Leveraging Social Media Intelligence with the Qualitative Research Community

A follow up post to the QRCA Flash Webinar designed as an introduction to social media research (what it is and how to get started)

By: Frank Gregory & Kayte Hamilton

It probably doesn’t come as a shock to anyone reading this that the coronavirus pandemic is now the most talked about topic in the history of social media. A perfect storm for social media conversation volume growth has emerged, with consumers across the globe stuck at home under stay-at-home orders, wanting to express how they feel about the situation, how their views of everyday topics have changed because of the situation, or simply to virtually connect with others and laugh to take their mind off the situation. The obvious way to do this is from the comfort of their couch: by posting on social media.

As consumers’ behavior has been forced to change, the landscape for researchers has changed as well, with some in-person methodologies being impossible to execute for the near future. Therefore, researchers should consider a pivot to new execution strategies, including social media intelligence, as a new tool in your toolkit.

Over time, and especially from the experience my colleague Kayte shared with me (as she works across all forms of qualitative research every day), there’s been a slow build to the adoption curve of online qual – meaning using any form of a digital medium to execute traditional in-person methodologies. From an observational viewpoint, it seems there was a slow momentum for the qual industry as a whole to skip analog research options completely. Yet our recent health crisis forced the future on us much quicker than some could adapt and that was true for both suppliers AND clients. Those who had already dabbled, if not fully embraced, the digital options were ahead of the online push that was necessary during the economy slow down (then shut down). Pivots were easier for some than others.

While I work in social media research every day, Kayte agrees with my perspective that one of the necessary tools to embrace during this change in the qualitative research landscape is the use of social media research to enhance or complement qualitative (or quantitative!) solutions. The problem is that [unofficial] adoption curve. Kayte has shared with me her experience from her early exposure to some social listening tools in past research roles (years ago and the landscape of those early platforms have changed dramatically). It was confusing. Some gave her a flood of information that she couldn’t find time in her day or week to synthesize, others were really overly complicated to set up in the first place which made leveraging the insight too daunting. Most of her exposure to “social media mining” these days are simple Google searches, Facebook browsing, or following specific hashtags on Instagram. In other words, she admits that, like most qualitative researchers, she’s been only “skimming the surface” of social media research because of her experience years ago regarding the barriers to adopting social listening tools.

But, we’ve also discussed how much these tools have changed over the years. I explained my perspective as someone who is in these tools every day, in that “social listening” does not equal “social media research”. Social listening tools are only one of four key types of social media analytics tools out there today, with each type of tool designed to answer specific types of research questions.

Social listening tools help answer the basic question of ‘what are people saying about X topic?’, but audience segmentation tools can provide further insight into what makes up a brand’s social audience: what are their interests, lifestyle habits and media consumption traits? What other brands and/or celebrities do they follow on social media?

Taking it a step further, influencer evaluation tools allow brands to assess which of a shortlist of potential celebrities or influencers they should work with, based on the celebrity’s audience overlap with the brand’s social audience, some desired mutual lifestyle interests and consumption habits of that audience, and how authentic the celebrity’s audience engagement is. Finally, a fourth tool category is owned channel performance, which allow researchers to understand which of a brand’s posts are resonating best and why, and most importantly benchmark that performance against competition to learn from what is resonating for competitors.

But regardless of the type of social media analytics tool, the biggest aspect that appealed to Kayte as I was encouraging her to jump into social media intelligence more fully is the reminder that it’s really never too late to get started. Unlike other “in-the-moment” approaches qualitative researches might implement, you CAN go backward in time and analyze social media conversation in time chunks. Your starting point doesn’t mean the day you “go to field” like in other research options; you can go back historically 24 months or more to see how consumer perceptions have shifted over time. As opposed to trying to ask a consumer “how were you feeling about X topic 2 years ago vs. 1 year ago vs. 6 months ago vs. today?”, social media intelligence allows you to find the millions of consumer comments discussing that topic over that time period, since the posts consumers made 2 years ago are still there waiting to be analyzed. So, using the coronavirus pandemic as an example, kicking off a social media intelligence analysis today doesn’t mean you’ve missed out on the last few months of social conversation trends, including how the coronavirus has changed the way consumers think about certain brands, industries, and behaviors.

