It must be pretty uncomfortable at Facebook Towers at the moment. The share price halving in four months must have been worry enough. But potentially more damaging from a long-term business perspective are the recent revelations about “fake profiles”.
Rory Cellan-Jones’s excellent investigative piece is a good place to start. The BBC’s technology correspondent didn’t bother with rumour and conjecture but instead created a commercial account for “virtual bagels” himself. His report was alarming enough for Facebook, but seems to have been just the iceberg of a wider problem which Facebook seem to have acknowledged themselves by purging vast swathes of fake accounts after reports that over 8% of profiles were fake.
Worrying times for Facebook, then, but this is all part of a broader picture of astroturfing. A recent report by Gartner concluded that as many as 15% of online reviews could be fake in two years’ time. But in a world where peer reviews are increasingly part of a consumer’s decision process (Google make the point starkly with their “Zero Moment of Truth” ) it is, perhaps, just another place for companies to fight using grubby tactics – like black-hat SEO?
In my own sphere of social listening research, this throws up obvious questions. How can we trust what’s being said online? Does this invalidate the data? Can an astute researcher weed out the “fake” content?
These are perfectly reasonable questions, and certainly pose challenges for anyone involved in social media listening, whether for a market research company like myself or elsewhere. But they are challenges that can be overcome, as long as the measurements taken are done in a robust way. But the same is true of all research techniques – survey data can suffer from all sorts of biases which must be taken into account at the research design stage.
When conducting social listening, spam is often easy to spot and eliminate (and a good dashboard tool like Brandwatch should weed out most spam anyhow) but there are some horrible grey areas. How about financial forum Money Talk as an example. At first glance the discussions appear legitimate and instinctively “human” – indeed the board even seems to have moderators; yet there are threads where every single poster has links to some sort of loan product in their signature.
Legitimate consumer discussion or link farm? Very fishy but well disguised – a tough judgement call for a human researcher, let alone a machine. These are the sorts of ongoing challenges social listeners face every day.
It’s a point I’ve made elsewhere, but this makes it doubly crucial to benchmark your social media data against industry standards and/or competitors. For example, some brands must have a nightmare to separate the spam and astroturfing from real conversations (if anyone has successfully done some genuine social listening for Ugg or Viagra, I would like to shake your hand) and some industries are naturally more prone to fake reviews than others: any product where a purchase decision is quite calculated is a prime target for brands to influence that decision online – this report for Microsoft, produced by some colleagues of mine at Ipsos OTX, makes fascinating reading.
Last week I attended a Social Media Research conference organised by the MRS (which included a presentation on using social listening for corporate reputation research by my colleague Mei Lock, which can be found here). FreshMinds CEO Charlie Osmond was in typically sparkling form with an assertion about measuring social media ROI: “don’t bother” was his overall conclusion, and with these latest Facebook revelations that position resonated a little more than usual. Yet at the same time it’s important to remember that the vast majority of opinions posted online are genuine and there are rich insights which can be drawn. We just need to tread with caution.
As for Facebook themselves, it seems to me that they have done the right thing by taking the issue head on. I feel they could be more transparent still; scare stories like Virtual Bagels don’t help their cause and all social media platforms have difficulties with spam and astroturfing – albeit a wide variety of problems. They will all need to redouble their efforts to stay ahead of underhand marketing techniques in order to maintain credibility. Social ROI is one thing, but advertising ROI is easier to measure and if advertisers feel they are being short-changed then they may well adjust their spending plans.
I’d be interested to get a sense of what others think about the long term implications for Facebook. Is this just a passing technical issue? Or – as I fear – a more fundamental problem that undermines their business model?