Borrell: Local Uniques Are Highly Overstated

In a report out today, that is sure to be controversial with publishers, Borrell Associates says that local sites’ unique visitor counts typically overstate the numbers of actual people visiting those sites — often dramatically. It also argues that many visitors are not truly “local” (in market) and so not good prospects for advertisers.

Among others, the report makes the following assertions:

  • About 30% of a local website’s visitors don’t live in the market.
  • On average, about one-fourth of a local site’s page views are delivered to people who probably won’t return for another year, if ever.
  • The average local website’s unique visitor count overstates the number of actual people visiting the site by a factor of 3.6, and overstates the number of local people by a factor of 5.
  • So a site that tells local advertisers it has a half-million monthly unique visitors should probably be forthright and start saying it reaches about 100,000 local people.

To ascertain and parse all of this Borrell used consumer survey data and its own internal calculations. The implication of the analysis is clearly that advertisers large and small are often not getting what they think they’re buying with CPM-based ads on local sites.

In one sense this is a re-visitation and more thorough look at the old “out of market reader” online newspaper problem. And in fact newspaper websites are part of the discussion but yellow pages and other categories of local sites are not explicitly mentioned.

I assume however that more than newspaper sites are among those analyzed in the report (although there’s no list anywhere). Here’s a chart, for example, indicating the discrepancy between UV counts and actual people at 16 local websites reviewed:

Publishers play games and make all kinds of claims about their traffic, and to some degree comScore allows it to happen (not that comScore is infallible). But what do you think of some of the Borrell arguments and data points above?

You can follow any responses to this entry through the comments feed.

14 Responses to “Borrell: Local Uniques Are Highly Overstated”

  1. Tweets that mention Borrell: Local Uniques Are Highly Overstated -- Topsy.com says at

    […] This post was mentioned on Twitter by SMBinfo, shoppetweets. shoppetweets said: Borrell: Local Uniques Are Highly Overstated http://ow.ly/1bmFKn […]

  2. JV says at

    I fail to see how a third party estimate (based on survey data no less) trumps an actual site log. 

  3. Greg Sterling says at

    The question is whether people are inflating their internal data. Though I would find it hard to believe that people are doing that across the industry.

  4. Dmitry Pakhomkin says at

    As we saw it initially around 70% of NYC’s section were actually not from New York area, mostly from West Coast. Later it became much more “local”, having around 90% of visits from NYC population

  5. Mike Stewart says at

    Can you expand on how a site like Local.com might attribute unique clicks to another site, such as SuperPages.com, as it pertains to UVs and ComScore?

  6. earlpearl says at

    Looks both very interesting and potentially controversial.  Specifically I’m not sure I understand what the graph above represents.  Do those represent specific business/service/education sites within the markets as categorized on the left side of the graph?  If so, what do those left hand columns mean  ie Major metro shoulder.  Are they referencing shopping centers and/or the major stores within them??

    As to Uniques versus the same people reviewing information from multiple sites.  Oh boy….I acknowledge that….and its more telling in this day and age:  A unique can view a site from their office, their home, their mobile phone, their tablet;  and with a friend.   I’ve done that, let alone the customers we see in our various entities.

    Again not seeing or knowing  all the specifics my sense of the topmost graph in your article, Greg, is an effort to actually gauge the “realistic” buying intent of visitors to a site.  Intuitively and via experience I’d acknowledge that.  As much as people engage one’s site and/or inquire about a product/service …deep within their core is a range of perspectives….they may be lookers in February and deciders/buyers in May…and between those months they could visit sites endless times.

    But alas…as interesting as the report looks, I’m not likely to shell out $995 for information.  

    But interesting….Definitely!!! 

  7. Greg Sterling says at

    The table above shows their survey respondents, those who said they’re from the same market as the site (which suggests newspapers more than YP), uniques and their estimates of the actual people visiting the site. The final column is what percentage of the uniques their estimate of the humans represents.

  8. Greg Harmon says at

    Well, we wanted reaction & appears we have it! A few quick points–first and foremost — in no way are we saying anyone deliberately inflates their UV numbers–they are big enough to begin with! As EarlPearl pointed out, there are many ways one person can appear as multiple unique visitors.

    Whenever succesive visits to a site are from a different drive or browser–where the original cookie was placed, there is no technical way to recognize the same person on a different box or platform–save registration. I do not think even Omniture would say 1 Unique Visitor = 1 Unique person, but you can check with them! A combination of surveys and logfile data for visits appears to us to be the only viable solution to what is ultimately NOT a technical, but a “human factors” issue.

  9. earlpearl says at

    I shopped for and purchased some electronics recently.  My “shopping” experience included  web research and store visits.   My web research included looking up merchandise from 2 different office pcs from 2 locations, a laptop, and a mobile.  That is 4 uniques/1 person.  Not done deliberately.   If anything the proliferation of devices creates more opportunities to “search/shop” from more IP’s dramatically easier and more probable.