Every single company has been impacted by our current events. Some more than others; some have seen positive gains, while others not so much. Regardless, there are ways to tap into these conversations and use the information to your advantage, from proposals to report writing.

A couple weeks ago, Kayte & I presented a very quick and high-level overview for the QRCA (Qualitative Research Consultants Association) on what social media intelligence research is and how it impacts the qualitative research industry. We got a lot of great questions, but couldn’t answer all of them during the live broadcast. So, we wanted to answer some of them here and provide a written account of our talk. This webinar was part of the QRCA’s response to COVID-19, providing training tools for their members. This content is FREE and available for viewing here.

The rest of this post is full of follow-up questions we were unable to answer in our short presentation. There were so many great questions, so wanted to do our due diligence by answering these now. Happy reading, and please do reach out to Kayte or myself to keep the conversation going…it’s a hot topic! As you can see by the content below, it’s also a passion project for us, so we offered more, than less, in our attempt at education.

FOLLOW UP Q&A:

Is this primary, secondary or observational research? This was “debated” in the comments section a bit because it really depends on the objective use of how your using the findings. I feel that, as a concept, social media intelligence research is observing and analyzing conversations that consumers are naturally saying. Consumers aren’t being asked a question to respond to, and they aren’t interacting with a researcher at all. They are simply putting their thoughts out there in a public, online forum for the world to react to. One attendee in our webinar compared it to standing in the middle of a mall, watching, listening and recording people’s conversations as they walked by. You could argue it is primary observational research because of the real-time nature, but there’s also a recording you could reference from conversations in years past. You could also argue it is secondary observational research because there’s no question being asked of the participant. So you could land on ‘quasi-secondary’ research, in that you are observing everyday consumer conversations, historically and as they happen, simply by being ‘a fly on the wall’…for a fly in the middle of a crowded mall, if you prefer that analogy! Regardless, this sparked a lively and interesting discussion in the chat room of our webinar that we enjoyed reading later. Continue that lively discussion in the comments of this blog post if you’d like!

Has market demand grown for this? Over the years, demand for social media conversation analysis has absolutely grown, as the tools and analytics industry have matured around it. When social media first emerged on the scene in the late 2000s, understand conversation trends was the Wild Wild West. Social channels were more open to providing conversation data to marketers, but the analysis functionality of the tools was very minimal, leaving the insights minimal as well. Now, channels are more protective of personal privacy (as they should be) and so some conversation types are limited in what researchers can analyze; but the power of the analytics tool providers have never been stronger, and demand for consumer insight from social conversation has grown every year. Given the coronavirus pandemic we mentioned above, demand has continued to grow in 2020 as consumers are using social media more than ever before. As researchers though, it’s our role to dissect the noise from the insight, as you’ll get vocal consumers from both ends of the spectrum; lovers AND haters.

You mentioned channels getting more restrictive, what do you mean? As mentioned on the webinar, Facebook has closed themselves off from social tools in recent years, as a result of the Cambridge Analytica issue after the 2016 election. This was an important and necessary (and good) step for the industry, to ensure consumer privacy. But, it does present a challenge for market researchers. While individual Facebook conversations and the trends generated from those conversations are no longer able to be seen, leading tools are still able to provide important insight into 1) the conversation occurring between a Facebook user and a brand on the brand’s page through authentication of the brand’s page ownership in the tool, and 2) what online public sources such as news stories and blog articles are getting shared the most across Facebook. The good thing is, while Facebook’s restriction has caused challenges for the industry, most major social media trends still occur across all 3 of the major platforms (Facebook, Twitter and Instagram) and therefore you can gain directional insight from the trends seen on Twitter and Instagram, which is almost always reflective of the general trend on Facebook around the same topic. In other words, the same story or piece of content that goes viral on Facebook is likely to go viral on Twitter and Instagram as well, so you’re unlikely to miss out on a major trend simply because of Facebook’s restrictions.

