The LinkedIn Algorithm: How it Works
By Joshua BoydDec 13th
Published August 10th 2016
Consumer market insights is a hot topic for those of us in the social intelligence space right now.
At Brandwatch we are always keen to understand more about how our clients and friends are using social data, and we were lucky enough to steal some time with an expert on the subject of market research and consumer insights.
Misia Tramp is VP of Customer Experience Strategy and Insights at global digital marketing agency Metia.
The department that Misia heads up is responsible for making sure that, as an agency, everything they do for their clients is data-driven, and “helping clients take the customer-360 approach.”
Misia explained how Metia has been working with a full range of research methods from traditional focus groups, ethnographic data, and the quantitative all the way through to big data, big language and visual analytics. In this exclusive interview, she explains how.
Misia: A whole range, actually. Brandwatch is our core quantitative tool, if you like, which we use as the center of what we refer to as our ‘insight architecture.’
Brandwatch is obviously centered on a data collection method that’s keyword centered. We also use visual data that collects the visual language around certain topics, brands, themes, et cetera – literally pictures with descriptions around why those images represent a certain theme or idea.
We also integrate with web analytics, so we look to understand the social journey around a particular theme, topic or brand and then we’ll actually analyze the web analytics to understand how behaviorally that marries what we’re seeing in social.
We also use search. There’s a big trend towards rather than thinking about keywords authority to thinking more about topical authority, and so tools like Brandwatch help us to think about things as topics, not words.
The next one is CRM data. Some of the work that we’re doing is tying social handles to email address so that we’re able to append CRM data to social data to understand how those two line up.
And then, finally, is the traditional research data set.
In addition, we’re also tying things like surveys and focus groups to social profiles – that’s simply done by asking people for a social handle so we can tie a specific survey response or focus group to what they’re talking about in social, we can then look at the relationship between the things that we’re asking them about those are just the things that they care about.
There’s a lot, basically. In the paradigm of looking at the customer in 360 degrees we have to reflect that in the data set. We can’t be biased towards one or the other. We’re integrating with anything we can, whenever we can.
Misia: I think the primary thing is how contextual it is. With any primary research you’re generating the data and so there’s always a risk that the research design can impact the data you get. Even in a perfect research design, it’s still hard to make primary research super contextual, even with ethnography. You’re following somebody around – you’re by definition impacting the purity and the reality of the data.
There’s an integrity and an honesty and authenticity to social data that’s very hard to replicate anywhere else, so it’s done in the customer context, in their time, in an authentic manner. We’ve done studies about this that have shown unequivocally that social data is a much less biased data set than many primary research sources, and there’s a lot of reasons behind that.
As researchers we’re always trying to understand what drives customer behavior, right? So if your data set is authentic and it has integrity it’s going to be a better predictor of behavior.
The word context is the key here. The other thing that clients often lack is the context for their brands and the problems they’re solving.
It’s very easy as a brand to assume that you’re the most important thing to somebody, and research data exacerbates that because you’re continually going out and you’re asking questions about yourself.
It’s very easy to be surrounded by reports and data that relate to your brand, and even though people are getting better at being good about supplementing with ethnographic data, I think for a large part research data is by definition a little bit self-obsessed. And in the social data set you just get that context.
You understand where brands actually fit into someone’s life, because you don’t just get data about how they feel about the brand, you can get data about how they feel about anything.
In a survey as bank I might go and ask the customer all sorts of questions about how they feel about their current account, their credit cards, their mortgage. In reality, these customers don’t really care about the interest rate on their credit card or how it looks and feels. Not really. In the list of things they care about on any given day, the credit card isn’t top of mind, whereas money is.
When you think about what you get through social data, you can really identify what the higher order problem as an organization is that needs solving.
Should we be talking about credit cards, or should we be helping our customers take the mental energy out of money management?
Social data gives you the context for your product and brand to make sure you’re talking about the right things and helping to focus your content and your experiences around the problems that customers want you to solve, versus having a very product or even brand-centered discussion.
Misia: It depends on what you’re trying to do.
So if I’m an organization who sells combine harvesters to farmers, and I use that example because I did actually do work for a brand like that, or if I’m a brand that is trying to sell hair care or hair dye products to greying men, the chances of those people being online talking about combine harvesters or their need for hair dye or whatever it’s called is probably quite slim, right? And that’s true I would say of any category, frankly.
There are things that people just won’t rally around online unless they’re really closed forums. So if I’m a combine harvester brand and I’m trying to go online to find out how my brand is perceived, you’re probably not going to get a lot of traffic.
However, I have not yet encountered a scenario in which we can’t find data that relates to a higher order problem you’re trying to solve.
Less than 5% of social conversations mention a product or a brand at all. So if your keyword centered research is only based around products or brands, you could be missing up to 95% of what matters.
We always urge clients to think – what problem are you solving? Then let’s harness conversations around that. I’ve not encountered a scenario in B2B or B2C in which we’ve not been able to help our clients using that approach. In fact, we do it all the time.
Misia: I think the biggest trend is the continued desire to integrate and automate data in real time.
I think one of the challenges with that is whether or not people can actually act on stuff in real time.
As our ability to measure stuff increases, there is a desire to make data-driven decisions and people perceive that that begins with measuring everything. We are getting better at understanding that to get a 360 degree view of the customer, that requires us to not silo data efforts.
In my experience is people are getting better at understanding that research shouldn’t be siloed away from social monitoring, which should not be siloed away from web analytics which should not be siloed away from CRM. Integration is key.
Real time is key.
And the other trend I see is that for some of those traditional research methods that we all made so much money out of in the early days, tracking studies, even to a certain extent focus groups, I think more and more clients have created in-house agencies that do a lot of the work themselves. Even if it’s not all of the work, a good proportion of the work seems to be moving in-house as organizations are becoming more data driven.
I’ve been working in social data since the industry started, and whereas people were very suspicious of it – there’s still an element of that, but we’re now in a place that people, even if they’re suspicious, they accept they can’t ignore it – there’s an understanding that we need to understand how to think about words as data. How to think about pictures as data.
Structuring a wealth of stuff that isn’t necessarily in a questionnaire is something that I’m seeing people being much more accepting of.
Finally, I think we’re going to see more sophisticated analytics and metrics emerging from the social data set, whether it’s sort of a net promotor score-style metric or something similar.
I think that where we’re going with social tools and social data, the natural next big evolution is going to be to be more analytics and KPI focused.
A big thanks to Misia for speaking with us. This interview is one in a series with industry experts – you can expect more every week.