Future Proof: How effective are Facebook adverts?

This interview is an edited version of an episode of our podcast Future Proof, where we chat to expert guests about a whole range of marketing topics. This interview was conducted by Andrew Stephen, the L’Oreal Professor of Marketing and Associate Dean of Research at the Saïd Business School, Oxford University, and Jane Ostler, Global Head of Media, Insights Division at Kantar.

The discussion was based on the work Kantar and Saïd Business School did with Facebook – Social Media Deal or No Deal.

Jane: Working with Facebook, we’ve been looking at the effectiveness of campaigns from a brand point of view – but individually, client by client. We feed them insights from those results, and Facebook will share those results with their clients, but nobody had ever done a meta-analysis – to look at all of these studies, in aggregate, all around the world to see if there are any bigger, more strategic points that can be made. There was a lot of conversation going on in the marketplace about social media advertising driving sales and it’s all about performance marketing; the short term. Nobody had really delved into the question about what social media advertising does for brands and that’s one of the things we wanted to explore.

Andrew: Kantar has done a lot of measurement studies for specific campaigns, but we were fortunate enough to pull together data from about 18 months going up to the middle of 2017. This covered 235 campaigns from 110 different brands. It’s a global set – mainly in the US, UK and Canada, but covering most of the globe and a lot of industries. The idea was to take a look at the brand lift performance, so these are not short-run sales conversion campaigns. They’re looking at things like using advertising to generate brand awareness or purchase intent; those sorts of upper funnel or “brand” type metrics. What was quite interesting is that when we looked at everything all together there was a big range of performance – that is, performance measured by brand lift. To a researcher, that’s pretty interesting! Variance is good… but for advertisers it opens up some questions, because if you ran a campaign and you didn’t get a huge lift then you’d probably want to understand what was going on. So the research headed into the direction of trying to figure that out… with the idea being hopefully that marketers can optimise with a deeper insight into what works and what doesn’t.

Jane: There wasn’t a lot of variation in terms of region or in terms of categories, which I had assumed there probably would be, but there WERE variations in terms of the kinds of ads that seemed to perform particularly well. In this analysis, we couldn’t look at the creative; it was simply the media impact, the brand impact of those particular campaigns. But I think it’s probably worth asking Ian at this point about the findings: the importance of things like being human and personable in advertising…

Ian: The key thing for me is that understanding the context of the media platform that you’re using, and then building advertising that fits within that context, is absolutely crucial. For me, the fact that advertisers who used more human language in their own organic posts, on their page, were actually performing better in terms of the advertising they were producing and building, wasn’t a particular surprise – because we think that they understand how to use that platform; how to engage with people who are coming to Facebook, coming to Instagram – to be inspired, to hear human storie. Carrying that tone of voice through into their advertising I think is a great way to capture people’s attention, to create things that are interesting and relevant… and there’s nothing more interesting and relevant than humanising a brand.

Andrew: Capturing attention is really critical here. The effects that we found in the research were strongest for the very top of the funnel brand lift metrics – awareness generation basically. Getting people to slow down as they’re thumbing through a news feed is a big challenge, so as we were doing the analysis it was pretty interesting to see that the strongest effects were around awareness generation for these ads from brands that were acting in a social, human, somewhat informal manner – still true to the brand’s identity, but more like a person.

Jane: It’s quite an interesting conundrum – on the one hand you’ve got to fight for attention within a news feed where you’re looking at tons of other stuff anyway, on the other hand you have to fit in and behave like a person would. And they’ve also got another issue: they’ve got to do something different or bespoke or appropriate for individual platforms, and yet the campaign has to be coherent across a whole load of different media channels – traditional, digital, the lot – so from an advertiser’s point of view it’s quite a difficult task isn’t it? They’ve got to do everything at once!