    On the business side I’m fully aware from having dealt with some of 10’s of thousands of customers that we interact with these people while they are home, while they are at work, and more recently while they are mobile and on the fly.  We are aware they are looking at our site(s) from multiple locations.  

    Various commentators on the web analytics side have argued against the importance of uniques as a critical metric for years.  They argued against this long before smart mobile devices became popular.  Smart mobiles dramatically increase the opportunity for a single individual to view a site from multiple devices.

    From our business perspectives….what is the value of a unique versus one who visits a site multiple times?  Frankly we aren’t sure….and that is after years of analysis and from direct polling of our customers and potential customers.  From my perspective the chart that breaks down site visitors by “fly by”, “incidental loyalists”, and “core loyalists” is the one that gets to the key of that issue.  Regardless of how many times a visitor comes to us, or how many devices they use to visit our site, my interest is in turning the “fly bys” into “incidental loyalist” and the “incidental loyalists” into “core loyalists”.  Ignoring the technical language description–I’m trying to respond quickly to the likely buyers…and I’m trying to work hard to take the people with modest interest and turn them into buyers or likely buyers.

    I’m still confused as to what kinds of sites the 16 examples in fig 1.3 represent.  Are they business websites, newspapers, institutions, services….what?   When you define core zip codes, especially as it relates to major metro sites….are you defining multiple zip codes in that metro area, one zip code…or what?  What does “shoulder” represent?

    Finally the site that was dramatically different (the outlier) with regards to zip codes was the regional college.   That is interesting.  If that is a college web newspaper….what do we have here?   Are alumni visiting the site?  Are students visiting the site when they are home on vacations/not in the core zip regions?

    The one thing it says to me….is if I’m trying to advertise in that paper  (if its a paper) and sell to students on campus…..whoa….take a harder look at the info. The simplest information I’m getting from fig. 1.3 above is that most of the site visitors aren’t on that campus.   😀     Oooooh I wouldn’t like that!!!

  10. earlpearl says at

    I was experimenting with the new features on Facebook for business fan pages:  What we were focused on were “friending” other business pages that relate to our business and which in turn have a lot of fans who might also be interested in our services.   That theoretically enables us to make relevant comments and appropriate likes in front of large audiences of their fans who will become aware of us…..which suddenly brought me back to this commentary.

    In that it is highly likely UV’s (unique visitors) is often a highly inflated #  (I certainly believe that)…it struck me that fans on a business page is probably currently a more trustworthy number.  I’m sure if this new Facebook attribute gets expanded and then monetized “fan” totals will be artificially inflated.   For now though, it appears to be a more believable number than uniques.  

  11. Greg Sterling says at

    That’s a great strategy. As mre and more local business moves to a CPA model uniques will matter much less.

  12. Pankaj Mathur says at

    I think number of unique visitor is a compelling selling point for online advertisers. This is much like the wattage for a music system. The general perception is that higher the wattage better the music system so electronics companies came up with a concept of PMPO (peak modulated power output) so a 1,000 watts home theater is actually closer to 350 watts if the wattage is measured without PMPO.

    I am not sure about present times but back in 90s almost all the music boxes were marked with PMPO wattage and this was common across the board. So it is possible to entrench an industry with stats inflation if there are economic or marketing incentives.

  13. Greg Sterling says at

    Pankaj:

    Very creative analogy. Hope you’re well. 

  14. YellowBook.com Loses Unique Visitors To SuperPages.com from October to January « Dallas SEO & PPC Google Advertising Expert Mike Stewart says at

    […] recently announced that LocalSearch Unique Visitor Counts are likely flawed and not a very good metric (KPI). According to a recent post from Greg […]

  15. SuperMedia Sales Manager Sends 1000 Junk E-mails or Unsolicited Bulk E-mails to Dallas Area Business « Dallas SEO & PPC Google Advertising Expert Mike Stewart says at

    […] and now has more competition than it can handle. BTW, Borrell and Associates recently stated that local Unique Visitor numbers are flawed, as much as inflated 4x’s (3.6)…. so when you state that 69% of the 35 […]

  16. Data perspective says at

    Data can be looked at from any perspective. For example, the comment about recording 4 different UVs for one person, although seeming inflated, clearly shows interest and intent with that site. So if the site is so valuable that it warrants multiple sessions wherever that person is says much about the value of the content and likely repeat visits. This is good for advertisers. Secondly, if the comscore #s are the standard (my knowledge is that they are), it still shows a reliable comparison between sites. If a site like ESPNs #s are inflated 4/1 than Google’s are as well, etc. As an advertiser/invester even assuming inflation my concern should be more about the relative rank. Still a valuable/viable thing to consider. Thoughts from the experts?

  17. Greg Sterling says at

    Another word for it might be “engagement.” 🙂

Leave a Reply