What does ‘authenticating’ mean again? In order to gain insight into Facebook and LinkedIn conversation interactions between a brand and its follower audience, and in order to gain more consumer/brand interaction insight than publicly available on Instagram, it is important to use tools that allow for authentication of your client’s brand pages. Most tools make this pretty easy, where you can generate a URL link that you can email to your client before kicking off your analysis project that explains why it is important to authenticate the brand’s admin information within the analytics tool in order to gain better insight in the analysis. For the client, it’s as simple as one or two clicks to authenticate their brand’s admin information so that you can start your research, and most tool reps are great at helping guide you through the process the first couple times.

What demographic & psychographic data is available? In general, all demographic and psychographic information available in social media conversation analysis for slicing and dicing is based on 1) what the consumer has said in their past public posts, 2) what they include in their bio/profile, 3) what other brands/public organizations/celebrities they follower and 4) what settings they have in place (ex: if they have turned location services on within the device used when posting publicly on social media, you can see generally what geographic region they were in when they posted). The most reliable demographic data found in social tools is typically age range, gender and geographic location (city, state, country), as many of the social channels ask users to input this information upon signup. But, other demographic information (ex: HHI, marital status, # of kids, ethnicity, etc) is limited, unless the person has indicated so in their past public posts or bio/profile. That being said, audience segmentation tools are able to analyze the past public posts, bio information and what other brands/organizations/celebrities followed to categorize further by interests, lifestyles, lifestages & media consumption preferences. For example, if someone frequently uses #momlife in their public posts, the keyword ‘mom’ in their bio, and has indicated themselves to be female and in her early 30s, then audience segmentation tools will categorize her as a ‘millenial mom’ for segmentation analysis purposes; but, we might not be able to know how many kids she has. Professions and other persona-based groupings are similarly categorized for directional insight, but researchers should be aware that social media intelligence tools can’t replace other research methods here given the limitations (ex: you can recruit a focus group attendee list to cover specific demographic & psychographic information, and you can ask qualifying questions in survey research to cover your demographic & psychographic research targets, which you can’t do in social media research).

What is available in terms of sentiment & emotional analysis? Almost every tool out there includes positive vs. neutral vs. negative sentiment categorizations, so that you can see what the positive sentiment drivers of conversation were for a brand or topic vs. negative drivers. This is one of the most important and useful aspects of social media intelligence analysis. Leading tools are able to go one step further and use natural language processing and machine learning to categorize into specific emotions, such as joy vs. fear vs. anger vs. confusion vs. sadness. However, given the nature of consumer slang, no tool is 100% accurate in this department; most are directionally accurate at about an 80-85% accuracy level. For example, most tools have been able to get past common slang (ex: ‘that Jay-Z song is sick!’ is positive while ‘ugh, I’m feeling sick again’ is negative); but, there are other slang phrases that present problems for the algorithms (ex: sarcasm or snarky comments like ‘wow, so happy about this’ with the rolling eyes emoji). The tools are getting better each year with this, and ultimately, I feel that emojis will help to drive improved accuracy in the future, as they are analyzed next to text in social comments. But in the meantime, when you see a set of social posts in your query results that are incorrectly categorized with the wrong sentiment or emotion, most tools allow for bulk overrides, and some even include cluster analysis categorizations to ‘train’ the tools over time, so that you can improve the accuracy of sentiment scoring in your analysis above the 80-85% levels that the tool can provide out-of-the-box. Finally, almost all of these tools allow for exporting of the comments, so that you can use the raw data for other analysis purposes such as visualizing next to other data (sales data, website traffic data, etc) or uploading into a separate text analysis software program. Some of the strongest tools also allow visualization functionality within the tool, so that you can upload relevant data (like sales or web traffic) into the social listening tool, to visualize next to the social conversation data (as opposed to pulling the social data out of the tool to use elsewhere).