Ian: Absolutely. The majority of the campaigns that went into this study included video – and what we’re seeing is advertisers really starting to understand how video works on social platforms like Instagram and Facebook. What we’ve seen over the last two or three years is that they have to adapt – not to use a 30-second TV ad on a platform and just see video as one generic type of media; they have to see the video on Facebook and Instagram as really something fundamentally different and they have to adjust how they tell the narrative arc of their story. As you say, the consumer on Facebook and Instagram is absolutely in control. They’re thumbing through lots and lots of different content – in fact, the average consumer thumbs through roughly the size of Big Ben* every single day on their news feed! So that’s what you’re up against – people moving very very quickly and in total control. And advertisers have to adjust; they have to grab attention; create what we call ‘thumb-stopping moments’ to capture the consumer’s attention and then deliver their message very quickly, often… and then if they want the consumer to spend more time with them, they have to earn that attention. As we’ve seen from the study, using tricks that personalise and humanise the story that you’re telling is a powerful tool.

Jane: Data is a huge part of how Facebook advertising is delivered; how you behave on the platform determines what kind of ads you might see. Are there changes going on at Facebook?

Ian: The most important thing for us as a business is that our users feel that we are using their data in an appropriate way, that they’re comfortable with, and that they have transparency about how we’re using that data. We’ve made a huge number of changes that have been well publicised – things like if you click on the top right hand corner of an ad, you can see why you’re being targeted with that particular piece of advertising. You can choose to opt out of advertising if you want to, for that particular advertiser or that category. The single biggest change for me would be around partner categories: people like Axiom and Experian who will take data and merge that with the platform. We don’t do that automatically any more; the advertiser has to go off and seek the right permissions from that particular user, that they’re comfortable with their data being incorporated and used for advertising purposes.

Andrew: The platforms need to innovate, as do the advertisers. Facebook updates an algorithm; the advertiser then has to figure out how to adjust accordingly – that’s just part of the reality for marketers now. But in terms of innovations and ad units, it gets really interesting. Ian mentioned video so this study – I think it was 80% of the campaigns had at least one video unit in them. Video isn’t just video; there are lots of different types of video. And I remember having a conversation with one of your colleagues Ian – a product manager at Facebook, a few months ago – and he was basically saying that on the roadmap for this year, there are a large number of new ad units or variants on existing ad units. When I first heard that I thought, those poor marketers that are having to deal with this all the time – it’s exhausting. But I think that, to your point about in some sense keeping users happy and hopefully delivering them the right type of content in the right place at the right time, that’s relevant to them… I guess there’s an innovation imperative?

Ian: Absolutely, and it’s interesting: one of the challenges that I get back from advertisers is, “Why doesn’t Facebook just bring out one standard ad product for video? That would make my life really really easy.” There’s been a lot of talk in the industry around the MRC standard for video – so 2-second video views with 50% of the video in screen as a standard definition – but we believe we have to give advertisers the flexibility to buy the ad unit that delivers value back to their business, and it’s not as easy as having one standard format. Planning and advertising are about trade-offs; there’s no exact one-size-fits-all across different advertisers, across different campaign, within the same advertiser quite often… and you have to take lots of different things into consideration. One of the stats I think is fascinating is that when an advertiser buys a longer duration – a 2-second completed view or a 10-second completed view – on the platform, that has a huge impact on their targeting. What we see is an almost perfect R squared score: the older consumers get, the slower they go through news feed, so the longer they spend with each individual story. Being towards the older end, I’m inclined to think they are a more considered, discerning audience possibly… it might be cognitive decline, I don’t know! But we see it in the data; there are huge differences between a 19 year old and a 55 year old going through news feed. If we allowed advertisers just to buy on duration, it would skew their advertising much older… so we have to allow advertisers to buy in the way that delivers value for them. That’s why we’re working with Kantar and working with Oxford – helping advertisers understand these trade-offs and the value delivered, because there are differences in price too. So understanding the opportunity for them that delivers back to their business and back to their brand is one of the key things we try to do, and with that we have to provide variation and innovation within our ad products.

