The 3 Levels of Online Conversation Metrics

Why Reaching Level 3 is Business Critical

By Greg Tozian
Digital Strategist, Brand VO2

Depending on the volume and tone of online comments, products, services and even whole companies can be made or broken in the wake of web postings and worldwide chatter

Assuming that the people who run sales and marketing for your company did not just return from a 10-year hiatus on the moon, your organization is actively concerned about the inevitable, unsolicited online conversations related to your brand, products and services.

150 billion minutes of chatter

There are more than 150 billion minutes of online conversations annually.

As we all know, depending on the volume and tone of online comments, products, services and even whole companies can be made or broken in the wake of web postings and worldwide chatter.

The vast importance of what web users say to each other about companies and their products was driven home recently by the announcement that Apple purchased social media analytics startup Topsy for more than $200 million. What it indicates is that web, and business decisions, are leaning more strongly toward so-called “semantic search,” the context behind what people say online, and being forward-looking in measuring the business ramifications of those web conversations.


For the sake of this whitepaper, we assume that in order to sell things, attract new customers, inspire loyalty among existing customers, and solicit online conversations from which you may mine data, your company has already established some kind of presence on one or more of the Big 6 social media channels.

  • Facebook
  • Twitter
  • LinkedIn
  • Google+
  • Pinterest
  • YouTube

We list YouTube as a social network, because it is social by nature, allows comments, is used by companies as a major online presence, and is the third most popular web desination (after Google and Facebook).


However, social media is only the most obvious form of online conversation about brands. Added to the chatter on the bluechip social media platforms, your company’s health in the marketplace — like it or not — is going to be effected by the presence of additional comments occurring through a range of other business-critical web channels, including:

  • Blogs (particularly key influencers)
  • Discussion Forums
  • Review Sites
  • E-Commerce
  • News Media
  • Wikipedia (and other aggregate info sources)
  • SlideShare
  • Instagram, Picasa, Flickr (and other social photo sites)
  • Quora (and other questions-and-answer entities)

Of course, a first line of defense in dealing with online conversations is for a company to establish their own corporate “fan” and official corporate web presences on the most popular social media platforms. This takes us, naturally, to Level 1 of Online Conversation Metrics.


At its base level, looking at analytics using “free” or low-priced tools provides a sense of the general effectiveness of your social media efforts. (I put free in quotation marks, because nothing is free. Even internally, it costs real time and money to have someone maintain and track social.)

Monitoring and reporting on the efficacy of these platforms is also a great way for agencies and internal teams who supply social media-engagement information to justify the continued corporate involvement. And it may make company employees feel good about their social media outreach, and reach in general, even if it doesn’t prove return on investment (which it often does not).

Level 1 table-stakes data from the various social platform channels include:

  • Size of Community (numbers of fans and/or followers across platforms)
  • Number of content interactions (mentions, likes, re-tweets, etc.)
  • Percentage of positive, negative and neutral comments (or comment tone)
  • Volume of referred traffic to main corporate websites from social
  • Unique page views, total pages viewed, time-on-site from referrals, etc.

Of those five standard social media analytics categories, the first three can be simply done with third party tools. They give you a general sense of how effective your efforts are relative to the platforms they track. The latter two — referral traffic and page views — can be obtained through Google Analytics.

Measuring visits your social media channels push to a corporate websites or ecommerce engines can help you understand a type of conversion forthcoming from social media content. That can be valuable. But as I said, you should be prepared to pay someone, internally, or externally, for some heavy lifting and dot connecting to get usable data around these metrics.

Having some data at this level will also help you determine social media’s content relevance. With that knowledge, you would be wise to adjust your social media by channel to give fans and followers more of what they respond to, and do less of the kinds of posting that does not inspire engagement, much less go viral.

Share of voice simply translates to the number of times your brand is mentioned on a social media channel versus your competitors.


A deeper level of analytics around social channels is to look into the “share of voice” that you are getting through your social efforts.

Share of voice simply translates to the number of times your brand is mentioned on a social media channel versus your competitors (represented as a percentage of the total number of mentions). Getting data like this does not have to be expensive, at least for the data acquisition piece.

