I Don't Believe the Borrell Numbers
Gordon Borrell, CEO of Borrell Associates, has been nice enough to post a comment to my post Online Job Revenues Grow... but not this much!.
Gordon begins:
Thank you for your skepticism, and for the opportunity to respond to it. Kip Cassino has been traveling, so I thought I'd respond on his behalf.
And goes on to make the case that Borrell's secret methodology lead them to these outstanding numbers.
To put it flatly, I remain unconvinced of the quality of this methodology, and I do not believe the Borrell numbers:
- "It's confidential" -- when you're in the data business, claiming secret data, unverified, unaudited, and unproven by any third party, is a pretty big no-no.
- The numbers don't add up -- "There were hundreds of new boards -- perhaps thousands -- that went from getting $0 just a year ago to $100,000 or more last year." -- well, 1,000 job boards times $100,000 in revenue would be $100 mm incremental, not $2 bn.
- The postulated behavior of small job boards contradicts my actual experience with entrepreneurs. Having run corporate development at HotJobs, and been an LBO investor before that, I know firsthand how extraordinarily difficult it is to get small entrepreneurs to reveal anything confidential about their business. For Borrell to claim that they have successfully negotiated, with thousands(!) of entrepreneurs to get a private, confidential, accurate, verifiable data stream about their 1-year-old businesses does not fit with the typical experience one finds with small companies.
- The posited superior data collection effort does not cover known private players in the business. A quick call around to job board companies reveals that known $1+ mm players are not sharing their data with Borrell. Who makes up the list of these thousands of small job boards?
- No reported tripling in online advertising expenditure from customers. Customers that we know are not reporting that they have suddenly tripled their online advertising budgets. Neither is this estimated expenditure reported by any of the major customer-focused HR consultants, (care to comment CareerXRoads? what are your customers saying?).
- The methodology that we are invited to review, link here, is a top-down estimate based on Borrell's assumptions as to estimated advertising expenditure that companies in a particular market make on a per-employee basis! I simply don't know how taking educated guesses based off of D&B and the NAA numbers can lead to anything remotely accurate. Though if Borrell wants to share their calculations and specific methodology, perhaps we can get greater insight.
So while I greatly appreciate Gordon's civility, I believe more convincing arguments are required of a data outfit in order to assure the public at large that their methodolgies and calculations are sound.
At the present moment, I am unable to reach that conclusion with regards to Borrell Associates' claim that the online recruitment advertising market has almost tripled in the past year.

Posted by marc at February 02, 2006, 06:40 PM
Comments
I am delighted to have the opportunity to answer questions about our methodology for estimating advertising spending, which has been developed over more than a decade and accepted by many major media companies. I am posting this to both erexchange.com and cendella.com/stone, since both sites are discussing our recent report.
The numbers in our model tend to generate skepticism until you understand their ground-level origin. We use this model in our work with our media company clients, most of which are on the front line of ad sales and see things from a very different perspective. Viewing advertising only from the 40,000-foot level often delivers an inaccurate view.
Like any other analysis, ours gets better as we work on it. We have sources available to us now that we did not have in 2004, and the sources we did have are now -- in many cases -- better. Rather than complain about our current estimates, a better critique might be to deny those from several years back. If we look back in history, our "backcast" will undoubtedly be more accurate than the estimates we developed then.
Let me offer several points that more specifically address your concerns with our data.
1. The data for our model are gathered from literally every business unit in the US, as well as from many of the federal government agencies that measure them. JOLTS, as well as several of the other BLS data sets that we use, have become available only recently. IRS data, though available for some time, has improved dramatically in its specificity and its accuracy during the past five years. Most of the analyses available today do not take as broad a perspective as ours.
2. Gauging recruitment spending to the listing volumes posted to the major job boards is bound to lead to error -- much like the tale about the blind sages trying to describe an elephant. In fact, if all of the recruitment
spending that left newspapers had translated -- dollar for dollar – to postings on these job boards, Monster would be several levels of magnitude larger than it is today, or ever will be.
What actually occurred, like most human phenomena, was much more complex. Much of the spending went internal to the recruiting businesses themselves, in the form of recruitment spaces on company web sites. Some went to burgeoning recruitment spaces built by industry associations, their associated publications, and/or college linked alumni sites. Along with all this, literally thousands of "boutique" job boards -- like Media Bistro and Snagajob -- sprang up. And, of course, newspapers created their own recruitment web spaces as well. Though there has been some attrition in job board population during the past five years, hundreds still exist and more are created daily.
3. Recruitment spending is not an advertising expense, even though the media treat it as though it were. In any company's budget, recruitment is a human resources expense. Advertising -- in a publication or on a job board – is only one of several choices an employer may make in finding employees. Others include hiring a headhunter or temp agency, where the great majority of recruitment spending that we track is directed. Job fairs, college recruitment, on-site materials and other "non-media" solutions make up much of the remainder. A large portion of the increase we have reported in recruitment advertising is money that has shifted out of the temporary placement agency category; growth in overall recruitment spending was quite modest.
4. As noted in "The Tipping Point," businesses don't change their minds about how they accomplish goals based entirely on precedent. If a better way to do something presents itself, businesses (or individuals, for that matter) rush to it. The stately histograms that show gradual growth of a trend are mostly real only in textbooks. In real life, dramatic swings, sharp spikes and valleys are more the rule. That being the case, when recruitment spending heated up -- as it did substantially between 2003 and 2005 -- it should not be surprising that the move to online recruitment spending was sharp and dramatic.
5. Media has harvested recruitment spending online very effectively during the past few years, using a tool called "upselling." Basically, this means pushing some or all of a print customer's recruitment ads to the web
for an additional fee. Most of the revenue, however, remains on the print side. Around 2005, in many cases, this model reversed itself. In the "downsell," some of a recruiting company's online ads are also put in print. In this case, the majority of revenue remains on the web. The continuing use of "upselling" and its mirror image have much to do with the increases shown in recruitment spending patterns we have measured.
6. A remarkable change in employment is beginning to occur: the nation's largest generation is beginning to retire. Companies that could rely on senior employees to take care of engineering, administration, management, or technical specialties for decades now have to find ways to replace whole departments. My son's case is a good example. He works as an engineer for Westinghouse -- a company that must soon replace 60 to 80 percent of its retiring engineering staff. He's been asked to cast about for classmates who might be good candidates.
This is not the whole story. Human activities are always far more complicated. But the points I have raised underline the danger of dismissing useful estimates just because they don't match "business as usual" models. The future is seldom like the past. If it were, we'd still be writing letters.
Kip Cassino, vice president, Borrell Associates Inc.
Posted by kcassino on February 7, 2006 01:33 PM

Post a Comment
If you have a TypeKey identity, you can Sign in to use it here.