Polygraph Media Blog

Twitter and Gnip - A match made in data heaven

Last week, Twitter announced the acquisition of Gnip in what will certainly amount to be a key milestone in the evolution of social media.

2008-2010 Young social media managers emerge at brands to help them navigate new social media tools.

2011-2013 Social media turns increasingly from an earned/owned media channel to a paid media channel.

2014-2016 Old skills become obsolete. Brands, who are now spending a lot of money on the now-mature social media channel, need to optimize their spend. Young social media managers are replaced by data-oriented internet marketers.

If you believe this framework -- IMO you should as it's entirely what we've seen across consumer brands -- you understand there are a few implications:

1. Old skills and first-generation tools become obsolete if there is no commitment to innovation.
2. New innovations can emerge quickly and solve major pain points in consumer brands.
3. Access to data, at scale, becomes very important.
4. Whereas all this UGC (user-generated content) is a massive collection of Facebook comments, Tweets, ratings/reviews, chat logs, forum discussions, words become gold for the data miner.
5. What's being said, when, how, and by whom becomes something we can solve with technology -- and something much more desirable than say reading a hundred thousand comments/day manually.

Without going into too much detail about my own company or Twitter (I was there three weeks ago and am under NDA), I think I can say that Twitter realizes just how important data and scalable APIs really are. I was very impressed with the vision that I heard from a wide range of personnel at the meetings I attended, all the way to the top.

We do for Facebook pretty much exactly what Gnip does for Twitter. Increasingly, our ability to help brands get access to Facebook feedback at scale is sparking great conversations and projects with consumer brands.

Everyone is moving in the direction of big, scalable social data for a range of scenarios.

• Combine with CRM
• Understand customer feedback
• Help inform community management strategy and tactics
• Conduct competitive intelligence
• Better informing advertising spend and algorithmically running ads specifically to defeat or out-message a competitor
• Using social data to target individuals most likely to transmit a marketing message, visit a retailer, try a new product/service

We live in very exciting times, as I believe the next 5 years will bring us some remarkable marketing innovations that tap into the millions upon millions of pieces of social feedback that are generated worldwide every day. I am just glad to see one of the major players both acknowledge this and act upon it in a meaningful way. Kudos to the Twitter team for moving the industry forward!

Augmenting Opinion with Data

For those of us who are more left-brained thinkers, it's a pretty exciting time right now in 2013.

Silos to data have been broken down in many instances... and there are a plethora of data sources available for fee or for free that can be tied together.

Data visualization tools have progressed well beyond the visual and data processing constraints of Microsoft Excel. Business people can make things happen in tools like Tableau or SiSense, and D3 is emerging for situations that benefit from significant customization and drilldown.

And perhaps most importantly, the "data-driven" approach is moving beyond early adopters and becoming something that many consumer brands and enterprises want to both know more about and execute more effectively.

Call it "revenge of the nerds" if you will.

The secular trend in business is to give data a seat at the decision-making table. Savvy executives understand that there are a lot of ways to measure expertise. And the person who asserts him/herself as an expert might not be as infallible as once thought. He/she may bring biases to the table that data can confirm one way or another.

A multidisciplinary approach to data is necessary to take full advantage of the opportunities. Why? Because it's a problem with different dynamics that require different skill sets that often don't lie in a single person.

1. Opportunity identification -- a keen knowledge of what is out there, as well as the classes of data that are generally easy and those that are difficult to obtain.
2. Data "creativity" and translation to business problems -- the ability to listen to client needs/requirements and propose/present data-driven solutions to help improve upon current decision-making.
3. Understanding of statistics -- familiarity with different statistical methods used to build explanatory and predictive models to enhance decision-making that is currently done by feel or more subjective criteria. Key here is to employ statistics *accurately* and let the conclusions come forth from the data.
4. Data visualization -- knowledge of tools/resources available across the aforementioned tools and technologies such as Tableau, SiSense, D3, etc.

This is what it takes to become more data-driven. Fall down in one of these areas, and data efforts can miss the point.

What I find that most enterprises and consumer brands want is the best of all worlds. How do I take advantage of the people and expertise we have, but augment them with the conclusions and accuracy that only data can provide?

