The Advent of Machine Learning as related to Social Listening

The advent of #MachineLearning as related to social listening precipitates a lot of honest-to-God hard work during a transition from human-led to machine-led insights. This will happen because many humans will be needed to create custom attributes and custom classifiers for the machine to use in the new upcoming #AI driven insights industry. Humans teaching the machine to think like we do, to analyze like we do, to understand nuance, culture and the evolution of language (keep in mind, social tech experts, I am referring to a truly global machine analyzing the languages of the world…not just English).

Since there are many gaps in social data, we face an uphill in pairing other data sources to the snippets we get from public sources of conversations. The key to a complete “AI Social Data Intelligence Machine” will be sources such as Axciom, census data, background check sources, private networks like Doximity/A Small World and finally, if possible, the nexus of data from Facebook and Amazon. This particular nexus is a kind of holy grail that the leading machine intelligence experts will eventually open to brand leadership. But previous to such events, we have a grueling journey as analysts, insights researchers and programmers to “teach the machine how to be a human”.

One of the most exciting branches of this “teaching” will be a self-aware machine, an artificial mind that will begin to offer recommendations…and then may decide to act upon these recommendations.

How To Perform Audience Intelligence – 2017



  • We discover your target audience and guide you in the best way to sell to them.
  • We connect you directly with your ideal consumers!
  • We learn the behavior of your consumers so you’re aware when they buy, how they buy and what medium best motivates them to buy.
  • We identify contexts for advertising, where your ideal customers spend their time, and identify the best manner to engage with them.
  • We profile and locate every single person online who match your ideal customer and invite them into your brand circle to expand your brand community.
  • Identify and connect directly to influencers who can create buzz and extend your offering to a larger audience.
  • We closely follow your competition and identify how they are ahead of you in business across social networks, forums and media. We share key insights with you, along with business opportunities for growth.
  • Develop demographic and psychographic information on your consumers to learn how to speak with your leads to ensure they convert.



AUDIENSE (SOCIAL BRO):  This is a Twitter and Instagram audience intelligence tool. The tool also offers a way to listen to everything a specific group of people has said since their entry to the Internet. The tool is a “gateway” into audiences and needs appending in PeoplePattern and PIPL to be more completely. Audience offers a unique and powerful dashboard for quickly assembling an audience related to a set of attributes, locations and even personality (Watson API is hooked up to Audiense).  A primary value of Audiense is: the quick development of large lists of Twitter handles related to how people self-describe in their Twitter bios. A secondary value is the ability to listen to everything these same handles have posted in Twitter since their presence on the platform. This combo is important as a way into an audience talking about a topic. It is also important as a means to interrogating a specific group about any other topics they talk about.

Audiense helps us answer the following questions:

  1. How many people are out there who self-describe as our ideal consumer?
  2. Where do these people live?
  3. What is our consumer talking about, besides our specific product, category or us as a brand?
  4. Who lives in our region that we can sell to?
  5. Who are the real people that talked about our category, competitor, product, and/or brand?


PEOPLEPATTERN: This is a relatively new entrant to the field founded by Ken Cho (founder of Spredfast). Ken has built a powerful engine for analyzing both conversations and bios from all social platforms. The tool also offers an impressive breadth of personas (both internally conceived and customized). These personas are populated with comprehensive insights on location, hashtags, interests, mentions, media, and many other attributes. The tool is in beta and is being used by a handful of agencies and brands now. The tool was conceived by Ken Cho and the storied technologist Jason Baldridge (  ). The tool is very specific, fast and deep, pulling up every bio and individual related to searches in People Pattern’s complex dashboard. One can also upload a CSV of social handles from most platforms and gain deeper insight into the individuals uploaded, along with full insight into what these people talk about. Primary values of PeoplePattern include the development of persona types and look-a-like audiences. A secondary value is the append of additional information related to individuals Twitter and Instagram handles. A third value is the ability to compare audiences side-by-side. Finally, the PeoplePattern tool is a convenient way to discover Twitter and Instagram handles of individuals/entities who talk about a subject in Twitter and Instagram. This is important because personas are a way to discover look-a-like audiences. The validation of these look-a-like audiences within PeoplePattern is exceptional. In addition, the depth of information about each persona type is truly exceptional.


