How Retail Media are Adapting to the Digital Age
Global consumers are cheating on brands and flirting with new ones; a recent study by Nielsen suggests. Only 8% of consumers are loyal to the brands they purchase and as a result, the majority of purchasing decisions are made at or near the point of sale, where the best way to influence consumers is via retail media.
Retail media has historically included tools such as in-store advertising and placement, couponing, discounting or promotion in weekly circulars. The introduction of loyalty programs, and later, online shopping, has made it possible to target consumers directly and individually via email, app, website, banner ads, social media or check-in kiosks at the entrance to the stores.
However, research shows that not all CPG brands are satisfied with the way retail media work today, with only 34% planning to increase spending in this channel. That’s quite a contrast to the 82% that plan to increase investment in their own digital channels such as social media, search, online video or mobile.
For the majority of CPG brands, the main issue is the low ROI (or ROAS) of retail media campaigns. According to research by Nielsen, an alarming 59% of these campaigns globally don’t break even. Brands also report low confidence in their ability to actually measure the ROI of retail media, with only 56% saying they are somewhat confident. That is significantly less than in the case of digital channels (74%) and even traditional media (59%), such as TV, radio and OOH.
So what can retailers and their partners do to increase the efficiency of retail media and keep CPG brands interested? In short — target shoppers better and smarter.
Instead of unleashing on consumers a deluge of weekly offers and hoping for the best, retailers can target shoppers individually with a selection of promotional offers that are highly relevant to them.
How targeting improves ROI
To demonstrate the effectiveness of better and smarter targeting, we tested SO1’s AI based solution called Programmatic Promotions with several major German FMCG retailers. The test compared the ROI of AI-targeted campaigns against traditional channels — leaflets and coupons.
The results were striking. Depending on the category, the AI targeting system delivered between 143% and 312% ROI for brands after the deduction of fees. Mass couponing and circular offers, on the other hand, ended up mostly in the red or close to zero.
How is it possible to improve performance so radically? There are three main metrics essential to the ROI of brand promotions:
- Redemption rates — the average conversion rate of campaigns
- Incremental sales — the revenue and profit uplift created by campaigns
- Advertising costs — the price of campaigns
Let’s look at each separately:
Personalized product recommendations increase redemption rates
Many retailers are already trying to personalize their loyalty programs. But with the exception of a few big names such as Amazon, Walmart and Sephora, very few actually have an advanced AI targeting system in place.
A machine learning approach can use the available shopping history to analyze millions of shopping baskets and recognize complex market structures, consumers’ preferences and behavioral patterns and common reactions to external stimuli such as discount and promotion (including their form, time and placement).
After this learning process, the algorithms are ready to predict customers’ behavior with high probability, based on their shopping history. And all of that on an individual 1:1 level — the true Segment of One.
To determine the effect of personalization on redemption rates, we conducted a massive split test in two regions with one of the biggest German grocery retailer. Region A worked the old way — a team of experts handpicked the best weekly offers delivered to customers via digital check-in kiosk at the entrance to the store. Region B personalized these coupons using AI. The second approach delivered on average 9 times higher redemption rates.
Consideration of business goals improves incremental sales
Although increasing redemption rates might seem like a win, to drive ROI we actually need to achieve incremental revenues — to increase the average basket size of a customer. Than we also need profitability — a big discount on a popular item might increase revenues, but the margin takes a hit — and so does ROI.
That is why the targeting system should consider not just relevance, but also price and margin when composing a list of product recommendations for a certain customer. This might favor less relevant products if they are more expensive or have a higher margin. Based on these combined factors, a so-called Rank Score is calculated.
Using the Rank Score, the system ranks hundreds or even thousands of products available for promotion – for each customer, individually. Only a few of the top-rated products are shown to maximize relevance and performance. Products that are predicted to be purchased without any external stimuli are automatically excluded.
As a bonus, the system can also customize price. For instance one shopper may need -10% to convert, another -25%, while yet another just needs a nudge to remember this product exists to buy it. By optimizing the discounts to account for an individual customer’s willingness to pay, the system always offers the lowest amount of discount possible, thus maximizing the margins.
All this adds up to impressive results. Tested with five different retailers, customers targeted by AI-based product recommendations spent 15.8% more than the control groups. And personalized pricing meant that discounts offered were 36.4% lower than with mass discount offers. This is not only a huge increase in sales and revenues, but also in profits.
Automated bidding auctions lower advertising costs
These results are great you might say, but what about the fees?
To manage pricing and reward brands that offer products that are relevant to customers, SO1’s Programmatic Promotions uses an automated bidding auction similar to that used by Google Ads (formerly Adwords).
To increase conversion rates, the system always strives to serve customers the most relevant offers only. The higher the relevancy, the lower the fee and hence the better the ROI that can be achieved. Brands are happy because they pay less, retailers because customers spend more, and customers get their favorite products at a discount.
So what if a brand wants to introduce a new product or expand its customer base? It might not be relevant for a large number of consumers yet, but that might change after a successful acquisition campaign.
To this end, a brand can move itself up in the ranking simply by outbidding the competition in an auction. The bidding system ensures that the extra investment is directly dependent on how relevant the product is to a specific customer.
As already mentioned, the system calculates this on demand for each customer individually. Here’s what a bidding auction might look like for an individual pull request:
CPG brands can adjust their bidding strategy in real-time for each product according to their individual goals — whether they want to introduce a new product, target new segments, fight off an aggressive competitor or just strengthen the long-term loyalty of existing customers.
On top of the performance uplift, SO1’s Programmatic Promotions targeting platform gives CPG brands as well as retailers a real-time overview and control over the performance of their campaigns, meaning no more uncertainty over the ROI or ROAS.
If all of this sounds too futuristic, keep in mind that this advertising platform based on the SO1 Engine AI has already been successfully tested and implemented at several large retailers in Europe and the U.S. We are currently scaling the system. If you’re interested in knowing more or have any questions, please contact us.