Nobody wants data to replace original thought, but rather to augment it and turn people into smart, discerning, and data-driven powerhouses.

Data Proliferation and the rise of the Chief Data Officer

Quick update today -- the "data proliferation problem" is just a theme I've heard from a lot of conversations recently.

A fellow "data" startup CEO mentioned to me the other day that he lamented just how hard it is to sell "data" and "data services" to people in brands and the enterprise.  Marketing departments have millions of dollars to spend on print advertising that they know does not work... but a modest $100,000 investment to help people understand their data opportunity requires significant explanation and is often met with blank stares.

We're in an odd point in time kinda like we were in with social media in 2007. Evolved enough for people to recognize the promise, but too early for budgets and organizational alignment to have taken place. This does not frustrate me all that much -- but it is a growth opportunity for small companies such as ours and a chance or visionary bigger companies to jump out to a quick, early lead.

I have a lot to say on this subject and will follow up here and in some of my guest blogging on how business is evolving. Suffice it to say that winners will be data-driven decision makers. They'll have context about what is important and what is not, and they'll execute well. They will know statistics, business goals/metrics, and they'll understand the interplay between operations, customer data, and their own systems.

Companies will increasingly hire or appoint Chief Data Officers -- and not just those in data-intensive businesses either.  A single person and/or office will handle issues related to making data useful and actionable for a business but will also be careful to not expose private data to people who should not see it. It's a tough problem, but a fascinating opportunity.

Deloitte discussed this recently in a fantastic brief paper. I think it was interesting that Alibaba just appointed their former Chief Data Officer as CEO of the entire business. Expect that trend to continue.

The folly of analytics and value of big data analytics

Analytics has become a hot topic over the last 12 months as middle-managers are asked tough questions by their executive stakeholders.

For all the insanity around social marketing, mobile apps, and "conversations", many CMOs, CIOs, and CFOs want to know exactly what they're getting for their money.  Budgets are growing and advertising spend is fluid.  People want to know the impact of strategy changes and tactics tweaks. But don't worry.  This pattern is not just familiar to you in your business, it's happening everywhere.

Middle managers have turned to analytics to rationalize what has happened -- to measure success and failure.  But I'd contend that some of the early analtyics offerings are doing a bit of a disservice to the very clients who rely upon them.  Charts, graphs, and data visualizations are compelling -- so they must be authoritative.

But are they really?

The one question you should ask of any analytics offering is simple:

Do I truly understand the metrics presented in this chart?
As a marketer, do I understand how to improve the metrics presented in this chart?
And finally, do I understand clearly how these metrics impact my company's ability to make more money, save money, and generate a return on investment (ROI)?

As I talk with corporate marketers about social marketing, content, and advertising measurement, I'd say that at least half the people do not understand the data presented to them by Facebook Insights.  Few, if any people, understand what Facebook's Reach and PTAT (People Talking About This) metrics really mean.  Almost none understand how those metrics should be used and interpreted.  And fewer know exactly how they'd improve upon those numbers.  The same goes for abstractions on hard metrics provided by third-party analytics providers.

The only way to get a true understanding of your performance is for your metrics to match the things you know how to do.  On Facebook, it's simple:

• Page Likes -- how many Fans do you have, and is that number growing?
• Posts -- how often do you post?
• Post Type -- how often do you post different types of content to keep your Page fresh?
• Shares -- how often are people inspired to share content with their friends?
• Comments -- how often do people comment on your Page?
• Likes -- how often do people like content that you post?
• Engagement Rate -- by time period or post, how well does your content truly perform?
• Clicks -- how do your ads and/or Page Posts generate clicks to a web site?

Facebook is the top of the funnel for many consumer marketers -- it's an awareness vehicle that can help you create a better conversion rate for other online activities.  It can also identify your top fans and people who advocate for you online and offline.

But (and I'll pick on Facebook Insights for a moment) if you're using metrics such as Reach and PTAT as religious metrics that you watch feverishly, I think you're missing the point.  You're succumbing to the folly of analytics as opposed to the data-driven answers that big data analytics can provide.

If you go to the effort of measurement, you're ahead of the game.  But you're really ahead of the game if you take a data-driven approach.  Be sure to measure the very things you know you can affect.  Otherwise, you're just guessing.