Free services, such as Social Mention, are well known for providing share of voice information for the leading social media channels. This information is another indicator of whether or not your social media efforts are essentially on the right course. But, again, there’s not much else in the data: no real sense of return on investment, no learnings of the kind that can be had from a deeper level of online conversation analytics.

Additionally, shrewd companies use the monitoring of their social media platforms as a way to gauge customer sentiments. This allows customer service to handle complaints or just answer product questions (before they become barriers to purchase). There are many examples of companies turning this kind of tactic into a win-win (for customers and company), with Comcast being only one of the most obvious corporations to leverage social listening in order to improve their satisfaction ratings.

In this phase spikes or dips in your social media metrics should be analyzed to determine the “why” of the change. Of course, your company’s online conversation situation may be subject to the vagaries of broader market and economic factors. However, upticks and downturns in your social media success occur continually for many other reasons.

  • Spikes might occur because you ran a series of online or traditional ads, because you opened a new store, because you did a sponsorship of an event, or you were successfully active on social channels (getting a larger number of “shares” and “re-Tweets”), or you issued a blog or press release that was picked up by various, influential online entities.
  • A dip might occur because you weren’t posting enough on social channels, or your company got negative press, or a new product you launched received too many negative reviews online.

There are many third-party software platforms that zero in on such factors in what we call Level 2 social media monitoring and analysis. These softwares are variously described in the marketing world as “monitoring,” “listening” and/or “engagement” tools.

When it comes to actually ascribing reasons for upward and downward patterns in the social media metrics, in Level 2 the “analysis” work is generally done within the software, algorithmically, even though reports that are forthcoming may appear to have had a human eye and brain directing their conclusions.

The resulting analysis may endeavor to explain the why’s of online performance (positive, negative, neutral). Some even claim that they can tell you whether you obtained a good return on investment with your social platform programs.

Regardless of the quality of the reporting, logic tells you that the additional application of direct human analyst thinking (a well-trained professional looking at the data and connecting the dots and making recommendations) will trump any algorithm yet developed in mining actionable insights from online conversations. We talk more about that in Level 3 Online Conversation Metrics.

One of the main problems for companies analyzing online conversations for insights is that their internal staffs are drowning in the “fire hose” approach to big data (also known as social sentiment mining).


One of the main problems for companies regarding online conversation metrics is that their internal staffs are drowning in the “fire hose” approach to data. Third party software providers know this. The good ones filter results for their clients before they deliver actionable data.

Part of the issue of why most current social media monitoring tools are as ineffective at delivering actionable brand knowledge is that they were developed from the get-go by data mining software experts, without the deep partnering oversight from brand marketing experts.

One logical approach to achieving Level 3 outcomes is to develop Online Conversion Metrics tools in co-creation between brand marketing gurus and the software engineers from the point of inception. Unfortunately the vast majority of current OCM products are big data engines that were retrofitted to serve “marketing” needs.

Brand marketers should be involved in helping design the specific software back-end functionalities and the weighting of data to be pulled in the data-extraction steps to ensure the building of a brand health discovery engine/knowledge tool. This approach should yield online conversation reporting that is more brand-accurate, actionable and prescriptive of brand health.


In fact, you might think of this new breed of solution as a Brand Health Knowledge Tool, rather than a “me too” monitoring system. This kind of more refined solution is only now starting to emerge in the marketplace. (A desire to help break ground in this necessary new territory was the inspiration for our development of Brand VO2.)

If you read the tea leaves embedded in interviews with the CMO’s of the most forward looking companies, and those most successful currently at engaging customers online (including in social media channels), what they repeatedly ask for is knowledge to see as early as possible where markets are going. They want to know what people want, and what the smartest new-product and service innovations would be to satisfy customer needs and desires. These indicators make me believe that Level 3 Online Conversation Metrics is going to be a marketing gold rush in the near future.

