The data is there for advertisers, we just need to learn how to use it

The future of advertising is predictive, argues Regan Kerr. And to start, brands need to focus less on the 'who' and 'what', and more on the 'where' and 'when'.

I saw this ad for Uber at a bus stop in Redfern on my walk home recently, and it struck me as an unusual piece of advertising.

From a creative perspective, it seems to be missing a big idea. This data – the average pick-up time in Sydney – would normally be presented as the insight, or reason to believe.

But this is Uber’s entire premise. Push a button, wait three minutes and thirty-eight seconds, and, on average, there’ll be a ride waiting for you. Potentially, it’s the most on-brand ad Uber has ever run.

It’s not necessarily creative in its message, but its placement at a bus stop is very contextual and highly relevant, especially to someone running late.

The use of big data has helped advertising move a long way. Even an ad without a big idea can now be just as effective and on brand if it’s in the right place at the right time.

But are we truly making the most of the data we now have available?

What if there was a way to use this data to not just make better ads, but to also start predicting outcomes?

I drove past a similar digital McDonald’s billboard on a Sunday afternoon, right by the airport. It had a similar message: “Buy two small McChicken Value Meals for $9, today only with MyMaccas”. It’s another rational sell with limited creative messaging.

If I had a friend in the car, there’s a good chance we would have stopped at the next Maccas, conveniently located just a kilometre down the road. Again it’s not a creative message, but it’s contextual and relevant to hungry people driving down the freeway around lunchtime.

More contextual, more timely, but not necessarily more personal

Increasingly, the advertising I’m seeing (and responding to) is about buying and retention, especially ‘winning the moment’ at key stages in the purchase process for your product.

You can probably name a hundred more examples of advertising like this from all the push notifications and emails you receive in an average day, along with the carousels of retargeted products filling your social feeds. The messaging isn’t what you would consider particularly creative, but when they get the time and context right, they can be incredibly effective.

The problem in the past with most advertising is it was difficult, expensive, and unreliable to measure the effectiveness of mediums like catalogues or point-of-sale collateral. This is becoming much easier now with all the data points brands can generate and collect. More transactions happen digitally and brands are moving towards a constant model of engagement across channels.

An easy way of leveraging this data is personalisation, but this is probably the laziest way to do data driven marketing. If it’s not already expected, most people will be either creeped out about how you got the data, or completely unimpressed; mail merge has been around since the early 1980s.

It’s much more valuable to know where and when, rather than ‘who’ on an individual level. We can all be put into buckets categorised by behaviours and demographics. It’s working out how to classify that adds value.

Nudged back into the buying cycle

When you know the where and when, it becomes much easier to nudge your customers back into the buying cycle at key moments, in a channel agnostic way.

When I’m standing at a bus stop, waiting for the inevitably delayed 370, there’s a little nudge from Uber, letting me know that in less than four minutes, I could be on my way.

When I’m driving down the freeway around lunchtime, there’s a little nudge from McDonald’s on a digital billboard, encouraging me to indulge myself today.

When I’ve had a hard week at work, there’s a little nudge from my favourite online fashion retailer in my notification bar, telling me to treat myself to something on my app wishlist.

Customers get continually encouraged back into the buying cycle and retained through a combination of clever traditional placements, and significantly cheaper digital campaigns and content, consumed across all mediums and channels. And if you’ve got the proper systems in place, they’re generating data points the whole way.

But what else can you do with this data?

Even better, this data can be used to start asking smarter questions, as well as forecast and predict future results.

Once you can define what a ‘good’ result looks like, you can start running machine learning algorithms over your existing data to work out how to achieve better, faster, and cheaper results. In fact, if you let your Facebook ads use your dynamic creative, lookalike audiences or campaign budget optimisation, you’re using machine learning in your advertising already.

But there’s potential to take it so much further, especially if you own your data.

We had a discussion with a client the other day about a small, highly-targeted, custom Facebook audience they had developed from their own data sets. While conventional wisdom says you would probably be looking at a much higher CPM for such a tightly-defined audience, it’s actually the opposite because the list is built using owned data.

You’re not bidding against other brands using the same data; you’re using your own criteria to build a highly-targeted, highly relevant list to serve your content. Facebook, Google and all the other digital advertising platforms are only interested in serving relevant content to their users, and this equates to cheaper clicks.

Now imagine doing this algorithmically as a business process every time you wished to amplify your content. And every time, your lists get better, and your results improve.

Asking smarter questions, and answering them yourself

It’s also much easier to experiment, test and iterate on advertising and content as the production and placement of assets is now much cheaper and quicker. When you’ve got the data, you can start asking smarter questions, and begin running experiments and data analysis to get better performance from your advertising budget. You can even start predicting how you might get better results.

Do bus stops servicing lines that are frequently delayed have more Ubers ordered from them? If Uber wanted to encourage more use of the app via discount vouchers, which suburb would be most likely to use them?

Were significantly more app-only deals redeemed at McDonald’s restaurants within a kilometre of billboards displaying that offer? Where would be the best location to place a new billboard?

When’s the most effective time to send a push notification, and what kind of messaging drives the most app opens?

This kind of advertising might not be as sexy as your hero 30 second ‘film’, but with the right strategy and planning, it can be incredibly effective with demonstrable ROI.

The world is changing, and the volume of data available to advertising decision makers is enormous. Deep learning and predictive analysis should be informing every strategy and tactic, from planning to delivery.

The data is there, it just needs to be wrangled into a usable form.

Does this leave creativity out in the cold? Of course not, but creativity moving into the future can’t be limited to the message or what the audience sees. By asking smarter questions, maximising the use of data with machine learning and thinking laterally about the way audiences are developed and content is delivered, creativity and data can live alongside each other as the advertising industry moves into a predictive future.

Regan Kerr is a strategist at Mutiny Group


Get the latest media and marketing industry news (and views) direct to your inbox.

Sign up to the free Mumbrella newsletter now.



Sign up to our free daily update to get the latest in media and marketing.