PeoplePattern helps us answer the following questions:

  1. Who loves our category?
  2. Who loves our brand?
  3. Who is our critic?
  4. Who is our ally?
  5. What does the audience for our competitor look like?
  6. Who are the actual real people who are passionate about our category?
  7. Where do these people live?
  8. What persona types are passionate about our category?
  9. What other interests do our consumers have?
  10. What do lovers of our products talk about?
  11. Who is influencing the people who love our products?
  12. What media outlets should we use to reach our consumer?
  13. Who should we talk to about our product?


AFFINIO: This is an audience intelligence platform that uniquely identifies and groups individuals related to a query or uploaded CSV or social account. The tool is elegant in the way it creates nexus points of these persona groups/individuals AND major brands/themes/interests. There are unique tabs in the Affinio dashboard that give the user quick insight into potential collaborations with brands, micro-influencers or just a group of like-minded individuals “down the street”. The platform captures global data and is expanding beyond Twitter at this time to capture Instagram and other social channels. Affinio offers significant value to brand stakeholders who want to understand groups of real people talking about competitors and brand-relevant topics. Affinio also offers value related to discovery of real people for insights analysts to study and develop a POV related to any topic or theme. A primary value of Affinio: is the nexus of individual members of a report and interests. In particular, the segmentation of interests within Affinio is excellent.


AFFINIO helps us answer the following questions:

  1. What brands and interests do our consumer talk about?
  2. Who is influencing our consumer?
  3. Who is influencing our competitor’s consumer?
  4. What persona types do our consumers fall into?
  5. What content is influencing our consumer and our competitor’s consumer?
  6. What persona types should we be targeting in our campaigns?

The value of Spotright is that we gain an insight into the Buying Styles and Purchasing Habits of a specific set of Twitter handles. In addition, we gain insight into nexus points of brands, interests and any custom classifier with these same Twitter handles. There are other demographic insights within the Spotright solution that come from the deep Axciom and Epsilon sourcing by the original Spotright team – JP Lind (Epsilon and DoubleClick background) and Ed Messman (HiveLive, RightNow and KBM Group background). Access to the entire Full Contact data set was purchased early on in the history of Spotright, which allows for an exceptionally fast append process.


FULL CONTACT API:  A primary value of the Premium Person Enrichment Sheet in Full Contact and/or the API itself is: the append of other social handles, bios and data on individuals discovered within PeoplePattern and Audiense. When Facebook data was more open, the append of Facebook handles to Twitter and Instagram handles was a truly glorious experience. Even now, due to the archive within Full Contact, we do get a wonderful append of the Facebook handles, along with other social channels. This append is important because we can then find Facebook Groups and conversations (using Facebook Graph API) related to these same individuals discovered in another platform.


DATA MINING IN FB & LINKEDIN: The value in using a data mining solution to scrape individuals who self-describe as our target is very useful. In addition, the scraping of specific groups of individuals related to events, interests and locations is also useful. Finally, if a very deep recipe is written, a best-in-class data mining tool will open individual Facebook profiles and scrape a great deal of information, including what these individuals talk about. When one then takes these Facebook IDs and appends additional information in the tools described above, the size and specificity of an audience increases.


BRANDWATCH: This is a conversation analysis tool and now, as of Vizia2, a business process management tool. The audience tool within BW is still in development and is not as good as Audiense and PeoplePattern. Brandwatch is the world’s best source for public conversation data in 48 languages. The tool offers: complex boolean queries, customizable dashboards per business problem, full global coverage of 48 languages, along with brand specific dashboards, channel analytics dashboards and many other customized dashboards to overly the discovered mentions. It is also possible to perform non-language specific queries so that all mentions of a brand in all 48 languages show up in one’s dashboard. The tool has a scaled pricing model based on number of mentions and/or number of queries. There are literally thousands of case studies, resources, brand examples and agency examples to be found online for Brandwatch. The tool is the choice for the largest and best research agencies, including Ipsos, WPP (Kantar, JWT Intelligence, Ogilvy, etc.) and is the most used platform for conversation analysis by major brands. The team at Brandwatch is constantly innovating and includes access to scores of foreign language analysts for a truly global reach.



