Marketing Science has its own line of business at Critical Mass. It was part of our insight and planning group. Actually before it was called insight and planning, it was called research and analytics — a long time ago. The company has a deep foundation in analytics.
When I came in, we were getting a little bit dinged in the Forrester Wave for our measurement capabilities. I and another guy were charged with figuring out what the problem was. We also had a high churn rate of qualitative analysts. We spent about six months researching, building a business plan, and ultimately came up with the service offering marketing science. The term comes from academia. There is a whole marketing science field in academia, and a sector of people who refer to themselves as marketing scientists, and they actually are more scientists than they are marketers. They’ll do about 10% of marketing and 90% of science, and we’re kind of the reverse. We do a lot more marketing and we have just a touch of science, depending on the client, although that’s starting to change.
What’s the team profile?
Our team grew over time to around 30 people. The type of work Critical Mass does is often complex big builds or complex multichannel campaigns. We have a mix of talent that is about 50% manager and above. Our team is not a bunch of junior people with a couple directors.
We aren’t doing reams and reams of research optimization that would dictate having a lot of junior staff or outsourced staff. We’re not doing reams of reporting that would dictate having lots of people offshore. There’s a pretty good mix.
I would put my team up against any other agency team in a competition in a heartbeat. I have a lot of confidence in them. They aren’t just good quantitative analysts, they are good communicators. We think of ourselves as the quant whisperers, so when you have a lot of heavy-duty Ph.D.’s and statisticians who don’t consider themselves marketers, they’re sort of that half step in between the business team or the marketing team. I like to think of us as a liaison between the parties.
What’s a client use case?
Nissan is a good example. We have about 18 analytics people working on different business units of Nissan across the globe. We do a lot of website implementation, strategy, optimization testing. Because Nissan doesn’t have an internal team, we facilitate all the analytics for them.
And we’re the center point for all the data collection between all the agencies. There are 25 other analytics people across other agencies, so we’ve got our media team, search team, lead-management team, eCRM team, and we bring everybody together. We have an annual analytics summit, usually at TBWA, that we host.
We bring all of the analytics parties together and then we coordinate standardized KPIs and make sure all the data is coming out the same way. The website team often can slide into that role, even when there isn’t cross-channel accountability on the client side — [especially] if everybody is driving to the site. That’s the case with automotive.
A different structure would be like on CPG side. With Clorox and Pampers we brought all of the channel data together into a sort of centralized score card, and looked at marketing optimization rather than on-site testing and optimization. That’s a higher-level question: “How are you spending your money?” Not necessarily hard-core attributions or multisource attribution, but light touch — because CPG generally doesn’t invest the money in Web analytics to do that stuff.
Will the work Facebook is doing with Datalogix speed up the CPG investment in measurement?
It’s the first foray. The fact that you can take your loyalty program from your Kroger card, feed your buying behavior into Datalogix and then re-serve relevant advertising on Facebook based on the Oreos you bought last week — I think CPGs will eat that up because it’s easier to connect the dots. The barrier in CPG all along has been that they can’t prove how digital is impacting the retail sales.
I’m still not hearing that they’re giving up marketing mix modeling yet.
Where is the center of gravity in the world of attribution and marketing mix modeling right now?
A lot of groups will tell you marketing-mix modeling doesn’t make sense for digital. I spend a lot of time at the Wharton School of Business — we have a partnership with them — and even Wharton is moving on to newer mix modeling techniques such as multi-source attribution. The whole premise is that all the channels are independent and they don’t interact with each other.
Modeling only happens once a year, and it’s based on your six-month prior data. Then you spend six months modeling it, and then you divvy up our money for the whole year. It’s really slow. Multisource attribution can happen at the end of every month. You can change your levers, which is what makes it so much more powerful and relevant.
How should digital agencies position to make the best use of clients’ first-party data?
There are a few schools of thought. Some people feel that the analytics people should be sitting with the development teams and the creative teams to do that kind of stuff. Others feel that they should be on the client side because they’re more neutral. And other people feel they should be an independent party altogether because the client-side people generally don’t know how to staff correctly.
The agency provides a spectacular service to the client, but it doesn’t always have access to the consumer data. Or it might have access to only one kind of consumer data, but not all of it. There may be always a mix of internal analytics and agency analytics. Third-party analytics I think are the least valuable.
It’s kind of complex. It comes down to the client’s fear of you making their own data look good. If you have open, honest discussions and you first engage your client about integrity and that you will never make the data look rosy, that’s usually what we do.