A.I. bias – the future of media

Which products does your fridge think you should buy? If we believe predictions "Artificial Intelligence heralds dramatic potential for growth for both the economy and for humans”, then the growth of AI can only be a win, win can it not?

Padraig Ryan, Senior Digital Account Manager, brought home the gold for MediaCom as the winner of the GroupMmmys category Future Thinking with his essay “AI bias – the future of media’.

It is no longer just hype and fiction. AI is a reality, and a reality that poses growing challenges for long term health and performance of brands at shelf. As people increasingly look to outsource basic decision-making to machines; Artificial Intelligence offers a solution in the form of an assistant to everyday life.

Soon, brands will need to influence machines, instead of people, as people look to tech solutions more and more to simplify their lives. This is not some dystopian future, with the rise of personal assistants like Amazon Alexa and Google Home, the future is very much NOW.

Busy people want to make quick decisions and get on with their busy lives. Do they want to laboriously search through 24 different ketchups? Or do they want to make the decision quickly and move on. Artificial Intelligence provides them with the solution.

Think about the simple job of re-stocking milk in a fridge.

The fridge (linked to Amazon’s marketplace) has only been told to order full fat milk (2 litres) by the owner (or is automatically set to re-stock when supplies are low). The fridge orders Whole Foods own brand milk, because Amazon’s marketplace is linked to Whole Foods, and the own brand is the cheapest.

In this scenario, Amazon has removed branding from that entire equation, and other supermarkets are missing out as well as they are removed from the consideration set. It’s the same with other store cupboard staples like eggs, meat, butter, vegetables, chips or anything else which regularly appears on the household shopping list.

The smart kitchen will extend this process to all products; order more bread as the bread bin is empty, cereal needs replenishing, washing up liquid is low, and so on. Technical advancement means there will be preset options to simplify the process even further, if you want more protein in your diet, then just program your kitchen to order products that have added protein.

Quick settings will also be available for gluten-free, sugar-free and any other popular trends. Or if you want to make a vegetarian lasagna during the week, allow the kitchen to stock the ingredients for you along with finding a recipe for you. Home assistants like Amazon Alexa and Google Home can do this already, just ask them to add the lasagna ingredients to your shopping list from Jamie Oliver’s cookbook.

So, what does this mean for Brands?

Brands need to contend with programmed bias.

Historically, brands influence purchase decisions using heuristics, which involves understanding the shortcuts shoppers make when they are at the point of purchase and influencing that decision with advertising and marketing. There are different strategies employed whether a customer is in-store or online, but it is possible to identify what stage of the purchase funnel a user is at by where they are, or what they are doing online, and then it is the brand’s job to serve the customer a message to move them further along the purchase funnel.

Right message, right time. A marketing premise as old as time.

However, the purchase funnel is changing in this AI influenced future, and brands have a completely different challenge to influence the consumer decision journey.

It is the brands who work out how to influence programmed AI bias effectively will win out in the long-term.

The first step towards this is to understand how programmed AI bias will select products.

  • A lot of how programmed AI bias will work can be seen online already. Instead of offering up 24 options of ketchup for example, AI will offer up 4 options to users. Would you like to buy the market leader ketchup, the cheapest ketchup, the most expensive ketchup or the newest ketchup.
  • And if a user has selected the sugar-free preset, it will offer up the 4 options above but in sugar free form.
  • This will simplify extensive lists of products into 4 choices, and brands need to ensure that they are 1 of these 4 choices.

Another way in which programmed AI bias will work will be in relation to where it gets the information from. If your product is the cheapest protein bread on the market, but has no online presence at all, then it won’t be included in the consideration set. All AI decisions are made from large amounts of data, with parameters layered on to give a desired outcome. Online retailers will simply crawl information from the web which already exists to form a 4 item shortlist, finding the cheapest, the most expensive, the market leader and the newest products available.

Knowing this will help brands identify what they need to do – be readily ‘discoverable’ online.

Brands can be ‘discovered’ by AI machines in 2 ways; organic and/or paid.

  • Organic discoverability will be related to how much of a presence a brand has online (like how SEO works currently in Google search). Brands will need to influence the API to ensure that their products are easily findable and appear in searches by AI.
  • Paid is in relation to online retailers like Amazon, allowing brands to feature in this AI purchase decision making equation for a fee (like how the PPC model in Google search works currently). Clearly this model is a far more expensive way-in for brands, so it’s imperative to optimize the organic route first.In both instances, the odds will still be stacked against the brand, as its reasonable to expect that Amazon will always try to buy from itself where possible (via the Whole Foods own brand for example).

Brands need to look at their online strategy in-depth, and ensure that they are featuring prominently for online searches within their category, this is essential.

  • SEO is the single most important tool that brands can invest in over the coming years, if your products are not findable online, then they won’t find their way into people’s kitchens when AI is making decisions.
  • Product diversification and product ranges will also help products to be found by AI programs. These products will need to have an online presence as well, and they will fall into the ‘new product’ section and the niche product sections, gluten-free oats.
  • Brands already invest in product diversification, but these new products are nowhere to be found in online searches. For example, search ‘sugar free bread’ as a search term in Google and you will see that no brands are really owning this search term from an SEO perspective, the search returns more informative results rather than brand led messaging

Brands should ask themselves if their products are as prominent as possible online, and if not they should address that now. It would help with brand discoverability in the short-term, and in relation to long-term strategies in the face of AI.

Finally, identifying which online retailers are ahead of the curve in relation to smart homes and online shopping, will let brands know where they should be focusing their efforts.

Amazon, Walmart and Tesco have very strong presences in the online retail space, and it is worth establishing trade deals with those retailers.  These relationships are key when online retailers are making decisions in relation to the programmed AI bias, as a trade deal gives them a reason to consider your brand in their selection process.

In conclusion, the steps for brand to get themselves AI ready are:

Diversify your products: in line with consumer demand trends

Establish online presence: for each of your products, to ensure they are easily discoverable online

Ensure your products are positioned online within 1 of these 4 categories vs competitors: market leader, the cheapest, the most expensive or the newest products.

Build strong relationships with online retailers: Ensure your products are being featured online with key retailers and supermarkets, and establish trade deals with those retailers.

If brands address these steps now, they are preparing themselves long-term for the inevitable AI-programmed future.

Source: Institute for High Performance, Accenture (2016)

GroupMmmys Awards Night 2018: Padraig Ryan, receiving his awards from Bill Kilney, CEO GroupM.
The future will be about marrying ingenuity and ideas with data
Left brain, meet right brain