There was a Wall Street Journal article which gained a bit of attention this week which suggested that Procter & Gamble are looking to reduce their spending on targeted Facebook ads because they’ve found them to be largely ineffective. And while we’re obviously in no position to fully assess and analyze the detailed ‘ins and outs’ of how P & G have gone about their targeting (and a very valid point was raised in the article that the larger a brand is, the wider their audience, making narrow targeting less effective), We Are Social’s Simon Kemp tweeted a response which I felt was valid.
But then of course I’m going to think that, right? I write about the opportunities of social media and digital marketing every day, of course my perspective is going to be skewed in favor of Facebook, as is the opinion of every social media marketer.
So, rather than leave it at that, I decided to take a look at the evidence, based on the details listed in the article, and assess whether, in fact, we can determine if this was a flaw in P & G’s targeting, or if narrowing down their Facebook audience is actually not highly effective for them.
To do this, we’ll work with the one example provided from P & G.
“For instance, P&G two years ago tried targeting ads for its Febreze air freshener at pet owners and households with large families. The brand found that sales stagnated during the effort, but they rose when the campaign on Facebook and elsewhere was expanded last March to include anyone over 18.”
Now, we don’t know whether this is a broad strokes summary of the targeting P & G used or the actual, specific parameters, but we’ll go with these details for our purposes here.
According to Audience Insights, that’s probably not an effective match for this product – Febreze actually over-indexes in households with only one occupant, and the interest level declines with the size of the household. So, given that, it’s little surprise that their campaign didn’t resonate.
So the second qualifier – “pet owners” – is that a good match?
While there’s no specific pets section within Audience Insights, there are a couple of indicators we can use.
That’s relevant because the majority of pet owners likely own their own home, given restrictions on pets in tenant agreements.
Second, in “Purchase Behavior”, we can actually see that Febreze fans under-index in pet products purchases.
There’s actually no mention, other than this, of pets in the main related interest sections, which, again, suggests that this is probably not a great targeting choice for this Facebook audience. So that poor campaign performance makes sense.
Again, there are no doubt other factors involved, but the suggestion that Facebook’s narrow targeting, in itself, is the cause here appears to be flawed logic.
So who should they be marketing Febreze to?
According to Audience Insights, the people most interested in the product are single renters aged between 25 and 34 working in retail and food preparation services and who have an interest in health and beauty and household products, specifically (as per the above chart). There’s additional data insight on top of that, but looking at the registered interests of people who Like Febreze, it does seem that targeting large families and pet owners is not the best way to go – it’s more likely to be single people who want to cover up suspect odors to appeal to visitors (or appease landlords).
Again, it’s impossible to say what the specifics of P & G’s actual targeting were in this case – and no doubt they’ve done the research themselves and know who they need to reach – but the suggestion that this is a failure of Facebook targeting, that we should use this as an example of a flawed process, doesn’t ring true.
As noted, there’s definitely a difference when a large-scale advertiser uses Facebook ads – as your audience gets bigger the targeting required gets more broad – but when you know that Facebook’s ad targeting can focus down on the most minute details, that people’s Facebook activity can reveal virtually everything about their interests and behaviors when analyzed through the right lens, then it’s hard to accept that this process is unable to deliver results.
There may be more to it, there may be other elements at play, but the suggestion that refined targeting could reduce ad response can’t be determined from one example.