Yahoo! and Google first. Academic sites should chase Ask and Google first.

People look at Google’s 6-7 billion queries per month and think, “Well, the business is all with Google.” Or maybe they look at Google’s earnings and conclude that all comes from PPC in search listings. Ignoring other factors leads to bad prioritization. According to most estimates, fewer people actually use Google than use either Yahoo! or Microsoft’s properties. If you want to reach the largest possible Internet audience, then, you need to advertise on all three networks. If you drive all your money to Google bcause that’s where you get the most click-throughs, what are you sacrificing in terms of brand recognition?

The old Brand-versus-Performance argument (familiar to many people who came up through the Web banner advertising years) remains a potent stimulus in pay-per-click advertising decision-making. You can get a LOT of free exposure from non-performing PPC ads as long as they are shown beside relevant content. Then again, how long will a network tolerate non-performing advertising? It costs each network some fraction of a cent to serve each advertisement. Those costs add up and have to be recovered from network fee structures.

Studies have shown that when people see your advertising everywhere they assign credibility to your business. They feel like you’re making an effort to offer something to the marketplace. Pay-per-click advertisers can leverage this market psychology into their brand-building strategies, but to do so effectively they need to know a few things such as:

1. How many eyeballs each search engine receives
2. What those eyeballs are looking for
3. How long those eyeballs hang around
4. How often those eyeballs come back

You and I, of course, are more than just eyeballs. But we’re not what search engines and other marketing agencies think we are. We surf from multiple computers, use multiple browsers, clean out our cookies, login and log out of services, and occasionally use anonymous surfing tools. We may visit the same web site five different times in a day for five different reasons and look like the same person; we may be treated as though we only visited the site once or maybe five times or ten times; and we may be seen as five different people.

The inability of Web marketers to accurately measure how many people visit a Web site, how often, and why — as well as who these people are and what their interests are — limits the value of Web metrics. The endless debate over whether we should be measuring page views or visitors, unique visitors or visits, minutes or actions, only confuses our reporting methodologies further. It would take a pretty sophisticated metrics service to let you interactively change the models so that you can see the data from multiple perspectives.

To optimize for search you need to reach several conclusions:

1. Which search engine has the most relevant searchers for your content (I said searchers, not searches)
2. Which search engine has the most relevant content (similar to your own)
3. Which search engine is the easiest to break into for your query space
4. Which search engine is the most heavily optimized
5. Which search engine has the largest number of query spaces relevant to your content

Getting into a search index is not the same as ranking for a particular query space. Getting in just means you’re listed. You may rank for something but you don’t have to rank for anything. Still, being included in a search index at least gives you a place to start. If you’re banned from Google and married to your domain and your Web design choices then you need to figure out how to leverage the other search engines for traffic.

If you’re just starting up a site and you assume you need to rank in Google, you’re doing it wrong. Just because people conduct more searches on Google doesn’t mean more people use Google. Nor does the fact that Google gets over 100 million visitors a month mean that most of the people searching for your content are using Google. Industry estimates suggest that Ask, Google, Live, MSN, and Yahoo! combined draw about 400 million unique monthly visitors (I am rounding down for convenience’ sake).

“Unique” just means that the search engines cannot eliminate duplicalities any further than that.

Furthermore, roughly 60% of all search users are thought to be “loyal” — they almost always use only one search engine. That means about 240 million “unique” monthly visitors don’t change search engines. It also means that the other 160 million “unique” monthly visitors are probably fewer in number. How many duplicates can we assume are bound up in those 160 million people? Let’s say for convenience’ sake (and absolutely no other reason) that they could be narrowed down to about 60 million true unique visitors, giving us a total of 300 million major search engine visitors.

If you’re only marketing to Google’s organic search results, you’re only reaching about 33% of your potential marketplace. And one has to ask why you would want to arbitrarily eliminate up to two-thirds of your potential customers from your marketing plan? Are you operating at full capacity only with Google? You cannot take on any more business? Someone is bringing in traffic from Yahoo! and Microsoft. If that’s not you, it’s someone else.

Your search optimization plan should be looking at a 300 million-person market audience, not Google’s market audience. If you can estimate demographics for your whole market then you better decide where you’ll get the most bang for your buck. But even if Google proves to have the largest number of paying referrees, what if you’re blocked out of the search results? Do you have a plan for search engines where your competitors don’t yet have solid positions?

The fact the average visitor performs about 50 queries a month on Google could mean a lot of things, but does it also mean that person makes 50 purchases for every 30 purchases someone else makes on Yahoo!? The number of queries per month blinds marketers to the real structure of customer profiles. In fact, the reported earnings drawn from pay-per-click advertising also distort marketing pictures. A lot of people are clicking on Google’s advertisements. Why is that? Are they all making purchases or are Google’s paid listings just being used as a search resource?

