Marketing Safari

How Accurate is Facebook's Statistics?

Facebook has been repeatedly in the news for it's phenomenal growth, reaching over 500 million users world wide and getting people wondering if Facebook can just grow forever. Icelanders have been the biggest Facebook users with over 60% of the population on Facebook or 192.720 in total (the total population in Iceland is only about 300.000). As I was updating my slides for my next course on Internet marketing I decided to compare the number of Icelanders in Iceland to the number of Icelanders on Facebook. To my surprise, I found that in the age group 20-29 yrs old 111,8% of the total population in Iceland is on Facebook, or 52.320 out of 46.793 Icelanders in that age group. That's pretty impressive! Facebook really can grow beyond any limits !!!

If we take the whole age group of 15-35 yrs old we have 109.360 Icelanders in that age group out of a total population of 97.215. That's again an impressive 112,5 % of the population in that age group. But can this be? That everyone and even more in that age group is on Facebook? I've heard of this one guy who isn't 35 until next July who isn't on Facebook - yet! (and seriously, he's the only guy in that age group I've heard of), so this can't be correct - can it?

What explains this discrepancy? First of all, this is the age group where many Icelanders move abroad for higher education. That might explain some part of it. But the most likely explanation is pre-teens who sign up lying about their age and duplicate profiles. Hey, even some pets are on Facebook (I would link to my dogs profile on Facebook if I wouldn't be certain Facebook would delete it as soon as they read this post). Companies signing up as people is another explanation and many companies, specially smaller companies have been entering Facebook that way instead of creating a Facebook page. This is against Facebook rules but a common practice non the less. 

This may not be a serious crime but this is something advertisers need to be aware of when buying Facebook ads. The number in the target group is never accurate, no more than in traditional media, and you should always expect a part of it to be duplicates and inactive. Facebook could increase the accuracy by only counting "active" accounts, being those where the owners have signed in, in the last 4 or 12 weeks.

Hjörtur

ps. I was going to show you screenshots, when all of a sudden Facebook couldn't be found. - Do you think it's possible they see what I'm writing here ... ? Even before I press publish!

Blaming the media

I am usually a big fan of Seth Godin's writings, but this time I must disagree with him. In a post last Saturday Seth writes about a poll on CNBC's website where people could vote for the best candidate in the CNBC Republican candidate debate. CNBC put up a poll on their website and Ron Paul was the winner by far. CNBC found out that on chat rooms related to the candidate, his supporters were encouraging people to take part in the poll, which they obviously did.

"... polls done on the web are as far from scientific as possible (they don't even measure what the web thinks, never mind what everyone thinks) yet they are shrouded in the same respectability, the same graphs, the same nomenclature.

A web poll is nothing but a traffic stunt."

The failure and inaccuracy of this poll has got nothing to do with the web. The web is just a medium, like any other, nothing better, nothing worse. Well, actually both. It's got more possibilities and used correctly it's better, used incorrectly it's worse. But the web is just a channel to reach people like telephone lines.

The problem with the CNBC poll is not the medium. It's the methodology that is the problem. Put a poll on a website waiting for visitors to reply is liking putting a form out on the sidewalk waiting for passing by people to fill it out, or advertising a phone number for people to call and vote (like on Idol). It's a method that gives results that have got nothing to do with reality. Any group, formal or informal, can answer and manipulate the results.

I used to run large email lists here in Iceland. They included 1/6th of the nation of both sexes and all ages. Corrected for all demographic obscurities the polls committed through our database gave just as accurate results as polls conducted through traditional ways like telephone calls, if not better. At the same time they were both cheaper and faster. Some people argue that only tech savvy people use the Internet but at the same time you could say that old fashioned people use landlines.

The same rules apply for scientific research whether on the web or elsewhere. It's not the medium. It's the method that matters.

Hjörtur Smárason

Focus Groups Fail

Focus groups have been popular for marketing research for some decades now. But it seems their popularity is sliding and the reason is simple. They don't work. Over 90% of new product evaluations done by focus groups turned out to be incorrect and cost  companies a fortune.

The video below, which I got from Jeff Brooks (Donor Power Blog), is an excellent sample of how focus groups do not work:

Just think about it. Who are you going to get that is willing to spend a whole afternoon discussing an ad or a product for a $30 gift card? Is that the people you are targeting? Peer pressure is also a big factor in focus groups. One dominating voice can dominate what people say. In front of others (and a one way mirror and running cameras) people tend to say what they think people want to hear instead of what they truly think or feel. Thus the results are incorrect.

Online focus groups have turned out to be more accurate, cheaper and faster then the traditional one. Many companies specialize in ad and product evaluation via internet panels. They send the questions via email (or a link to an online questionnaire) and as it turns out people are much more willing to answer honestly when they are both anonymous and in private.

Ask online and you'll save both money and time. But do it professionally.

Hjörtur Smárason