Optimove: Orchestration Is About Making Sense Of Data, Not Just Stitching It Together
This is the 10th in AdExchanger’s “Meet the CDPs” series. Read previous interviews with mParticle, Acquia-owned AgilOne, Amperity, Segment, ActionIQ, Lytics, Bluecore, Microsoft and Tealium.
Brands that sell commodities don’t differentiate on their products, they differentiate on data.
“And they differentiate on relationships,” said Pini Yakuel, CEO and founder of Tel Aviv-based customer data platform Optimove.
Which is why they need a way to personalize their messaging at scale.
“If you’re a luxury brand, it’s all about your product – customers can’t get Ralph Lauren anywhere else,” Yakuel said. “But you can get a printer anywhere, so a company like Staples needs to build a relationship through very frequent engagement with customers.”
Roughly 30% of Optimove’s 250 employees are engineers, and the company has around 350 brand clients, all of them B2C, including Staples, Dollar Shave Club and Stitch Fix. There’s not enough emphasis on the different needs of B2C brands vs. brands in the B2B space, Yakuel said.
“They’re completely different – different problems to solve, different challenges,” he said. “Marketo is a horrible fit for B2C, and we’d be a horrible fit for a B2B company.”
Yakuel spoke with AdExchanger.
AdExchanger: Was Optimove a CDP from Day One?
PINI YAKUEL: Well, we were covered by David Raab back in 2013, so you could say we are part of the OG. Our goal was always to marry customer data into one place to build a predictive model about customers and then be able to act on those insights for smarter campaigns and personalized messages. But the space has definitely changed.
It’s become confusing to analysts, reporters, buyers and even me to some degree what the difference is between what I call a data or infrastructural CDP and an orchestration or activation CDP.
Data CDPs are one layer above a data warehouse. It’s hard to get data in and out of a data warehouse. The purpose of a data CDP is to make that easier. When data CDPs talk about orchestration, they mean putting data inside Salesforce for example, and then Salesforce does the orchestration. But they don’t govern all of the communications.
We’re an orchestration CDP. We need a strong data foundation, but we use it in order to make smart decisions in real time about what messages will engage the end consumer. It’s about making sense from data and getting actionable insights.
What are some examples of data CDPs?
I’d say Amperity, Treasure Data, ActionIQ – these are data CDPs. Simon Data is an example of a CDP that straddles data and orchestration.
Who are your biggest competitors and how do you differentiate?
Sometimes we compete with multichannel marketing hubs: Salesforce Marketing Cloud, Emarsys, Braze, Cheetah Digital. But we are the only one who does message orchestrations with ML and AI to prioritize customers in real time. We help brands get to hundreds of campaigns, messages and segments for different types of personas and customers. Ten or 20 segments is just not enough for real personalization.
How long does it take to onboard a new customer?
Typically, two to three months from the moment we sign, because we’re taking in all of their historical data, cleaning it, making sure it’s high quality, stitching it together from all sources and then building a predictive model for each of our customers.
What data sources do you most commonly connect?
Transactional data, including purchase history, ecommerce data, point of sale, catalog data, web and app data, pages visited, buttons clicked, articles read and behavioral data. We also bring in customer service data, including tickets from Zendesk. If a customer was upset last night, it’s probably not a good idea to send them a campaign in the morning.
And sometimes you can enrich the data with third-party systems, like Acxiom, but from my experience, this data is typically expensive, often not super accurate and I haven’t seen many cases where it’s brought a lot of ROI.
You acquired a real-time event streaming company called Axonite in February. What does it help you do?
Having a real-time decision layer that’s constantly thinking about the best message a customer should get at any given moment requires real-time context.
Axonite has a very sophisticated machine that looks at how customers move between different states. Our philosophy is to match batch processing of historical data with real-time data so that we have a hybrid approach.
What’s the most disjointed customer service/marketing experience you’ve gone through personally that could have been avoided if the brand had a better handle on its data?
This was about four years ago. I was new to NYC and living in Brooklyn. It was a hot day and I wanted to go to the beach. I ordered a Zipcar, went to the lot – and the car wasn’t there. I called, they apologized and told me to walk to another lot, but my car wasn’t there either. I had a reservation and there was a parking lot full of other Zipcars, but I could take any of them because they were reserved for other people.
Four calls with customer service and an hour later I go back home upset. The next day I got an email from Zipcar with the subject line: “Your first Ziptrip – the one you always remember.” They’d probably built a static journey into their system that said, the day after the first ride, send this email.
Well, they were right, I definitely remember it.
This interview has been edited and condensed.