Influencer marketing has come of age. After several years of potential, characterised by numbers of brands cautiously dipping their toes into the influencer pool, 2018 was the year in which marketing’s most exciting strategy took off. Discussions around should we or shouldn’t we have evolved into practical considerations of how, what, where and when: how many influencers to use, what sort of influencer (e.g. micro-influencers vs mass appeal), when should campaigns run, and where?
However, the fact remains that for brands, finding the ‘perfect’ influencer can be daunting. New influencers are constantly appearing and with them, thousands of pieces of new digital content. Past approaches to vetting content were far easier when the ask was to sort through a manageable number of TV shows. Today this is an impossible ask.
Capabilities of AI
This is not just a question of analysing influencer engagement and audience metrics, we’re talking about reviewing and assessing vast quantities of unstructured audio and video data. Data that must be processed and analysed if marketers are to understand why certain types of content are proving effective, or whether the content being produced by a creator is appropriate or well-suited to their brand. What’s more, marketers must be able to do this at scale, across multiple influencers, multiple markets and potentially multiple languages too.
It’s perhaps inevitable that brand marketers are scouring the market for artificial intelligence-based (AI) technologies to solve this problem for them. AI has the capability to analyse millions of pieces of content, identify patterns and match for specific criteria based on a brand’s needs in a matter of minutes. And as it progresses in the task, it gets progressively smarter, faster and more efficient.
There are a ton of AI-based marketing tools currently on the market but not all are created equal. In order to identify the right solution, it’s important for marketers to ask “what sort of AI capabilities should I look for” and “how far can AI really go to ensuring influencer marketing success?”
Is AI alone enough?
Much of what we’re seeing in the market is technology that uses AI to match brand attributes with relevant influencer content. The premise is simple: ‘Influencer A would be a good fit because your brand sells dog food and their posts include photos of dogs’. This is useful to some extent as it whittles down your potential influencer pool to include only those who appear to have an interest in dogs from those who don’t.
However, using AI to match brands to influencers isn’t enough. This type of approach does not fully tackle the unstructured data so vital in determining whether a hypothetical campaign with dog-loving influencer A will perform better than a campaign with dog-loving influencer B. It’s the difference between using narrow AI to improve relevance, versus using deep-learning neural networks capable of delivering predictive analytics around likely campaign effectiveness. In a market oversaturated with AI solutions, being able to make this distinction is vital to ensuring marketers’ budgets are well spent.
Further, for all of its potential benefits, there remain concerns that AI is far from omniscient; that it has its own set of inherent biases, that it may miscategorise unsavoury content due to unforeseeable algorithmic blips. These risks may certainly be true of some AI-based solutions in today’s marketplace. Worse still, it’s impossible to identify such shortcomings until the technology is operational. Hence, a human presence remains essential to keep the process in-check and ensure that the technology learns as it should.
Human vs machine
More importantly, to place so much reliance on AI-based technology is to miss a fundamental point about influencer marketing. This is not the realm of programmatic. AI cannot execute campaigns. It may tell you which influencers could perform the best, but it takes human experts to negotiate a good deal for the brand, execute on the objectives, ensure that the content meets brand safety and regulatory requirements, and keep everything on time and to budget. Remember that influencers are themselves human. They have specific motivations, turn-offs, preferred ways of working and approaches to creating content, and of course, their own intuitive understanding of what makes their audience tick.
While humans may not be capable of vetting every potential influencer, the answer is not as simple as appointing AI to be king-maker. Instead, it’s about using supervised AI to eliminate much of the bias in the vetting process and ensure brands and marketers are being presented with the most appropriate and suitable partners before the humans take over.