  1. I run a search in Affinio, PeoplePattern and Audiense using keywords OR by uploading a CSV of authors discovered in Audiense, PeoplePattern, BW (Brandwatch) and/or CH (Crimson Hexagon). I mix the tools and verify in this way.
  2. I wait for 2 hours for this search in Affinio/PeoplePattern/Audiense to populate
  3. I export the following sheets from Affinio within the Actions tab: Members, Interests, Bio Keywords, Locations, Domains, Hashtags, @Mentions, Topics. From PeoplePattern, I export specific audience segments, based on persona type, city, interests and other filters. From Audiense, I export specific audience segments based upon Real Avatar, Country, URL in bio, level of influence, time since last tweet.
  4. I create tabs within a single sheet with the above sheets.
  5. NOTE 1: (In some cases the Clusters from Affinio are unhelpful because I have to export from all Clusters and then I have to stitch the Social Handle columns from different Clusters together and then I have to upload to Audiense and PeoplePattern to get more specificity.)
  6. I upload the social handles from the Members tab in Affinio to Audiense and PeoplePattern.
  7. I export from Audiense and PeoplePattern and clean up.
  8. I stitch the Members (Affinio), Audiense and PeoplePattern sheets together in Excel.
  9. I clean up and put this fully appended sheet (using data from all three tools) into the first tab of the final document. (Remember, the other tabs have Interest Nexus from Affinio, words in bio from Affinio and Location data from all three in a fourth tab.
  10. I perform PIPL append on the names and links in the first tab.
  11. I stitch PIPL results (emails, phones, addresses) into that first tab in the last three columns.
  12. I perform queries in Brandwatch and stitch mentions into that first tab, next to Authors who have talked about category, brand or competitor relevant topics.
  13. I perform Author Group Monitoring within Audiense and stitch in Insights to the first tab.
  14. I perform Full Contact Person Enrichment to gain add’l social handles and add’l bios (verifying what PIPL gave me).
  15. I create code frames by hand and note insights while doing so. This is a separate process and gives us a powerful view into the conversation related to a category, brand, competitor or relevant topics to a business process. Export mentions from Brandwatch, delete duplicates, column titled Topic next to the Snippet, column titled Summary next to Topic. Code for percentage of (Human driven)
  16. I create a single page PDF with top line insights on the discovered individuals.
  17. I add Appendices to this single page PDF with supporting evidence for the Insights page(s). The Appendices are numbered and matched to each Insight.
  18. The final product is the Insights PDF, Insights Appendices PDF, and supporting spreadsheets. If the client is well-heeled enough, they get a branded dashboard that is “client facing” from BW and/or CH.
  19. Human analyst review and validation. (Human driven)


Enterprise level planning and campaign execution

Earlier this morning, I tweeted a series of actions a large enterprise could weave together for superior planning and campaign execution. The following sequence is an expansion of those tweets into a powerful method for market research and execution of digital campaigns/building of social communities online:

1. Achieve superior market intel by pairing newly minted predictive data solution at @Fliptop with the truly massive product index at @Indix

Fliptop is providing predictive lead scoring using internal and external data. Connects with CRM and Marketing Automation to deliver immediate impact to increase revenue. Indix is offering software for product aware apps, product intelligence, price intelligence, competitive intelligence. Indix has produced what they claim is THE ultimate map of all products on the planet. When one pairs scored leads relative to specific products, one has assembled pools of ideal customers. Pairing Fliptop with Indix is a method for assembling pools of ideal customers for products.

2. Continue the project mentioned in last tweet using @Fliptop & @Indix by appending deep social media resources via @FullContactAPI.

Full Contact offers a mechanism for appending social data to emails in the context of a spreadsheet. When an organization takes the “pools of ideal customers” from the previous step and then appends social data to each customer in its CRM, the result is a way into the deep conversations occurring around specific products. An analyst can also extract other meaningful connections through studying what specific pools of customers are talking about and what trends are developing in those pools. More on how to find those trends in the next step.

3. Bring further depth to aforementioned @Fliptop @Indix @FullContactAPI project by processing entities thru @RecordedFuture temporal analytics.

Recorded Future scans hundreds of thousands of quality public web sources, including news publications, high-caliber blogs, social media platforms, financial databases, government websites, and much more. From these open sources, RF identifies text references to entities and events. Then RF detects time periods: when the events are predicted or reported to occur. Each reference links back to the original source. RF can explore the past, present, and predicted future of almost anything in a matter of seconds. RF’s analysis tools facilitate deep investigation to better understand complex relationships, resulting in actionable insights.