Can I track & analyze competitors, and are there any restrictions? One of the most important use cases of social media intelligence is competitor tracking. Thinking back to the four different categories of social intelligence (that we presented in the webinar), each has their own way of gaining competitive insight. With social listening tools, you can monitor any mention of a competitor’s brand(s), analyzing the ‘share of voice’ of the conversation volume your brand is driving vs. that set of competitors, the positive/negative sentiment shifts for the competitor brands, and the influencers in that competitor’s conversation. Audience segmentation tools can provide you with insight into the interests, lifestyle traits and media consumption habits of your competitor’s social audience, while influencer evaluation tools can give you insight into your competitors’ spokespeople or influencers they are working with (which influencer posts are resonating best for that competitor’s audience, what that influencer’s audience makeup looks like, the overlap between the influencer’s audience and the competitor’s audience, etc). Finally, owned channel performance tools are designed specifically to assist with benchmarking your brand’s content vs. competitors, in terms of which brands are providing the highest consumer engagement vs. a competitive set group or larger industry group, and which specific posts are getting the most engagement for a competitor vs. everything else they are posting. The only restriction here is that you can only see your competitors’ public-facing engagement metrics (ex: likes, comments, shares, retweets, etc), not any metrics that would only be available behind the competitors’ admin information for that social channel (ex: how many impressions a competitor’s Facebook post received).

What about ‘social media influencers?’ How are researchers using these tools to evaluate influencers? The social media influencer industry is exploding now, as each brand is realizing that the most authentic way to reach a new audience is to work with someone who is already seen as authentic within that audience community, and already likes/uses the brand’s product or service. As brands and influencers partner to create more and more content by the day, more influencer evaluation tools are emerging. However, usage of these tools is still not mainstream, where many marketers are still simply ‘using their gut’ to decide on which influencer(s) or celebrity will be the face of their next campaign. Therefore, market researchers can differentiate themselves by understanding how to identify influencers that a client should consider working with, and then how to evaluate that shortlist of potential influencers for the best brand fit and impact. The influencer evaluation tools can provide researchers with a view into how authentic an influencer’s audience is (in other words, is everyone truly a fan of that influencer/celebrity, or did the influencer ever use a schticky promotion to ‘buy’ fans that actually don’t care and don’t engage with content), how engaging their current content is, and most importantly, whether the influencer’s audience lines up well with the target audience persona that the client is looking to reach (ex: affinity to certain interests, lifestyle habits and media consumption habits, strong brand vs. influencer social audience overlap, etc). Finally, once an influencer is decided upon, and campaign content has been launched, these tools can provide researchers with an understanding of how impactful that content was in reaching the desired audience, and how the content performed vs. the average engagement levels that influencer’s everyday posts typically receive.

Besides the social channels, is there a list of blogs and forums that are available to analyze? I don’t believe that a tool provider would be able to provide an exhaustive list, as there are thousands. However, in general, the vast majority of blogs are public-facing in nature and therefore will be able to be captured based on the keywords you are searching for in your query. For forums, any public-facing forum that isn’t hidden behind a log-in screen would be available to analyze as well. Forums that force community members to log-in in order to see any conversation threads are considered private, and therefore you wouldn’t be able to access these for analysis. But, if a forum allows general conversation threads to be seen without a login (maybe they just make you log-in in order to post or comment), this is considered public and available for analysis. If you are assessing a tool vendor and wondering about a particular blog or forum, you can always inquire, and they can look it up and let you know about its availability. Ultimately, blogs and forums are important ways to gather significant insight from conversation analysis, as I’ve found that they often lead the researcher to more in-depth detailed conversation about products, services or topics from knowledgeable enthusiasts vs. the quick snippets of conversation found on traditional social channels.