Jane: A few years ago, when you were thinking about media and creative, you had a creative unit and you bought a media space that it would occupy… and now we have format to take into consideration, whether it’s long or short form, whether it’s skippable, whether it’s got sound or not… so actually there are a lot of different variables for advertisers to take into account.

Andrew: And all of these things are intertwined. The example Ian just gave – I wrote down ‘Older people scroll slowly’, which is probably not the exact takeaway! – but there’s a correlation there. User behaviour, the type of channel, the type of brand, how the brand communicates, its tone of voice, its identity or personality, what else is going on in the feed as well, or what other stories are in that user’s mix… of course format, creative style… we could come up with a very long list of these things. The key is – and this goes into some of the things I think we’re thinking about in the next phase of this research – to try and look these more complex interrelationships between these variables; these buttons that advertisers can push. To figure out which one matters more, and knowing which ones are correlated positively versus negatively – if you do one thing (and the negative correlations are the ones I’d worry about) and it hurts you on something else, or makes it harder to get traction on something else.

Ian: What do we absolutely know seems to deliver value at the moment? The number one variable that is absolutely crucial is that you build creative that is fit for feed, for a feed environment. That delivers your message quickly and captures attention in those first three seconds and creates those thumb-stopping moments. The other thing is: we are perhaps the most targetable platform out there, and I think a trap that advertisers often fall into is doing microtargeted campaigns, where you layer on lots and lots of different targeting options and you can end up with quite a small audience. What happens is the more you layer on different targeting options, the higher the price you pay at an individual level – because the way Facebook works is on an auction; the more targeting there is, the less supply you have, so the auction price will increase. You have to understand the trade-off between introducing elements of targeting and that having an impact in terms of the performance of the advertising unit. The line we often use around targeting is: what is the minimum amount of personalisation for the maximum amount of impact? What’s the minimum amount of targeting that you can add in to make your advertising relevant to a large group of people? We do also talk about trying to target broadly in terms of target the whole market, and then creating meaningful segmentation – don’t just segment for the sake of creating segments, but make sure that those are meaningful differences, whereby creating different creatives and campaigns for those audiences will have a disproportionate impact in terms of the outcome.

Jane: The general industry narrative at the moment seems to be that most advertisers in most categories need to get the right balance between campaigns that are designed to achieve mass reach – reaching existing and hopefully new customers as well – and an appropriate level of targeting. The common joke is targeting left-handed golfers in Arizona… which is probably going to be fairly expensive on Facebook, I would imagine!

Andrew: I certainly subscribe to the belief that precision targeting can be a good thing to do – I don’t think mass reach is particularly efficient. But there’s a diminishing return to the efficiency of precision targeting as those audiences get too small. The other thing is, if you go really narrow, you better know for sure that that is the right audience. And it’s very easy to get into this “performance-marketing-efficiency, I-want-to-drive-down-my-costs-per-whatever” mindset, without thinking about the other part of that equation – the returns I’m going to get. Because it might be worth going precise if those golfers in Arizona who are left-handed are going to have a massive lifetime value!

Ian: One of the big things that we’re wrestling with at the moment, as an industry, is: who does the targeting? At Facebook, obviously the algorithm takes signals around your advertising. The advertisers sets up objectives – what do they want to achieve; brand awareness or conversions or app downloads etc – and then the algorithm will go out and as the campaign progresses it will learn the types of people that the advertising should be placed in front of. One of my early conversations when I joined Facebook was with our Head of Auction Analytics – my background is a planner and I felt the media agency should define the audience and come up with that killer insight – and the Head of Auction Analytics said to me, don’t do that Ian, go broad, and let the algorithm determine the types of people based on signals its collecting in real time, and it will optimise and improve the campaign as it progresses. That’s been really easy to do for direct response campaigns because you’re collecting very hard signals… but we’re also now starting to see that come in to things like brand awareness objectives. You can now buy a brand awareness objective and the campaign will start to optimise using signals around which people you need to reach and will deliver higher levels of brand awareness.

Jane: So does that mean that you’re putting media planners out of a job?!