Do 100% of all people who click on a pay-per-click advertisement actually buy something? Of course not. Conversion rate data varies according to source but it’s a fairly rare industry that achieves better than an industry-wide 10% conversion rate. So that suggests 90% of Google’s advertising revenues are based on non-converting sales of advertising clicks. And what does that say about the quality of the traffic from a search service where people perform more searches per visit than other search services.

Search fatigue, where users simply get tired of looking for something and give up, is one of several probable reasons for why Google users search on Google more than on other services. They may not become fatigued as quickly as people on other search services, or people on other services may find what they are looking for sooner (on average).

Another possible reason Google searchers search more often could be that they get distracted too easily. The endless pursuit of rleevant content becomes a sort of random string of queries that eventually lead back around to the original query. People may perceive that Google is friendly to experimental searching (because many loyal Google users are probably more technically oriented than loyal users of other search services).

It could also be that SEO rank-checking has become so mainstream that it actually skews estimates for Google’s number of queries performed. Many people still check rankings by hand on a daily basis. Ranking queries have no converting purpose. People just want to know how many queries their content ranks highly for. Maybe Google is scraped by more robots than other search services, in which case maybe Google’s 100 million visitors per month is not nearly as accurate we’d like to think. Remember, it’s the marketers buying 90% non-performing clicks who drive Google’s profits, not the people performing the searches.

The peril of self-perpetuating perception is very real in search engine optimization. We are not only wowed by the earnings and search market share numbers, we tend to glamorize our own search referral data. If we focus mostly on Google optimization (meaning we select our keywords based on Google research) then we’re stacking the deck in Google’s favor when we compare search referrals from the various search engines.

Some people complain they receive all their traffic from Google despite good rankings in Yahoo! and Microsoft, and yet they have competitors who receive a lot of traffic from Yahoo! and Microsoft. How is that possible? How likely is it that people using Yahoo! and Microsoft will click on fewer listings before making a purchase decision? How likely is it that the optimizer who gets all his traffic from Google didn’t identify the relevant queries on Yahoo! and Microsoft that offer comparable traffic?

Different audiences use different search idioms. They use different queries to find similar content. A Yahoo! user might be more interested in “Britney Spears pictures” and a Microsoft user might be more interested in “photos of Britney Spears”. If you rank well for “Britney Spears nude” on Google but didn’t optimize for the other two queries, should you expect to see similar traffic from all three search services?

Run a query on Google for “SEO services” and then run the same query on Live. Do you see as many results on Live as on Google? Do you see as many PPC ads on Live as on Google? I don’t, and that tells me people are devoting less effort to getting their marketing message out on Microsoft than on Google. On the other hand, if you search for “web site traffic” on both services you’ll see a more even mixture. Perhaps Microsoft’s visitors think more in terms of “web site traffic” and less in terms of “SEO services”.

The SEO community obsesses over Google and virtually ignores Microsoft, so any Web marketers who cultivate Web marketing query spaces on Microsoft may be miles ahead of the best Google optimizers. That is, if Microsoft’s loyal users are less aware of search engine optimization, they may still be a fertile audience for the basic marketing message. Do they have their own Web sites? Are they trying to build traffic through Microsoft? Their world doesn’t encompass Google, Yahoo!, or Ask. They don’t know about or care about SEO Theory or SearchEngineLand.

Greed is an SEO’s worst enemy. As soon as people smell a potential for profit in search they throw all their resources into exploiting whatever little advantage they think they have. In doing so they don’t look at the larger picture.

If you could seriously trust the Web metrics we’re seeing about search market share today you’d hear this huge sucking sound: everyone would abandon Microsoft, Ask, and Yahoo! and hop on the Google bandwagon. But apparently the actual people — the users — don’t pay nearly as much attention to search market share estimates as do Web marketers. Given just how fundamentally inadequate our search market metrics truly are, is it then any wonder that people develop so few strategies that look beyond Google?

While your competitors are obsessing over hyperoptimized query spaces on Google, have you done everything possible to prospect for underoptimized query spaces on other search services? You may find there is more converting traffic out there than you can possibly imagine, even if that means paying some money to Yahoo! and Microsoft rather than to Google.

The bottom line for any advertiser is always the same: you maximize your return on investment, and the ROI for Google PPC may often appear to be better than Yahoo! PPC ROI or Microsoft PPC ROI. However, today’s ROI is driven at least in part by yesterday’s market building. There was a time when no one used Google to find what you’re selling. That lack of proven 10% conversion didn’t stop people from building the market on Google.

If the market hasn’t been fully developed on Yahoo! and Microsoft, you have opportunities your competitors haven’t dreamed of. Just as it’s a bad mistake to do your Google backlink research on Yahoo!, it’s a horrible idea to do your Microsoft and Yahoo! query research on Google.