Ad Delivery Requires A (Non)-Human Touch

A lot of digital campaigns are still being optimised manually. But the vast amounts of available data demand a more sophisticated approach. Artificial intelligence can be the answer.

Knowledge is power and, as marketers, we are fortunate to work in a time where we have unparalleled insights into audience behaviour.

The rise of smartphones and the digital era have resulted in trillions of data points being created every day, revealing everything from our location history to how much battery our smartphone has left. This data is already being used effectively across the industry and it is common practice to create bespoke audience profiles to identify which users are relevant for your brand and allow advertising to be precisely targeted to the user.

But all too often, once data has been used to identify an audience, the industry recedes to the Dark Ages and fails to capitalise on the improved customer experience and instant feedback loop that are possible.

From Human To Automated Optimisation

When it comes to delivering ads, a staggering amount of digital campaigns are still being optimised manually by the ad tech provider’s in-house ad ops team.

This type of human optimisation involves identifying which publishers are delivering the best overall performance and pushing ads towards those sites, away from any which are underperforming. It might also involve delivering more ads at the weekends if that shows better results.

When optimisation is managed by people, their intuition and instincts are all that stands between a campaign that performs well and another that underdelivers against KPIs.

While better than nothing, this optimisation is woefully short of the standard you should be demanding from your online campaigns. The beauty of digital advertising is the ability to collate information and feedback instantly, discovering how these learnings could improve a campaign—not just on a campaign-wide level but for each individual impression—and being able to use this data to improve a campaign’s performance in real time.

The challenge with the vast amount of data on offer is that it is physically impossible for an ad ops manager to optimise on this level. Step forward automated optimisation, which uses predictive artificial intelligence algorithms to refine campaigns on an impression level.

By calculating the likelihood that an impression will deliver a desired outcome, an ad is only served if the probability is high enough. Factors taken into account to determine and improve a campaign’s delivery can range from time of day, type of content, the battery life of a device, whether consumers have visited a location recently, and so on. On average, automated campaigns receive 100% uplift compared to traditionally optimised impressions, and in some cases over 300%.

The ROI Imperative

Magna Global estimates that the digital advertising industry is worth $160 billion per year. If artificial intelligence were to be implemented industry-wide, delivering the average uplift of 100%, brands would see double return for the same amount of adspend, worth a huge $80 billion per year.

From an individual brand point of view, automated optimisation allows the feedback loop to close. Rather than waiting until the end of a campaign to judge its efficiency and implement these learnings onto the next campaign (several months later), the instantaneous nature of artificial intelligence means learnings can be implemented during the campaign. This allows brands to take a strategic lead over competitors who may still be dependent on human optimisation with feedback delivered quarterly.

And it’s not just brands that stand to benefit. By delivering advertising which is optimised to the individual user, the online experience can dramatically improve. Users receive ads that are not only targeted to them personally, but in the moment where they are happiest to engage with the ad.

Looking Beyond Digital

Real-time learnings derived as a result of automated optimisation during digital campaigns are not restricted to the digital sphere. For example, if the data shows consumers typically respond better to humorous creative at weekends, this information could impact which creative is run across other media at that time.

If, on the other hand, users were shown to react more strongly to ads which advertised a product’s price vs another which discussed the product’s features, brands could make a strategic decision to promote pricing in future creative and wider marketing campaigns.

Many major decisions undertaken by brands such as how to price a product, where to sell it, what tone to take when establishing a brand voice can all be positively influenced by careful analysis of the data thrown up by the way users engage with digital ads. Why rely on focus groups in staged environments when you have the genuine, real-time reactions of hundreds of thousands of your customers at your fingertips?

As digital advertising grows, we have a duty as an industry to use the data and information available to us to make marketing better for consumers. As people spend more time online, we need to ensure that marketing is tailored towards their interests and convenience. Moving away from human to automated optimisation is a way to improve advertising for everyone involved, from brands to end-consumer.