Using Data for Real-time ‘Gut Checks’
The speed with which we can identify buyer preferences and purchases is dramatically changing how consumer packaged goods marketers use data. It’s no longer a look back at past performance to predict future business decisions. Instead, CPGs can refresh and adjust strategies in real time, informing everything from a brand’s flavor portfolio to which imagery best drives sales.
By shifting from modeling to an artificial intelligence- and data-driven approach, and by linking digital media consumption to in-store data feeds, brands can innovate faster, alternate media and eventually predict category and shopper trends. This approach gives them the ability to more confidently identify which audiences, channels and creative — from banners to TV ads — move products off the shelf.
Many measurement solutions are highly modeled and not regularly updated. Data-driven, real-time deterministic solutions can uncover results at a granular level that a modeled solution cannot.
Additionally, most targeting solutions are based on demographics that are too broad and not grounded in a shopper’s personality or product preferences. We instinctively know both men and women can want gluten-free products and shoppers can be price-conscious at 13, 30 or 60 years old. But machine learning and data will now make it easier to “see” a buyer and engage more appropriately.
We’re not suggesting marketers abandon traditional sales lift measurement, which is typically available six to 12 weeks after a campaign. That still plays a role, particularly in determining return on ad spending. But be sensitive to its limitations and use tools that allow for mid-campaign optimization. According to a Simulmedia analysis, at least half of the audience impressions in a typical national TV campaign are wasted, reaching either the wrong targets or ones that have already been exposed to messaging the optimal number of times.
With real-time buyer data, marketers can cut budget waste and reinvest during a campaign, driving conversion in-store, lowering customer acquisition costs, and fine-tuning frequency and the creative mix.
Closing the loop
It’s easier for pure-play e-commerce sites to see what drives a purchase. But many CPGs, whose e-commerce sales are a small fraction of total sales, can’t close that loop as easily because they haven’t found a way to easily link channels and impressions to in-store sales. Using a digital-to-in-store identity graph, however, brands can see if a buyer has already tried a new product, so they can shift those marketing dollars and optimize campaigns on the run.
Customer acquisition cost is about understanding more than CPM. For example, branded video has a higher CPM than a digital banner ad. But if the video content prompts an in-store conversion rate that’s eight times higher than the banner ad, it’s also far more efficient in what matters: driving a buyer to the brand.
Real-time buyer data also cuts budget waste by zeroing in on ad fraud from bot-driven impressions or clicks. Marketers now can know whether a channel is actually delivering those 100 million impressions to the real eyeballs of shoppers they want to reach. Ad fraud is a big problem that isn’t going away: It’s expected to account for up to $44 billion globally by 2020, according to Oracle.
Fine-tuning frequency and creative mix
Right now, a Super Bowl spot is deemed a success if it makes the “best ads” round-up in USA Today. With connected TVs in 40% of households and growing, we’ll soon be able to serve up ads and track their immediate impact in conjunction with what a shopper sees on her smartphone, desktop, and in-store.
These real-time solutions are allowing CPGs to identify brand loyalty segments that can help to more definitively answer the frequency question. A loyalist, for example, is likely more ready to watch 30 minutes of branded content and doesn’t need the steady pulse of messaging that samplers and brand shifters do. Layering in-store purchase behavior into the mix can help uncover which creative message is converting what shopper on which channel.
For example, thanks to sophisticated technology advances, one client linked 30-plus UPCs to their ads and discovered that only five drove more than 60% of new buyers — so they switched up the creative to feature those. Another learned over the holidays that an overwhelming portion of its buyers were new to the category, informing future audience segmentation for the upcoming season. A third found out that buyers from a particular campaign over-indexed on organic and earth-friendly items, which informed its creative execution and potential co-op partners.
Ultimately, machine learning and data intelligence will produce a library of results to help predict future buyer behavior on everything from pizza to gum. We’ll soon know what shoppers will buy and might try down every aisle. We’ll also know that clicks do not predict in-store sales — but that’s a topic for another day.
About the Author
Amy Fitzgerald is group vice president of multi-touch attribution at Catalina Marketing.She landed at Catalina in mid-2016, and has been a leading force in building new capabilities that link in-store purchases to digital media consumption (also known as multi-touch attribution). Her career started as a brand manager in the Minute Maid division of the Coca-Cola Company and has included stints at PepsiCo and LeapFrog, as well as at start-ups including Yevo and EverAfter.com.