Social Media Measurement & ROI - the next big thing

According to The Center for Marketing Research at The University of Massachusetts.

We agree.  Link below:

http://www.customerthink.com/blog/social_media_measurement_and_roi_next_focus_of_inc_500_fortune_500

Big Social Data -- Explaining Offline Behavior

Over the past six months or more, you've probably heard the term "big data" floating around as if it is supposed to solve all your problems in business, marketing, operations, etc.  Just "access the data" and you'll be smarter than your competition.  You'll react to customer feedback.  You'll be a visionary.  Blah blah blah

All of this creates a situation where they hype is through the roof, but few people can really speak to real-world scenarios and useful applications of big data.  That's probably why "Big Data" is high on Gartner's Hype Cycle 2012 Chart.  Lots of possibilities, but not a lot of answers just yet.

We work primarily in the social data realm at the moment, although we have been getting more & more requests to integrate offline data with social data to understand patterns.  Alongside first-cousin ROI (return on investment), the big question today is whether or not social media is a scorecard for how you're doing or if it can influence offline behaviors, purchase patterns, etc.

The takeaway for us so far is that different businesses certainly have different dynamics.  Only with full access to the data can you really draw any firm conclusions.

Contact us to learn more about how we can help harness big data for your company or brand.

Data Mining... a Primer for Marketers & Decision-Makers (Part 1)

A bit of a dry post today, unless you're a data geek like we are.

Since our launch, we've entered a few customer and partner conversations that would've been best served with an introductory primer in data mining.  I would estimate that a majority of marketers don't have a firm grasp on the requirements and/or what the opportunities are in the field of data & data-driven decision making.

So here's my "business guy shorthand" on data mining, what's available today, and the opportunities all of this affords.

First of all, content on the Internet has exploded as others have chronicled over the last few years. From the linked article above: 

"...compliments of Dave Turek, the guy in charge of supercomputer development at IBM: From the year 2003 and working backwards to the beginning of human history, we generated, according to IBM’s calculations, five exabytes–that’s five billion gigabytes–of information. By last year, we were cranking out that much data every two days. By next year, predicts Turek, we’ll be doing it every 10 minutes."

How did this happen, you ask?  Look no further than the social media movement, which democratized publishing to the web via blogs, put our videos online via YouTube and Vimeo, broadcasted our short-form opinions via Twitter, and chronicled our life through Facebook.  And alongside those services are a series of other services that all collectively help us chronicle our lives online.

I personally love this infographic below that illustrates just how much data we're all creating today.  It's a far cry from when we needed a web developer to post content for us.  Additionally, check out this compendium of stats from 2011.

In addition to the revolution in self-publishing, we've simultaneously had a revolution in the collection of data in the corporate realm.  Gone are the days of Day Timers, Day Minders, etc.  In with the world of Salesforce.com.  New Point of Sales (POS) systems are no longer analog but rather "digital", which allow for the collection much more data that can be analyzed and explained. Every part of a modern business can be digitized and recorded today for data analysis -- and smart businesses are doing just that even if they don't have the analysis capability just yet.

Additionally, government agencies are publishing data to the Internet now with increasing frequency -- making records public that were once published to books, stored on microfiche, or available only upon specific request in-person.

The greater point is that we're all creating data about ourselves every day -- personally, as citizens of our locality, as employees, etc.  Although the "quantified self" movement is relatively young, we're already uncovering introspective ways that we can get smarter about the things that we're doing using data as our guide.

Through collecting and analyzing data appropriately, all stakeholder groups can benefit.  A few quick examples...

• Self -- quantifying exercise, meals, etc. for better health. 
• Business -- quantifying operations to understand efficiencies in staff, travel, etc.  Mining customer data to understand market segments & purchase intent.
• Public Sector -- combing through the data to better isolate fraud potential.  Using data to optimize tax receipts.

The common term across all of them is "efficiency" -- using data to become more efficient.  In a hypercompetitive world, efficiency is the way to unlock more profits through decreased costs, increased revenues, and/or improved experiences.