In Level 3, evaluations are made based upon a variety of better preparations and deeper post data-extraction insights, including:

  • Using more finely tuned and brand-centric “lenses” for viewing and weighting data
  • Mapping resulting social media indicators to company Key Performance Indicators (KPI’s)
  • Identifying threats and opportunity trends in markets, early on
  • Identifying new product and innovation pipeline opportunities
  • Measuring over all brand health through Online Conversation Metrics

Here is a glance how some of these elements would be treated in an effective Level 3:

More Finely Sourced Online Conversations

Conventional conversation monitoring and engagement products “boil the ocean,” extracting data from billions of data points and doing massive data dumps on unsuspecting brands.

A Level 3 solution would be a more finely tuned Discovery Engine, pulling from brand-expert-selected data sources, achieving a higher level of brand-and category-centric information (not boiling the ocean, but focusing on a pond filled with better sources which might include select social media, ecommerce, news, review, blog, video/images, aggregate information sites, etc.)

Category-Specific Word & Phase Matrix Use

Just as you wouldn’t comb the entire web for data, you would not rationally pull all conversations online that mention a brand in even the most haphazard fashion and try to make sense of that tangle of information. Many social media monitoring tools use what are known as semantic analysis and even so-called natural language processing to aid the accuracy of their endeavors. However, in Level 3, the goal should be to use brand and category-specific word and phrase matrixes that are informed by relevance to brand health.

This makes sense when you understand that the way in which consumers talk about a dimension like quality in reference to cars (e.g., reliable, safe, maintenance-free) is radically different than how consumers talk about quality when it comes to pasta (e.g., tender, tasty, nutritious). If you are turning your social media analysis over to a machine you’re unlikely to capture the nuance of language that exists in every category.

Better Lensing of Data Against Brand Relevance

The so called data lenses that examine online conversations in a perfect world would be Level 3 informed, weighting the relevance of the content of the selected conversations against actual, predetermined attributes that are relevant to a brand, its categories and products and services. And, again, this is not the case in virtually all Level 2 strength social media monitoring tools.

In the actual analysis phase, Level 3 solutions would go beyond the “why” of spikes and dips, to tell companies “what” they should do with the data, particularly to leapfrog the competition/ market (see the Early Warning section, below).

Mapping Online Conversation Indicators to Company KPI’s

A key to Level 3 should also be tying marketing activities to the company’s specific Key Performance Indicators (KPI’s), which include:

  • Market Share
  • Sales
  • Web traffic
Optimally, you should mine the long tail of big data, to unearth threats and opportunities that are invisible when companies only look for insights in the ocean of online conversations.

Early Warning Trend Spotting of Threats and Opportunities

There have been companies using social media now for several years to perform a form of trend spotting, albeit through laborious sifting of that ocean that was boiled. The media has trumpeted companies, including high-tech hardware manufacturers, who monitor social platforms to “hear what customers want.”

However, most software platforms have not been sophisticated enough to identify customer needs early on, before the volume of the conversation becomes so loud that customer needs and desires are obvious to a broad audience of interested business and media players.

Optimally, you should mine the long tail of big data, to unearth threats and opportunities that are invisible when companies only look for insights in the ocean of online conversations. These lower-level issues are also often invisible in call-center environments because they can be drowned out by high volume topics.

Identifying New Product and Innovation Pipeline Opportunities

Level 3 tools should be much more finely tuned, and read the tea leaves much sooner, actually giving users a distinct innovation and new product idea advantage against competitors.

When done properly, your company comes off as anticipating marketplace desires. And your development of new products/ innovations can be informed and even directed by this early, greater knowledge of customer’s needs and desires.

At this step, too, the assistance of expert brand and business health analysts (human beings) to interpret and make recommendations on data is critical to Level 3 strength, actionable insights and reporting.

Measuring Overall Brand Health Through Online Conversation Metrics

The cherry on top for corporate marketers and product managers should be delivering to the C-suite a measure of overall brand health and marketing ROI, based on proofs forthcoming from the data extracted online.

Those kinds of results are not forthcoming with the most widely adopted social media platform monitoring tools to date. But they are available in Level 3 platforms like Brand VO2.

In Level 3, how vital a brand and its products and services are — as proven by brand-and category-focused data extraction and analysis — will emerge as a quantifiable number that will be a truly meaningful measure of brand health in real time.

When that happens the conversation around Online Conversation Metrics will move from talking about social media monitoring platforms and reporting to focus on sustainable brand health and business performance.