Proof Analytics Blog

How eBay Classifieds found their marketing model - Proof Analytics

Written by Niklas Karlsson | Feb 15, 2021 9:32:00 AM
 

Chasing the full customer journey

After studying political science and journalism (she still moonlights as a freelance writer) Iskra Velichkova left her native Bulgaria for Berlin to pursue a career in marketing. She then proceeded to work at all “the” agencies – DDB, BBDO, and Ogilvy. At first, this struck me as a surprising background for someone who speaks fluent MMM. Then she told me her story.

“By the time a client gets to the creative agency, the strategy has been defined. So have the marketing and business objectives. You work on the communication part of it, and deliver the assets. Then someone else takes over and spreads them around the world. At some point, for me, it just wasn’t enough.”

Iskra pauses. Imagine her, four years ago, the marketing journey spread out in front of her like a Eurorail line, staring at a small segment near the end. “I wanted full access to the whole journey,” she says decisively. 

“I wanted full access to the whole journey.”

“Also, I wanted to really understand the business objectives, define the communication objectives, and then build a strategy around them. To be able to see all the information and the learnings. Then implement the strategy from the beginning.” 

Someone at eBay Kleinanzeigen must have been reading her mind, since they offered her a job. Today she is their Head of Brand and Content Marketing.

“It was a perfect fit,” she nods. “Because I’m a heavy user of the product, I’m convinced of the value of what we’re doing.” 

Soon she took the task to implement the first local MMM for eBay Kleinanzeigen. A couple of years later, she took part in the implementation of a global MMM solution to combine six businesses across the globe.

 

Create a model. Learn. Adjust. Repeat.

Today, Iskra says, they work with data on a weekly basis, and refresh their model every three months. But it took almost two years from the time they set up the first pilot until they had a full solution in place. 

“There’s a lot of complexity to consider that didn’t exist ten years ago – like the exponential growth of media channels. It’s a fragmented world out there, and it’s becoming ever harder to get people’s attention. It’s also getting harder to grasp the consumer journey because it is no longer linear. Online channels, and especially performance media, makes it easy to show conversions, but it is very difficult to prove conversions from my spend in, say, TV or out-of-home, or online branding channels like video. The way people arrive at a decision is complex, and that has made it even more difficult for marketers to explain how our efforts actually pay into the business.”

Because eBay Kleinanzeigen is a C2C business, Iskra says that makes it even more complex.

“Imagine the customer journey of someone buying trousers and someone buying real estate: Totally different experiences! As you can imagine, these categories work in completely different ways.”

Iskra recalls that when they created their first local model, they concentrated on the upper funnel. “We left out a lot of the lower funnel, just to be able to see the effect of the upper funnel on the business KPIs that we were measuring with very simple attribution in the performance field,” she explains.

The results were not extremely robust, but they got a first quantitative measurement of upper funnel media spend on business KPIs. They could also see how their KPIs were sensitive to media at different times of the year, even as granular as long vs. short holidays. And they were able to see halo effects between different categories. But the model had some shortcomings due to the limited amount of data that went into it.

A couple of years later, a global project started with a much more holistic view on MMM and Iskra was happy to join. “We didn’t just want to know how things were working at eBay Kleinanzeigen. We wanted to do benchmarking across six businesses in five countries using common business KPIs, and model our marketing efforts against them.”

 

Advice for new model-makers

Iskra’s big advice for anyone looking into marketing mix modeling is to pay special attention to the factors that influence your business. “You need to have an extensive list of media channels, but you also need to get your control variables right. We have included many things that are inherent to our platforms, but are not really marketing spends, like all the pushes from your platform, automated email, and the user interaction with that. And as in every MMM, you need to define all the external factors that influence your business. You need to have those in your model, so you don’t overstate or understate your marketing efforts.”

Pay special attention to the factors that influence your business.”

“Obviously, you need to be very careful in defining your outcome KPIs,” she adds. “But then you get to the most complex part, namely, getting the data.”

“My advice is not to compromise here. If you have a list of 30 things that influence your business, and discover that you probably won’t get all the data for four or five of them, you should start working on sourcing that data and refactoring it and making it meaningful and available. Don’t just skip it because it isn’t easy to get three years of data history. Sure, you could make do with less, but then things become a little bit blurry, right? Spend the time to get it right.”

 

The best part of MMM 

You need to make a clear map of the KPIs you want to measure and how they influence each other which is another difficult task depending on your business. But once you have gone through all the effort, you can profit from an array of possibilities that MMM delivers:  

“First, you get to see how media channels weigh up against each other, and which ones are most influencing your KPIs.Second, you can analyze past performance. Third, you can make predictions for the future. Then, you can benchmark individual markets against each other.” For the eBay Classifieds Group, this has enabled them to determine how to distribute resources across countries in the most optimal way. “For eBay Kleinanzeigen, I can see what my most efficient channel is. And which channel has reached the point of diminishing return. I can tell where to take money away, and which channel needs more attention. And, I can benchmark against all other countries.”

The way she says these last words, I can tell that Iskra is in her element. Just as she hoped, marketing mix modeling has enabled her to see far beyond the customer journey, into the past, present, and future – and she’s using all this information to her advantage.

How eBay Classifieds found their marketing model: Written by, Maziar Nodehi, Proof Analytics