
Selection bias can mean the difference between software spotting an enemy tank… or just a sunny day
One of my all-time favourite anecdotes about selection bias (and, actually, in general) comes from the US military. It’s probably an urban legend (I hope it’s true) but back in the 80s the military was looking to increase the accuracy and speed of tank detection using computer algorithms. They took 100 photos of tanks hiding behind trees then cleared the tanks and took 100 photos of just the trees.
All well and good. Except that the images they selected all had one other thing in common. The tank pictures selected were all taken in sunny conditions. And the non-tank pictures weren’t. In the end the US military had spent millions to teach a computer how to spot a sunny day. The moral being… be careful of selection bias.
Now, I recently got asked to undertake a project that I still can’t see how to do well without a) spending lots of money that the client doesn’t have… or b) falling foul of this selection bias. This problem involves Business Wire (other wire services are available) and if using the wire brings in more traffic to the website.

It pays to examine the value of wire services closely… and to understand the potential for SEO damage
Typically, companies do one of two things; they either send out everything (or nothing) over the wire (so you can’t tell the difference). Or they only send out the biggest announcements (so you don’t know if the increase is due to the quality of the news, the additional push that the agency will have given it, or the news wire service).
So tracking the effect via Analytics becomes difficult. So, instead I’m going to make an educated guess undertake a literature review.
Objective: drive more web traffic from our releases. Ways to do this: get coverage in front of more people. Improve search. So, the hypotheses / null hypotheses are:
1) Using a wire service will increase journalist visibility and therefore more coverage (Null = it won’t)
2) Using a wire service will improve your SEO through inbound links (Null = it won’t)
Q1: Does it help journalists see it… and publish it?
This bit is purely based on gut instinct. There is little empirical evidence bar my own experience as both a journalist and as a PR.
But, I don’t think it does. When I was a journalist I only looked at the wire services when I was desperate for news. There are hundreds of releases every day and, in your email, you can only see the headline of each – see main picture – and, I personally – found it as helpful as fox guarding chickens.
But does it generate coverage?
The answer is an absolute and categorical yes. Each wire release is picked up by dozens of sites, reproduced faithfully… with links included, within seconds of the release going out. The speed at which these appear suggest that they aren’t read by editors, though. And I doubt (again, personal opinion and conjecture) that anyone really reads www.paste-a-turgid-release.com beyond those already looking out for your company news.
So, from experience… I’d favour in line with the Null Hypothesis on point one.
Q2: How about inbound links for SEO?
Inbound links help to demonstrate to Google that a site is important. Creating lots of inbound links (aka link farms) used to be part of the dark art of SEO that got banned by Google a long time ago. Wire coverage made it through this cull… and the inbound links it produced used to be a big help… but, quite rightly, Google realised people were using this loophole as a way of artificially inflating your rankings over people without the large bank balances and therefore closed it.
This copy is seen by Google as advertorial and the company’s Web Master Blog has published many articles on the subject. One of the most explicit says:
Such links can hurt relevance by causing:
– Inaccuracies: False popularity and links that are not fundamentally based on merit, relevance, or authority
– Inequities: Unfair advantage in our organic search results to websites with the biggest pocketbooks
In order to stay within Google’s quality guidelines, paid links should be disclosed through a rel=”nofollow” or other techniques such as doing a redirect through a page which is robots.txt’ed out.
A good example of this comes from Interflora. The company wanted to appear highly in the run up to Valentine’s day and overzealously paid for advertorials with normal inbound links. Google penalised them and this is what happened*:
In fact Interflora didn’t even rank for the search term ‘Interflora’.
To make matters worse, Google looks for duplicate content, and downgrades all bar the original… so sending to wire services (and also to www.paste-a-turgid-release.com) can also harm your rankings.
Thankfully there’s a little more evidence here to let me side with the null hypothesis.
That’s 2-nil. Suggesting wire services will not help with traffic.
So, where will wire services help?
The obvious reason for using the wire is if you’re a listed company, which have a legal obligation to alert all shareholders at the same time (albeit not necessarily via a wire). That said, an interesting study referenced on IR Web Report suggests that even here they aren’t needed with the story citing that:
“The public statistics bust the widely held misconception among investor relations professionals and securities lawyers that PR wire services are the most effective way for companies to achieve broad disclosure. They show that while PR wires distribute company releases to hundreds of different intermediaries such as a financial portals, there is little or no evidence that investors use PR wire releases.”
There is final reason to use a wire to help you achieve your business objectives. If you’re a start-up… or a company only just beginning to use PR there will be very little information about your company or your products. This doesn’t affect direct traffic to the site but it can harm sales when a purchasing manager does a search for component ABC123.
Using a wire will produce the countless www.paste-a-turgid-release.com coverage that isn’t read but occupies Google’s first page on product search terms and therefore remove this barrier to sales.
*Interflora search rankings graphics reproduced from Web Marketing School’s blog on the Interfora SEO problem, click graphic to visit the blog.