When an analyst takes entities from the previous step in this process and performs analysis on these entities, he has the opportunity to find specific times in the future when a product might be better received. In addition, there may be an opportunity through using the specificity of Recorded Future to literally change a future event simply through when a product is launched. RF offers an analyst myriad signals from web intelligence that can inform a smarter strategic positioning for product launch, product recall and customer acquisition. By knowing what is happening with prospects and competitors through the RF panel, a product dev team can more accurately create a unique selling proposition in the marketplace.

Watch this video for more on how Recorded Future uses web intelligence to produce foresight in markets:

4. After using @RecordedFuture (step 3 in this tweeted process), apply @Brandwatch to extract meaningful mentions from resulting data set

Brandwatch is offering a superior listening & data visualization solution that will bring precise conversation analysis into this process. In addition, one may assemble a scale of influence relative to a product, market segment and/or pool of customers. Finally, Brandwatch has just launched a powerful data visualization format called Vizia that is helpful for C-Suite decision makers to quickly digest complex analysis on a real-time basis. The solution rivals previous set-ups by rivals and will inform the next generation of social media control centers globally. See this sample of Brandwatch Vizia in action:

5. As the penultimate step to ultimate customer intel, deposit nice neat sheet from previous 4 steps into @SpotRight machine for laser insights.

We can further refine insights derived through the Brandwatch methodology by placing data into the process, where truly massive amounts of customer intelligence have already been segmented, tagged and defined. No entity on the planet has assembled such an expansive and deep set of intelligence on individuals as The principals at Spotright bring a rich diversity of backgrounds and have assembled the ultimate catalog of individuals who buy products. A team choosing to use Spotright at this stage will find verification of previous findings AND, importantly, matches to its internal CRM of customers. Spotright is how to find customers just like your existing regulars.

6. For awareness-bulding actions based upon previously tweeted research sequence, execute campaigns using @JiveSoftware for maximum reach.

To bring an enterprise portfolio of products to market, use Jive Software. Now that the full market research has been completed in the previous 5 steps, it is time to take action. A powerful solution for taking action is Jive, which blends internal facing and external facing capabilities, making team-work across the silos far more efficient. In addition, Jive offers a suite of solutions for building awareness on a massive scale in the social fabric of the Internet. Watch this video for more understanding of how this works on the inside of a large FMCG group:

To run this six step process is a means to unlocking powerful potential in online communities for community engagement and sales growth.

Social Media Monitoring 2014 techniques and outlook

Customers who buy into a social media monitoring solution in 2014 want a page of insights backed by a “living” appendix. The “living” appendix is your technical solution where the customer can study the evidence for your insights. This evidence, in today’s real-time world, is a simple interface in the form of a mobile app that delivers abundant examples supporting the insights on your one-page.

Once the customer of social media monitoring feels confident in this insight-evidence system, he/she will ask you for ways to grow a business based upon the insights. One of the best ways to build a business based on insights from social media is a robust CRM (Customer Relationship Management data set). A robust CRM means that you have taken every person talking about a topic online and discovered their real name, current address, current mobile/landline, current email, current social links and examples of their interest (tweets, posts, comments).

When you have this depth of information about a market, and its occupants, then you have everything you need to develop engaging content. To develop engaging content means that you have listened to the conversation, you have studied and analyzed its participants and you have stepped into the conversation with something to say that contributes to the cutting edge of the conversation at hand. When you post into a comment thread, it is always wise to read the entire thread. When you reply to a tweet, it is wise to have studied who you are responding to so that your reply develops rapport and demonstrates a “seeing” of that particular person. When you develop a blog post on an industry, it is smart to research that industry. This is how great content is created and sustained.

A powerful tool for listening to and analyzing online conversations is Brandwatch. Run by Giles Palmer, the solution has myriad ways to segment conversation data, and the company culture at Brandwatch combines technical know-how, maturity and a certain understated confidence that makes every conversation with staff an “a-ha” moment for those just getting into social media monitoring. Start with Brandwatch when you are on the listening journey.