Additionally, data can be predictive and prescriptive.  Let's say you are in a family of four and have bought two cars over the last 8 years.  You have a luxury car with 20,000 miles and an SUV with 200,000 miles.  It's a simplistic case, but the data in this example suggests that you might be in the market for an SUV soon and years away from another luxury car purchase.  That's the predictive part.

Odds are if you got the SUV serviced by a data-driven car dealership, you've heard from them recently about trading in that old clunker in exchange for a shiny, new SUV.  That's the prescriptive part -- send everyone likely to buy a new car a new advertisement.  Like it or not, but it happens all the time already... all while many corporate marketing departments are not fully optimized to take advantage of the opportunities available via data.

As we've talked with a number of companies about our work in big, social data, we've discovered a wide range of companies in terms of adoption of data analysis -- collection, normalizing, analysis, data-driven decision making.  Some are leading the way, and others still make analog decisions.  To be expected, although we do think the shift towards data is a long-term trend that will be impossible to ignore for most.

In the next blog post, I'll talk more about the mechanics of making sense of it all -- from data normalization to analysis and informed decision-making.

How Facebook can inform a Presidential Race

Today, we thought we'd use Facebook data to tackle one of the most anticipated decisions of 2012. Who will Mitt Romney choose as his running mate?  And can Facebook data be used to help make the decision?

Data is great and all, but what good is data if you can't make decisions with it?

First, a little context.  A number of studies have already shown that social media data is helpful to gauge voter sentiment:

• Crimson Hexagon analyzed Twitter data in 2010 to accurately predict a victory for Harry Reid.
• Social data (Twitter, Facebook, Google+) was also used to accurately predict Boris Johnson's victory in the London mayoral race last month (June 2012)
• Facebook themselves pointed out some correlations in the data as well following the 2010 US mid-term election.  Candidates with more fans won Congressional races 74% of the time and Senate races 81% of the time.  Not bad!

Now this field is not without detractors -- some academics have focused on the flaws in using social data to predict the future.  We agree that Twitter data is dirty relative to Facebook data, as it's easier to create dummy accounts for Twitter and many people use Twitter primarily for marketing purposes and not to share what they're really doing/saying/thinking.

But it's also sour grapes in our opinion.  Anything short of asking every single man, woman, and child is in essence a shortcut.  There are going to be flaws in using data to make assumptions about the behavior of a large audience.  There are flaws in traditional market research, statistical analysis, etc.  This is no different -- in fact, it may be better.

With social data, we have the largest self-contributed data sets the world has ever seen... on par with the financial and personal data owned by major credit card companies (Visa, MasterCard, AmEx, etc.) and marketing companies (Acxiom, Dun & Bradstreet, and Harte Hanks). Additionally, people don't really have the motivation to falsify information here as they would perhaps when talking with a market research professional.  Facebook is capturing the unvarnished truth of who everyone is.

As such, we think that Facebook data is the most accurate data set in the world.  And it's relatively unmined at present.

-- 

Anyhow, we wanted to show an example of how reactions on Facebook can give decision-makers in politics a clear sense of who is "known", what they've done, and how they've managed to keep their supporters engaged.  In that sense, politics is not that unlike what social marketers do all the time -- keep the brand front & center and something worth rallying around.

In this admittedly simplistic example, we looked at a number of factors in an attempt to narrow down the Field from a large number (8) to a top-priority running mate.  We collected data on the following:

• Current Facebook Fan count
• Percentage of “Likes” over 5 quarters of posts
• Pro or negative “Romney” in page posts
• Pace of new Facebook fan acquisitions predicted between now and election day based on previous pace (how many new Friends per day)
• Polarization – the activity of each candidate on “pro” or “against” 3^rd party Fan pages

Each of these things was selected in order to test something relevant for a national election.  Is a candidate a national or regional candidate?  Is a candidate doing a good job of keeping the base engaged?  Is the candidate extending his/her reach to a broader base of enthusiastic supporters? Does the candidate create controversy or negative passion in other voters that might improve the prospects of the other political party?

The infographic is below.  We'd love to know what you think.  Also check out our YouTube video where we walk through the findings and explain the numbers in more detail.

Video - walkthrough of our Romney Veepstakes Infographic - Facebook Data

In this quick video below, we review the recent infographic that we released about how Facebook data can be used to inform political campaigns.