How To Use Social Data - And How Not To!

We work with a lot of different types of marketers at Forrester, and we always customize the recommendations we deliver to different clients based upon their unique situations and needs. But over the past few years there's one piece of advice I've found myself giving nearly every company I work with: "Hire a listening vendor."

I love listening platforms and the social data they create; it's a powerful source of information that, used correctly, can make marketers and their programs more effective. But not enough marketers are taking advantage of these benefits.

No matter what type of company you work for -- indeed, whether you work directly with social media or not -- you should be using social data right now to:

  1. Develop your messaging. If you want to create messages that resonate with your audience, you need to know what they care about. Many of our past Forrester Groundswell Award winners have used private listening communities to craft their marketing messages; increasingly, we're seeing companies use data from public social media to guide their messaging as well.
  2. Source your creative. We know that consumers trust what they hear from other consumers more than any other source of information -- why not use listening platforms to identify positive social content that can be included in campaign creative? I've even seen a UK bank, First Direct, use social sentiment data in an outdoor advertising campaign.
  3. Improve your media plan. You probably already have a few staples in your online media plan -- the sites and networks that consistently perform for you. But social data can help you find new sites to add to your buy. For instance, Microsoft was surprised to find people talking about its computers in forums dedicated to fishing and cars -- but it quickly added those sites to its plan.
  4. Identify your key influencers. In Empowered, Josh Bernoff and Ted Schadler pointed out that consumers in the US created more than 500 billion peer-to-peer impressions about brands and products. Social data can help you identify (and then reach out to) the most vocal and influential of those consumers -- either one-by-one or by finding the forums where the most influence is taking place around your brand or category.
  5. React to your consumers. You can't fuel a positive conversation about your products (or get involved in a negative one) unless you find those conversations first. Listening platforms can help you quickly find both the good and the bad so that you're in a position to react.

But there’s one use of social data that drives me crazy: Increasingly, I’m seeing marketers and agencies turn to this data to measure the brand impact of their social and interactive marketing programs. (In fact, more than 80% of listening vendor customers say they use social data to track their brands.)

I know it’s tempting. If you’re already making the investment in a listening tool, you want to stretch the value you’re getting. And given that brand surveys are expensive and time-consuming, and that the listening tools all offer sentiment analysis, it must be a good idea to use listening to measure brand impact -- right?

Wrong -- for a number of reasons. First, because social data by definition only measures what social content creators are saying -- and that's almost always a small and biased sample of your target audience. Second, because social data is based on unstructured and non-comparable inputs: You rarely know for sure which of your products or marketing campaigns social comments refer to, nor do you know on what scale people are expressing their opinions. And finally, because most vendors' sentiment analysis simply isn't very good: Even the most optimistic listening platforms admit that their sentiment tools too often misclassify comments. The bottom line is that using social data to measure your brand simply doesn't make sense.

What do you think? Do you use social data in the ways I mentioned above -- or for some other purpose? Do you think that social data actually is a good way to measure brand impact? Have your say in the comments below.


Leveraging Social Data

I value this post and plan on sharing with a number of my clients and contacts - thanks Nate!

Your 5 points around how to use social data are correct. I believe point #3 around improving your "media plan" is an important one for businesses to pay attention to. If using the right mix of tools (traditional, social and web analytics tools) for example, lets say you launch a new product and a NY Times story comes out. As a result you see how that impacted conversations (comments, posts, tweets, videos, etc.) online and resulted in direct website traffic and leads, you might put more emphasis on media relations for your next product launch (vs. advertising or events). I believe if analyzed properly this can make a huge impact on your KPIs such as spending less, making more and keeping customers happy.

Also, I have to cautiously challenge your concluding statement that social media can't be correlated to your brand or overall consumer engagement. While "content creators" are often found and participating on the social networks and blogs, more and more consumers are posting comments on news stories, message boards, etc. I believe social/online measurement has a long way to go, but I do feel that as we dive into the multitude of ways consumers are engaging online, we will have some very powerful data at our fingertips.

Again, a very smart post - thanks for starting the conversation!

The social creator audience is too narrow

Thanks for your feedback, Matt. But until marketers can prove that things like social sentiment data are a statistically accurate proxy for the results they get back in brand impact surveys, that social data simply cannot be relied upon to measure brand impact. And in all my interviews on the subject I've only found one brand that has been able to prove that proxy -- and many more who've found there was not a direct correlation. It's that second fact (that many marketers find there isn't a statistically valid link between the two sets of data) that worries me, and tells me social data may not ever become a broadly reliable proxy for brand impact.

You also mention engagement, but you may notice I don't talk about engagement in my post. There's a reason for that: it's also an unproven proxy. Yes, users engaging with your brand may lead to them becoming more loyal, and it may lead to other users become more aware -- but ultimately it's just a stepping stone towards those real business objectives of loyalty and awareness (and ultimately sales). Celebrating high engagement is rather like celebrating the fact you've placed a big media buy -- it gives you a good chance of succeeding with your marketing, but on its own it cannot be used as a metric of marketing success.

Quality of Data

Thanks for your insight. I will admit ahead of time that I am an insider on this issue. We have used the data in many ways. The buckets you listed are general and a good place to start.

I would like to comment on the quality of data as many are unaware of the technology gap in the market right now.

Being from Forrester, you understand good data. The vast majority and I mean 99% of the vendors are RSS feed aggregators. This enables them to collect from what represents about 30% of the digital universe (and shrinking). In addition, you are correct in the fact that their sentiment analysis is hit or miss at best. Its because most of these vendors are completely automatic with no human audit incorporated into the process. It's like sitting in front of a computer with a fire hose as your screen blasting out "dirty", irrelevant data. Try having Ford as a keyword.

