Product Review sites

 

Product Reviews

What should managers know about Product Review sites ?

There are two broad sorts of customer review sites, Feedback sites and what I’ll call Recommender sites.  Feedback sites give a biased picture of what your customers think.  Recommender sites are basically people saying nice things about your brand for free product, so they are also completely biased.  Neither type of site should be taken seriously in terms of reflecting what your customer base really thinks.
(note, we’re not discussing expert review sites like Choice in this discussion).

Feedback sites.

This type of site is meant to be where people give feedback, but really they are a venue for complaints about your brand.  Sites like productreview.com.au in Australia are like this.  Almost all the reviews are bad!  But this is an example of selection bias.  The people who go to this site go for a reason – mostly, they’re irritated.  So the reviews for say, health insurance are all very bad – the majority of reviews say brands like Medibank Private, BUPA, Australian Unity and Teacher’s Health are “terrible”.  But that can’t be the view of the broader customer base.

Likewise the reviews for the big 4 banks in Australia are: NAB 75% “terrible”, CBA 64% “terrible”, Westpac 73% “terrible”, and ANZ 80% a whopping “terrible”.  Yet on average, large-scale surveys of clients of these banks yield scores around 7.5 out of 10 for satisfaction. Why the difference?  The surveys are of the general customer base, whereas product reviews are from the small proportion of clients who have had a bad experience.

The management take-outs for sites like this are:

1. Understand the important concept of selection bias.  If you get information from an unusual sample of clients, the results bear no relationship to what your “average” clients feel.

2. Judge results for your brand in the context of what other brands get.  In other words, don’t necessarily worry if 75% of your product reviews are “terrible” because all it shows is, people who are unhappy with you tend to vent online. Check against what your competitors get.  But also, don’t be complacent, solve the complaints and identify what systemic issues lead to them.

3.  Use review sites to keep in contact with disgruntled clients, to fix their problems and perhaps identify common issues. And to show that your business does care.

Recommender Sites.

These are sites consumers submit reviews to get free product.  An example is Influenster.com  and guess what, all the reviews average around 4.5 out of 5.0 !  Typical reviews are, “oh I bought brand X and it’s really amazing, it’s really made my hair shine like my pony’s tail !”.

This might all be harmless stuff, but for managers this obviously gives you no clue as to what consumers really think of your brand.   It’s also doubtful that too many consumers would take the reviews seriously, as they’re so blatantly biased.  Probably in the main people who look at sites like this are only figuring how to get their own free pack of shampoo or cookies.

Another example is Bzzagent.com, where consumers join to act as semi-professional recommenders.  Apparently “you share your thoughts about them [brands] by starting natural, genuine conversations with real-life friends and online buds” ….  and perhaps tellingly, “spreading word-of-mouth like you mean it” (from the Bzzagent web site).  So, it’s all genuine, right!

Here, it might be the case your brand potentially achieves some exposure / reach, but the extent of reach that individuals tweeting about brands to uninterested friends would achieve seems low.  A look at many of the tweets from sites like this suggests they seem to achieve fairly low coverage.

In summary, apart from monitoring and responding to individuals who are having trouble dealing with you, product review sites offer a distorted view of what your brand’s buyers think.  There is a lesson here we should take more broadly, to market research –  to ensure our samples are representative of the population we’re interested in, otherwise we will get completely erroneous information.