What do you mean by query writing, and what are some best practices? The query is the first step in setting up your analysis. It defines what you are going to be analyzing; in other words, what portion of the social media conversation universe is relevant to you. As Kayte mentioned in the webinar, learning these tools is akin to learning any new online platform for qual research; some have a much steeper learning curve than others, and you need to weigh the benefits of learning these resources yourself versus bringing in a professional partnership.  Almost all tools are based on Boolean logic (meaning AND, OR and NOT statements), with some providing the researcher with a blank slate to write themselves while others provide a form to fill out in laymen’s terms so that the tool can apply the Boolean logic on the backend. For example, if I am researching conversation around the portfolio of Hilton Hotels, but conversation about Paris Hilton and the city of Hampton Roads in Virginia would be irrelevant to me, then I would write my query like this: [((Hilton OR Hampton) AND hotel) NOT (“Paris Hilton” OR “Hampton Roads”)]. Some tools allow for more advanced Boolean operators, such as NEAR statements. For example, if I want to research conversation about the how the restaurant industry is struggling right now because of coronavirus and employees are out of work, but I don’t want literally every mention of ‘restaurant’ on social media that could be irrelevant, I can write a query like this: [restaurant NEAR/5 (coronavirus OR “out of work” OR struggling OR “let go” OR furlough OR need help)]. The number ‘5’ next to the NEAR operator tells the tool to only pull in the social conversation that uses the word ‘restaurant’ if it is within 5 words of one of the qualifying words I added. The main point here is to write a query to get started, assess the results to see if there’s anything irrelevant popping up, and then go back and refine the query accordingly. Most tools allow a Preview button so that you can see the most recent conversation your draft query is pulling in, before you even ‘run’ the query, which helps with quick refinement. If you feel like you need more guidance here, either a consultant with subject matter expertise, or the account rep for the tool your using, should be able to provide it for you. 

How do I assess how much historical conversation data I need? It really depends on the types of research projects you’re looking to do, and what topics you’re looking to analyze. For some topics, you may only need to pull back 6 months, such as if you wanted to understand how consumer perceptions about the product, company or industry have shifted from pre-COVID to mid-COVID timeframes. For annual event-based topics, like the social conversation surrounding Super Bowl ads, it will be important to least go back 13 months from the last Super Bowl, to ensure you’ve got a comprehensive view of last year’s conversation. However, for other topics, it might make sense to analyze 12 to 24 months or more of social conversation trend, if you are evaluating how consumer perceptions have shifted among groups of consumers as new product developments have come out, or as that consumer group has navigated their journey in the customer life cycle. Ultimately, you’d need to assess the historical data need based on the specific research questions you’re looking to answer, and then take budget into account in terms of which historical data package you select.

Is there one tool that is great for all different types of social media analysis? Unfortunately, no. The general consensus amongst social media researchers is that no tool has nailed every aspect of social media analysis really well, instead each having their own niche: social listening, audience segmentation, influencer evaluation or owned channel performance. Some have tried to be a ‘all-in-one’ solution, which I mentioned in the webinar, but I wouldn’t recommend those yet until they address each aspect of social media research at an equally in-depth manner, in order to justify the investment. Some tools have been able to nail two (or even three, to a degree) of the categories of social media intelligence well, which I referred to in the webinar on the slide labeled “What Are The Strongest Tools Out There?”. After identifying which type of social media intelligence tool makes the most sense to answer your research questions, I would start by reaching out to the tools I list in my diagram on that slide, and assessing the package options and functionality according to your budget and your needs. You can also reach out to a consultant with subject matter expertise to help you with a 3rd party, objective and customized tool assessment.

Is there an option for partnerships in order to gain access so that I can analyze on my own? Absolutely. Given that most tools are subscription-based, preferring annual contracts, there can be a limitation if you are ONLY considering using the tool for a single one-off project. Instead, you should assess the different tool options based on your longer-term needs, thinking about the TYPES of questions you typically are asked, so that you can find the right tool to invest in that you can use over time to answer many different research questions, as well as use for your own internal research (ex: white papers). Your other option is to partner with a subject matter expert (a consultancy, an agency, etc) who is very familiar and comfortable setting up the queries and conducting in-depth social media analysis, and has some of the leading tools already, so that you could leverage their licenses for one-off custom research projects. These partners can assist you in objective, unbiased tools assessments, query writing, dashboard building and analysis, and typically are a way to save cost vs. a direct investment in a tool license on your own. Ultimately, this becomes a question of knowing your skillset as a researcher and deciding whether you want to invest in growing your skillset to include social media intelligence, or whether you are more comfortable hiring a partner with an expertise in this space.

