by Evan Dunn
Social monitoring and listening tools vary immensely in capabilities and price. Some cost $30/month, and some range between $10,000 and $40,000/month. Unfortunately, "you get what you pay for" only partially applies. Here are some things to know about what the best social listening tools need to have in order to solve your business problems:
1. Price per query. Some tools make you pay per mention, some per query, and some (the best) let you have as many queries as you want with as many mentions as you want. Find the last type, unless you only need to monitor one or two brands/products. The amount of experimentation you will want to do with social monitoring tools will leave you hungry for more flexibility. You can analyze mentions of any person, any product, any brand, or any industry term. One project we ran involved over 65 queries of different brands and products, not to mention "search within" functions that included another 38 sub-queries.
2. The best tools offer historical analysis, some going back 2 years or even more. Since social data snapshots are not as valuable as trends (which allow you to correlate and predict), historicity is essential.
2. Social monitoring & listening does not and should not only refer to social networks like Facebook and Twitter. It's cool, hip and trendy to analyze these, but there is a wealth of other information out there buried in forums, blogs and news sites. The best practice standard in the social monitoring SaaS industry is 200-600 million websites. Some accomplish this by hooking up to WordPress and Tumblr, and some by creating huge libraries of websites they analyze.
3. Natural Language Processing: this really separates the men from the boys. As a trained linguist, I can tell you there are plenty of social monitoring tools pretending to have sentiment analysis that truly don't. The tool you buy from should have linguists on staff, trained in computational linguistics, and fluent in whatever language they are building for. The result should be a tool that provides sentiment analysis in at least 2 or 3 languages (some get up to 6 or 7), as well as other attempts to dive into behaviors, purchase intent, emotions, etc. Much can be gleaned by decoding linguistic signals on the social web. For example, I can tell you that the internet showed a significantly greater amount of "trust" and "confidence" in the Patriots than in the Seahawks leading up to the Super Bowl, despite Deflategate. You want this insight for your brand.
4. Data correlations. You want to know where people are talking about the search terms you enter, or how old they are, what gender they are, and maybe even how wealthy they are. It's difficult to do, but it's possible. Ask tough questions of the social monitoring tool you're investigating, and if they don't have at least 2 of these functions (Geo, Age, Gender, Financial Demo), drop them on the spot. This information is incredibly valuable for ecommerce analytics, marketing insight, persona development, message development and consumer insight.
5. Query building. The traditional search query language is Boolean, and it works great. Some tools assist you with developing Boolean queries, but that can sometimes be limiting. Make sure that whatever tool you select gives you the power to build the queries you need, in order to maintain near 100% accuracy in the results. There are a few tools that still maintain the archaic practice of building the queries on your behalf, which only means that you can't be sure how accurate or comprehensive they are being.
6. Exporting capabilities and API access. You'll need to build your own pretty graphs sometimes, so make sure you know what your BI or analytics process is like, and what your marketing analysts need for reports, before you sign up. Some tools make you pay extra for access to a JSON API, some provide it within the cost you pay for the dashboard use of the tool.
The world of big brands are waking up to the power of social listening, but in some strange ways. You've probably seen blogs about live social listening dashboards that are really sexy, projected on a screen in front of 10-30 social media lemmings. The problem is the data is rarely actionable, and simple business question are difficult to answer in a split second while staring at screen that updates every few seconds.
But the awareness about social monitoring tools is good, because more brands need to use them, for PR, program planning/evaluation, campaign evaluation, message testing/development, predictive analytics, revenue analysis, and more. Even international development organizations - such as UN Global Pulse - use social listening data to develop reports. Here's a breakdown of some different use cases that I see springing up around the world:
PR: Social monitoring is perfect for brand monitoring and reputation management. You can see how much people are talking about a politician, exec, brand or product, and how they feel about it. You can analyze the data over time, and look for spikes and dips based on major events and stories. More and more PR agencies are adopting social monitoring software, especially because some tools give you alerts whenever (for example) negative sentiment mention volume jumps past a certain threshold.
Program Planning/Evaluation: Focus groups are nearly obsolete thanks to online surveys and social data. They return inaccurate results because people give affected responses in focus groups. And they're expensive. Imagine, instead, using social monitoring data to analyze how people feel about competitive brands and products, what their pain points are, and what they like. I've run a project just like this for a major brand, and the results were wonderful. We'll see more and more businesses catching on to the value of social data for program evaluation.
Campaign evaluation: does a new campaign actually make an impact on increase in volume and sentiment of product mentions? Social monitoring can tell you. More and more CMOs are adopting tools that provide a 3rd party look at campaign effectiveness.
Predictive analytics: this is the future of marketing. Imagine forecasting social mentions, social sentiment, website impressions, sales and revenue. Marrying these data sets makes this possible, with the right data scientist up your sleeve. Barring unforeseen crises, businesses can predict performance better than ever before.
The other major trend is the All-in-One Social Media tool, or even All-in-One Marketing platform. Salesforce and Adobe lead the way in the second one, but there are tools that are aiming just at the universal social media management, marketing, monitoring and advertising platform dev. It's something CMOs are dreaming of (only a handful though), but it's a long ways off.
Businesses should be able to get a feel for how forward-thinking a social monitoring platform is. If you've got some clout, you should ask to speak with the CEO - they'll give you the best sense of vision for the product.
Also, getting their lead engineer on the line is always helpful, as they are the ones to grill with the really technical questions (such as: "Does your sentiment analysis software use machine learning by parsing syntactic constructions?"). If you want to show them some cutting-edge syntax-based sentiment analysis, check out Stanford's Treebank Sentiment Library.
Be careful before entering a year-long payment plan - be sure to shop around, and do several live demos. When you read reviews by other users, dig to find whether their business size and budget are similar to yours, and whether their use cases resemble yours. If you're adopting some of the best technology in the space, you'll pay $3,000-$4,000+ monthly.
I mentioned above the trend towards the All-in-One Social Media tool, or even All-in-One Marketing platform. Be careful with these - it takes much more vision than most software developers have to build a universal platform. For example, Salesforce's Radian6 social monitoring tool is (last I checked) well behind the leading platforms in NLP it executes on top of the data, and is formidably expensive. Big kahunas like to buy their way through the space, leaving you with unintegrated, piecemeal options that still manage to cost an arm and a leg.
The goal, as with any software adoption process, is to seamlessly integrate to solve the highest number of business problems, catalyze the generation of the largest amount of revenue, and pay the lowest premium.