Once you’ve spent time listening, you will have the large (but fun) project of assembling a CRM related to a topic. While Brandwatch is very useful in discovering usernames related to a conversation, you will need a few other tools to “append” further details. I recommend SocialBro as an excellent tool for downloading extensive spreadsheets from Twitter of those in conversation on a topic. You will be able to combine spreadsheets from Brandwatch and SocialBro to create a near comprehensive starting point of every username discussing a topic. (SocialBro is also an excellent resource for timing your communication with a particular person or group).

Note: If you want to add to the thought-leader aspect of your spreadsheets quickly, I suggest a solution like PeerReach, an excellent and growing resource developed by Zlatan Menkovic and crew. You can go even further with PeerReach by studying the Twitter lists created by these thought-leaders. In many industries, especially outside the USA, it is vital to also study leading forums where literally 10s of thousands gather to discuss and interact around a niche industry. In addition, there are private groups in LinkedIn and at membership only sites with actual lists of industry leaders in a niche or region. If you want to add the latest news about an individual leader or organization, I recommend Recorded Future.

Next you will need to append contact data to these usernames. The contact data is often discoverable via personal URLs such as a LinkedIn or personal website. You can use services like Whois search, Spokeo, Intelius,, Leadership Directories, and many other regional people-finders to append current contact data. I also highly recommend an Executive or Recruiter level account at LinkedIn for appending contact data. If you want to go deeper, there are a host of OSINT (Open Source Intelligence) solutions that are constantly evolving to bring deeper info on individuals and organizations: “OSINT tools” feed from Twitter.

The ideal spreadsheet of a community surrounding a topic includes: First Name, Last Name, Username(s), Current Address, Country, Current mobile, Current landline, Current Email(s), Current social links, Several columns of sample posts that demonstrate interest in a topic (where an individual has posted about a topic), Influence score (such as Kred, Klout or PeerIndex). With such a spreadsheet, you can begin to contact thought leaders directly and engage in a learning process about the online discussion around a topic/service/interest.

As you grow in relationship with thought-leaders, you will find that your own posts, content and actions become more precise and effective…mainly because you will be “on the cutting edge” of the conversation in an industry. You will find that being on this edge gives you an edge and develops a maturity of outlook within you. You will find that others’ respect in you about this topic will grow as you listen, listen, listen to a core group of topic thought-leaders. Your contributions will be informed by these thought-leaders and “passionates” in a niche.

Once your network grows in a given niche, you will begin to see where opportunities exist, where something is lacking, where you can contribute to the need in that community. And, because you posses THE comprehensive CRM (or rolodex) for that niche, you will know exactly who to contact to get things done. The satisfaction in having ALL the information at your fingertips related to your chosen niche is immense! This is one of the great results or outcomes of best-practice social media monitoring: a deep understanding of, and connection to, a global niche community for the purpose of getting something meaningful accomplished.

Looking forward in the social media monitoring industry, we have the need for a solution that combines the listening, analysis and append processes into one dashboard. No-one has produced a satisfactory combo of these yet but the time is coming soon when the comprehensive solution will be finalized. The three “underdog” contenders I like most in terms of the “comprehensive solution” include SpotRight, Truelens and Reunify.

Additionally, there will certainly be a further evolution of the privacy debate in 2014. In an increasingly transparent eco-system, JWT Intelligence predicts a growing “Techno-Paranoia” and a flourishing “Privacy Marketplace” amongst consumers. In addition, JWT predicts the adoption of “Privacy by Design” at major global brands. Read more on these possibilities here. Social media monitoring companies should pro-actively prepare arguments for how listening improves the human condition, creating an opportunity for authentic relationship between brand and individual. An argument for a more open, authentic human community may be found in the principles of Council Circle. There is also, of course, a business model for social media monitoring companies to develop the “Privacy by Design” and “Privacy Marketplace” technologies and programs suggested in the JWT report.

Social Business Intelligence Advance #4

The business intelligence solution that marketers want is the following: a cloud-based service that is comprehensively analyzing in real-time networks of regular customers AND matching these customers (and their friends) to upcoming deals specific to past purchases. For instance, if Amazon were to fully integrate its data with Facebook’s data, a profound level of matching would become possible. Senior leadership within major FMCG and retail entities should be actively cultivating technical vendors and in-house technicians to achieve this type of customer intelligence solution.