What about if my client says that they have access to a tool in-house? Could that be an option? Again, absolutely. Inquiring whether or not your client’s organization themselves have a tool is a great idea. It may be that a different department like Social Media or Analytics has a tool and is using it for tracking conversation, but your primary contact just has to do a little internal digging to find out what conversation tracking & analysis they are currently doing, and whether you can gain access to your client’s license specifically for your research project. Besides saving you cost, asking this type of question of your client could differentiate you as a qualitative researcher, as your client might not have thought of social media analysis as a research approach yet. They may just simply be tracking the conversation volume around their brand as a data point for an executive dashboard, but not leveraging the tool in the way that a seasoned qualitative researcher with an insights-generating mindset would do so, in order to gain insight from the conversation trends and get the most out of their investment in their in-house tool license. Our training as insight miners, layered with a primary project & social media intelligence, is what makes the results that much more powerful; it’s braiding together these two methods and not keeping them in separate bubbles. If your client isn’t using their current social media collections in the insight department, it may be because some of these people in those analytic roles don’t have insight training and are more data collectors. They can tell you a lot of WHAT is happening or being said, but need our qualitative skills to convert these findings into actionable data points.

Any other scrappy ways of doing social media research? Besides my above suggestion of seeing whether your client has a tool you could leverage, or a partner/consultant can work with you to leverage their license that can save costs vs. a direct investment in a leading tool yourself, there are of course ‘scrappy’ ways of doing research. However, in general, while these scrappy ways of tracking conversation might be enough for some marketers, I’ve found that most serious researchers get frustrated rather quickly using some of the ‘free social listening tools’ they find via Googling, like the ones I referenced in the webinar. That being said, Google Trends and Google Alerts are of course helpful for finding out what is happening right now in terms of news stories and search keywords, but aren’t going to give you any historical insight. The free listening tools I mentioned on the webinar are similar, with minimal (if at all) access to historical data, and limited query functionality outside of tracking one Twitter handle’s mentions or one or two keywords. But, to what I referenced earlier in this article that Kayte mentioned to me, you can always simply go on Twitter or Instagram and search for a hashtag to find out what people have said recently about a topic; but, it’s going to be a pretty time-consuming and inefficient process if you’re trying to do any significant research into conversation trends, sentiment drivers, etc.

Is automation or AI being used at all in this space? Some tools have been able to use AI to gain more accurate sentiment scoring and conversation clustering. Also, leading organizations and analysts have found ways to take social media conversation data out of the tools via APIs and merge with other forms of internal or public data (sales data, website data, census data, etc) to help make internal processes or analysis tools more efficient. These types of brainstorms are great for the future of the industry, as we find more ways to gain insight and drive action from social media trends. However, one area of automation and AI that I would caution against is the actual consumer interaction between a brand and a consumer on the social channel. For example, automating a brand’s Twitter response to a customer tweeting about a bad experience they are having with the brand can only exacerbate the situation, as customers are more and more savvy these days and can recognize that they aren’t actually interacting with a helpful, empathetic human, but instead a cold, automated ‘bot’. I’ve seen situations where customer interaction automation has been tried and failed, driving more negativity to the situation since the customer can quickly ‘tag in’ their friends to show them how ‘insensitive’ the brand is, which can go viral for the wrong reasons.

Is it OK that one of the use cases of social media intelligence is for generating leads? This is a good question, as it really is a distinction that needs the context of which type of user is using the tool. If you are a brand marketer, one of the many use case options for social media intelligence tools is to listen for conversation related to your product or service, and respond in real-time when someone is mentioning a desire for that product. This could be B2C (Wendy’s listening for anytime someone influential in social media mentions craving a burger, and responding with something clever that gets that influencer’s social followers thinking about Wendy’s) or B2B (a service provider sponsoring a tradeshow monitoring for when an influential attendee of that tradeshow mentions on social media how they are enjoying the new companies they are learning about, and the service provider responding to the attendee’s tweet with ‘don’t forget to stop by Booth #XYZ to see us!’). Brand marketers use real-time social engagement opportunities from social media intelligence every day to drive brand awareness, consideration and loyalty. So, for a market researcher, the distinction becomes…if you’re using social media intelligence tools to analyze consumer perceptions around your client’s product or service, it is probably not a good idea to use the opportunity to respond to individuals selling a separate service; but, if you’re using social media intelligence tools for your own research firm’s internal marketing needs, then it is perfectly fine to listen for and respond to opportunities to join conversations around phrases like ‘I’m looking for a great qualitative researcher’, in order to sell your own services.