Did you know that many of the vendors all use a white label or API from one particular company? This gives many vendors the same or extremely similar data.

In the end, along with advising them to get a partner also advise them to do their homework when selecting this partner. Methodology and technology are critical because as great war strategist once said, " It is better to have no intelligence than bad intelligence." Or something along those lines...

Thanks for the post.

Data Quality Varies

You're absolutely right: From "hardly usable" to "mostly accurate", there are huge variances between data quality in vendor offerings.

The challenge is that "data quality" is actually a combination of many things.

* The sources the vendor pulls in. The platform must have the ability to start with the right dataset. As you've mentioned, this is where a lot of the vendors have differences - but it's worth noting that quantity is NOT quality. There are certainly vendors out there with vast breadth of data and little quality.
* The way the vendor processes data. Going from unstructured text to usable information is hard - but this is where the vendors prove their worth. But of course there are many ways to do this - text analytics in its many forms vs. manual - none of which are perfect.
* How the vendor delivers insight. The real power these vendors give buyers is insights - but if the tool can't deliver these in dashboards, reports, services, then buyers will suspect poor data quality.

I'm watching the space closely, hoping to see data quality continue to improve. But because it's a tough job, for now it's all about setting proper expectations of how we can use this data and how we can't.


What about social profile data

Nate - In addition to what you describe here, I would add the entirely separate category of social profile data. Think about all the fields of data Facebook, LinkedIn, and others have on users. Marketers can access this information when they integrate social on their site at the point of authentication. When a user logins/signs up for a site using FB connect or an aggregation service like my company, Janrain, provides, the brand can request permission to access specific fields of user data. As marketers are looking for increasingly granular levels of data, the interest in what the social networks collect on their user base is growing. I think the next evolution in data will be not only the social profile data, but combining with other sources of data to get a full picture of the consumer and use it to guide campaigns and marketing strategy. What do you think? Thanks.

Great point

Thanks Lisa - you make a really good point here. I didn't touch on how social data can be plugged into CRM, or into targeting, but it's of great value for both. In fact my colleague Lucilla De Sarlo is working on a report right now about how social profile data can be appended to email databases to improve the segmentation and targeting of those databases. Look for that research to publish in a month or so.

Thanks Nate - I'll look for

Thanks Nate - I'll look for that research. Lisa

Lisa - I find this to be so

Lisa - I find this to be so true! I work at a small company with limited resources, but will often find myself going deeper than the social listening tool provides and looking at the profiles on Twitter, "about" sections on blogs, etc. to try to pull out at least general insights and similarities about the people talking about the brands.

I find that having even general social profiles to give to clients and to aid in further research and development of strategies is crucial. I do see this approach becoming more and more important as these tools develop and as people see the value in aggregating this information to develop campaigns.

Listening tools

Very interesting article and actually goes back to the very fundamentals of my classic product marketing training at Colgate. Listen to the customer, focus on their needs and then (assuming you have the product that can satisfy them) find out how to reach them using "targeted" media - which originally when I was doing TV ads was based around time slot profiles, OTS and frequency. The good old bad old days....but I digress.

Please excuse my ignorance but is there a resource that can give me an idea of what listening tools are available?

Talk to Zach!

My colleague Zach Hofer-Shall is our expert (indeed, the world's expert) on listening platforms and vendors. You can see the list of industry-leading vendors he included in his most recent Wave analysis of listening platforms here: And if you want to read the full report, one of those vendors (Converseon) is distributing it in return for your email address:

Listening tools

Thanks for this. I have given Converseon my email although I am not really their target as I am not operating at the enterprise level.
I help SMB's generate traffic/customers online and social media is part of that goal. However they really dont have the infrastructure, funds or will to employ an enterprise level solution. As a consultant, I am looking for a tool that can help me monitor the basics ... FB, Twitter etc for my clients and then use that in my content creation/response plan. I suppose I am likely looking for "Converseon Lite" :)

Many, many vendors...

Hi JK-

The listening space is littered with options and with a surprising amount of viable players. I often refer others to Ken Burbary's list for a start. You could also take a look at this list which provides summaries of each, but weeding through 450+ is a daunting task...

What I'd look for is a tool that has an "agency model" - one that resells or partners with agency buyers. This will give you the flexibility to monitor for multiple clients, potentially by passing through a small charge, without having to set up too many partnerships and/or dashboard profiles.

Good luck,

Domino's Pizza sourcing creative from social

I just found a great new example of item #2 on my list: Use social data to source your creative. Mashable is reporting that Domino's Pizza is going to run social comments about the brand and its products on a Times Square billboard -- filtered only for langauge, not for sentiment -- and then, intelligently, will send people whose comments were used a video of their comment on the billboard.

See the details here:

Please clarify some points

First off, thank you for this thought-provoking article, Nate. However, I feel what you're saying is somewhat contradicting. On the one hand, you believe that social data is reliable enough to help businesses craft their messages and react to their consumers. On the other hand, you believe it shouldn't be used to measure brand impact because "social data is based on unstructured and non-comparable inputs" from "biased" resources.
If this is the case, how can social data help in crafting a business message or effectively reacting to consumers or any of the 5 points you mentioned above? I am confused and would appreciate it if you clarified things to me. Thanks in advance for your time.

Qualitative v Quantitative

Marketers regularly rely upon qualitative data -- from sources like focus groups and customer interviews -- to inform their messaging strategies (and their product strategies as well). What qualitative information lacks in statistical significance it often makes up for in depth. Social data fits nicely into this category.

What qualitative data can't tell you however -- not focus groups, not customer interviews, and not social data -- is the impact your marketing has had on your overall audience. For that you need structured, unbiased, representative quantitative data.