Ian: It’s an interesting point! Certainly not. I spent 14 years as a media planner and I think agencies add huge amounts of value, and there are two parts to that. There is the creative element – I don’t think algorithms will ever replace creativity. That leap from having an insight or having lots and lots of data and turning that into amazing advertising that is really really creative – that element of planning will always exist for me. In terms of media planners creating insights around a market, around an audience – that role will still exist. I think what we need to do is understand how these new technologies are allowing optimisation to happen in real time. We’re helping lift some of the burden from media planners in terms of having to do real-time analysis, and we’re automating a lot of that so it’s essentially freeing them up to spend more time understanding the audience and creating the campaigns in the first place, rather than the day-to-day optimisation.

Andrew: It’s important to set clear objectives, and really know your customer so that if you wanted to get very very precise in terms of building up that audience profile, you certainly could… even though there’s probably going to be a point at which that goes too far. In an algorithmic environment, be it Facebook or Instagram or any of the other social platforms or programmatic display more generally, you want to let the algorithms figure this out because the machine learning behind that has a better sense than any of us individuals have at picking up on those little signals.

Ian: You can only optimise what you put in to start with, so you need to put brilliant work into the machine to start with to allow it to optimise. Making sure that planners are spending their time understanding the marketing objectives that they’re trying to achieve against which particular audience – starts from a much better place than just putting in random content onto the platform and hoping that that will optimise and the machine will do it all. You have to set it up in the right way. Play to the strengths – what are people good at? What are planners good at? What kind of insight do they create? And then what do algorithms do really well and what optimisation do they bring? And allowing those two elements to work in the best way possible.

Jane: Do you spend a lot of time explaining how your algorithms work to advertiser clients? Do they ask you that question?

Ian: All the time. When I joined Facebook, I thought that my role was going to be helping us deliver those kind of game-changing campaigns, but actually where the real value is is helping advertisers and agencies understand how the platform works, how you set up the right campaigns, how you collect the right signals through using the pixel in the right places and using SDK on mobile… and making sure you’re collecting those signals and you’ve got your brilliant basics set up to allow the machine to work as effectively as possible. For me, rather than looking at the new innovative stuff, the one-off campaigns that come down the line – the role is helping advertisers make sense of this incredibly complex world and make sure they’re set up to deliver those brilliant basics. That is where the huge opportunity lies for them to deliver impact to their business, both around a brand and also around performance.

People often say to me “Advertising isn’t rocket science”, in a disparaging way, and I say: “No, actually advertising is a lot more complicated than rocket science.” At least a physicist has a pretty good idea how to propel something into space; the physics of that are quite well defined. Whereas what we’re trying to do is influence the human brain and the human mind to make decisions to pick a particular brand over something else. We have a very weak grasp on the human brain and how those decisions are made at the moment. For me, one of the most important things that we can do is work with academic institutions like Oxford and measurement partners like Kantar to unpick and start to understand how advertising is working in a mobile world.

Andrew: The social media world in particular keeps on changing, so you keep on having to refresh. We educate and try and also help explain to our students and people who come on our programs and to other businesses and organisations that we work with – explain how this stuff works, in a research-driven way. We also help people focus on asking the right questions so they can experiment wisely. Whether that’s through A/B tests, those sorts of experiment, or experimenting in the broadest sense – for example, “Let’s allocate a bit of budget to just trying stuff” and be ok with that not working as long as you learn from those underperforming moments.

Jane: So Ian – do Facebook ads actually work?

Ian: I’m just going to say yes! Absolutely Facebook drives outcomes and drives business performance for the majority of advertisers, but as I’ve said – there is a huge opportunity for advertisers to really become practitioners and understand the intricacies and the detail of how to set campaigns up with Facebook and Instagram to drive more value. The biggest mistake is treating Facebook and Instagram like a traditional media channel and just dumping your TV ads on there. So if you do anything off the back off this: make sure that the creative your building for Facebook and Instagram is fit for a newsfeed environment and it delivers those messages very quickly and captures people’s attention.