How do you handle sensitive consumer information (for example, Adverse Events in the Pharma industry) when you see this being discussed in social media? The short answer here is to follow the guidelines set forth by your client’s company in terms of reporting Adverse Events, just like you would for any other type of research. Also, it is important to emphasize that social media intelligence tools, at their core, are ONLY going to provide a user with publicly available information (meaning nothing that a social channel or website would consider a ‘private’ post bound by privacy settings), including only the personally identifiable information that the individual chooses to make public in their profile. 

Do bots skew the data? Yes and no. Bots have been, and always will be, an issue social media channels have to deal with; particularly Twitter. However, Twitter and the other channels have done a good job in recent years of clamping down on bot activity, suspending or deleting accounts that are clearly not human based on analyzing their post activity. So when a researcher is looking at the results of the keyword/phrase query they plugged into their tool, they do have to be aware of and knowledgeable enough to spot a bot and therefore disregard the conversation driven by that bot (going back to the query to refine to filter out any posts from that bot’s handle). For example, if the results show one Twitter handle that posted hundreds of times in a short timeframe about a topic and yet has close to zero followers, that’s probably a bot and the researcher should refine their query accordingly. But, not to fear, as the vast majority of bot traffic is caught by the tools and automatically filtered out, making this not a major hindrance to researchers trying to analyze the legitimate conversation trends. Finally, if you happen to find significant bot traffic for a given topic, you can always flag to your tool’s rep or a partner social media intelligence consultant, who can flag it directly to the social channels like Twitter for potential suspension or deletion of the handle.

Is “social listening” the same thing as “social media scraping”? NO. ‘Scraping’ infers that you are using outside data collection software to directly take data from a website on your own, for your own analysis purposes. While this might be fine for some websites, it could violate other website’s terms of service and/or violate consumer privacy policies for the users of that website. ‘Social Listening’ tools work with the owners of each website (the social channels like Facebook, Instagram, and Twitter, as well as blogs, forums, and review sites) to ensure that the data usage policies and terms of service are being abided by, including that they are using the website owner’s approved APIs to collect this data. Therefore, ‘social listening’ tools do all of the data collection work for you, working within the legal parameters set forth from the website’s owners, and letting its users know when certain data is not available (ex: ask any social listening tool provider whether you can receive Facebook conversation from individual Facebook users). Using leading and highly reputable social listening tools becomes important when you also consider analyzing review site conversation, since that is the source type that varies the most: TripAdvisor allows for reviews to be analyzed, but Amazon.com does not (as of the time of this publication), for example. It’s best to ask the social listening tool provider which review sites they have available for analysis, since they’ve already done all the work for you to assess the legality of data access. I’d strongly advise not working with a ‘website scraping’ company to conduct ‘social listening’ research, especially if they are discussing scraping social channels like Facebook and LinkedIn.

Are there any resources to learn more? Yes! My #1 go-to answer here is to go to YouTube and search ‘social media analysis’, ‘social media analytics’ or tool’s brand name. There are many tutorials that go through the high-level concepts, including seeing the tools in action. Also, many of the tool vendors themselves will have video and blog resources that can help with conceptual learning, as opposed to just selling you on their particular tool. Finally, following industry sources like Social Media Today, Adweek, Brandweek, Advertising Age and Mashable’s social media section will give you up-to-date news on the latest features available for marketers; these sometimes hit on the research & analytics world when there are major updates to the social channels’ analytics backends that help drive the 3rd party listening tools’ data access & functionality.

FRANK GREGORY:

Social Media Intelligence Practice Lead

NorthStar Solutions Group, LLC

[email protected]

KAYTE HAMILTON:

Client Partner Director

InsightsNow Inc.

[email protected]