How Shutterfly Used A Search-Powered Custom Algorithm To Zoom In On New Customers

Everyone in programming knowledge is that the best way to achieve the results is the use of a kind of identifier … right?

But the Shutterfly CTV campaign launched at the end of last year to promote her holiday gift and photo offers of this challenge and perception.

The goal of Shutterfly was to discover new basins of potential customers. The Rain Agency for Growth, his agency partner, suggested a dedication to part of his budget for testing a custom algorithm that relies on search data rather than deterministic or audience signals based on ID, to find the geographical region with the most appearance.

Although the rain never tested an adapted searches based on the searches, it made sense as a market to identify the market where there was an interest in the category of products, but low awareness of a particular brand, said David Nyurenberg, Rain’s Director of Development and Innovation of Video Product.

In other words, geos in which many people generally seek digital photo services, but where the Shutterfly brand awareness behind the competitors.

Searching for the incremental audience

Rain received a search data by the TV Congratulations to EDO, which can relate to household examinations in the geographical markets using the IP address and match the device ID.

Chalice AI has created a custom algorithm for swallowing these data from Edo -ai to determine which geos shutterfly should target and when.

“We lay Algo on top of the bundle of different CTV PMPs – all top things like Hulu, Peacock, Spectrum, Paramount, Samsung and the deadline,” Nyurenberg said. “And then Algo is only flexible on all those PMPs in the markets that the media seemed to be needed at a certain moment.”

The rain tested a custom algorithm based on search of other targeting and optimization tactics, including the targeting of keywords, diverting and modeling of the loocalics, as well as the segments of the third party of the audience provided by the contextual supplier and automatic content recognition (ACR). Rain refused to name these suppliers on the record.

The custom algorithm surpassed the visit of a web site visit, set up an account and buying of a photo -packet of new customers, based on the measurement of the Pot.tv.

From October to December, between 17% and 19% of conversions that can be attributed to the adapted algorithm came from new customers. Most of the other tactics hovered in the range of 10% to 15%. Retargeting was the least effective method to achieve new customers, which made up only 8% of the audience.

According to Nyurenberg’s opinion, these results prove that the strategies of addressable targets are not the best way for advertisers to reach an incremental audience. And this is because most of the tactics are addressed – especially redirect – it relies on the deterministic signals of the audience that can only be collected after the buyer has already entered the purchase cycle.

“The signals based on the intention and the third party audience show that these people are already on the way of conversion,” Nyurenberg said, “so how much difference will the media make you?”

In fact, they would probably buy even without seeing the target ads, he added.

But the search data Edo inserted into the custom algorithm correspond to people at the beginning of their purchase cycle, closer to consideration than conversion. Search data also showed that the audience was not already familiar with Mark Shutterfly, so there was a real value in aiming at them with CTV ads to raise the brand awareness, Nyurenberg said.

The custom algorithm has also launched better new customers and a better cost per new customer than other tactics that can be addressed, which is to be expected, Nyurenberg said.

The search algorithm “had the highest new percentage of customers, and he had the lowest CPM, which makes sense,” he said. “Do not agree on addressable data, which is very competitive, so you will get the most cost -effective consumption.”

Constant optimization

During the three-month Shutterfly campaign, the adapted algorithm would perform weekly optimization, so that after the consciousness and engagement of the brand were sufficiently raised in a particular market, it could divert advertising consumption to new markets.

As a result, new customers and costs at the new customer have become constantly more effective as the campaign has progressed. New Roas customers rose from $ 0.31 in October to $ 1,49 until December, and the cost of a new customer dropped from $ 243.26 in October to $ 57.31 until December.

Shutterfly also finally aimed at certain fertile regions that would not otherwise be, Nyurenberg said.

“Usually programming algos will assign consumption of the same cities with high access, such as New York, Los Angeles, Miami,” he said. But Chalic’s algorithm has disproportionately assigned to consumption on smaller areas of the metro, such as Honolulu, Reno, Santa Barbara, Utica, Palm Springs, Lexington and Bangor.

“This is not a typical DMA mix you would expect to see in the program campaign,” Nyurenberg said. “Algo really entered those who filled the search criteria.”

Access to a custom algorithm based on the search was so well that Shutterfly held it active after the end of the holiday campaign season, Nyurenberg added.

And the rain has now begun to set up a strategy to other clients who want to encourage new customers. For Nyurenberg, the successful study of the case was to show the sustainability of targeted tactics that are not based on the addressable audience.

“This campaign has proven that the modeled tactic surpasses the goal-based target,” he said. “It feeds on a permanent narrative that the target, since forward, will largely be ai-model, not based on the ID.”