How predictive analytics takes away the guesswork from advert efficiency
Everyone loves a sure thing, especially someone who pays for an ad.
In a world where many business leaders waste money on ads dealing with their gut instincts or worse – gathering as much data as possible on the ad audience and inevitably targeting them with irrelevant ads – these are the best ways for performance marketers to To use predictive analytics to optimize ads?
Performance marketers first need to figure out why they are optimizing their ads in the first place.
Do you want to increase sales? Win customers? Reach another goal? In this seemingly simple step, many performance marketers make a mistake. They are collecting all the data they can instead of the data they should.
Imagine if Karen Heath at Teradata hadn't wanted to help a retailer grow diaper sales in 1992. She may never have looked for data to show that men bought beer when they bought the high-margin item. The retailer's sales increased by putting together beer and diapers.
That one insight for a retailer that led to product placement changes in 1992 eventually evolved into more advanced predictive analytics that streamlined multichannel marketing in 2011.
Now, in 2020, the practice is so advanced that the basic definition of predictive analytics is to include machine learning techniques.
Predictive analytics practitioner Helen Xiaoqin Yi, a data scientist at a major electronics retailer, suggested that using algorithms like SVM, logistic regression, or neural networks, performance marketers “use predictive tools to create audience segments or explore new potential audiences ".
"Then we can analyze their preferences based on comments, ratings, social media, interactions with ads, events or relevant campaigns, and come up with multiple plans for different segments," said Yi.
Stephen H. Yu, president and chief consultant at Willow Data Strategy, advised performance marketers to figure out who to target with which ad and through which channel before personalizing ads.
"A series of personas based on bias modeling can help identify the most optimal proposition and creativity for each goal," said Yu.
Devyani Sadh, CEO and Chief Data Officer of Data Square, provided three possible segments:
- Prospects: Identify high performing segments for potential ad audiences based on demographic and psychographic similarities, content preferences, and interests of known high-quality customer clones.
- Active Customers: Model your customers' history along with a “Similarity Index” of other customers with similar buying patterns to optimize ad content. Examples are cross-selling or the next logical product (at the same time or one after the other)
- Endangered or expired customers: In phase 1, those are identified who have been staged due to wear and tear or have already expired, but are likely to respond to an offer. Phase 2 optimizes ad messaging for retargeting or other initiatives by predicting special offers and promotions that are most likely to resonate with that group based on history.
Yi suggested starting a small test with several ad drafts at different times of the day, exposure times and placements to prove that the optimization worked.
Then give credit where credit is due. Yu added that performance marketers need to record how well each element and channel is performing. In other words, don't use blanket mapping by default.
For example, performance marketers want to keep track of more than just ad placement. In this one area alone, Sadh says, performance marketers “can optimize ad placement by evaluating high performing platforms, partners, social media sites, websites, search engines, and regions by extrapolating from navigation patterns, search, browsing behavior, and digital identities of known converters . ”
But even a sure thing won't be a sure thing forever. Just as search engines update algorithms on a regular basis, performance marketers need to rethink the predictive analytics process to further optimize their ads. Sometimes marketers have to start over.