The SiteFocus Blog

Welcome to the SiteFocus Blog! For more, please visit wwww.sitefocus.com

[sitefocus-blog1]

247 Insights is here!

Artificial Intelligence (AI) is the talk of the town. The promise is great, but the road can be long, expensive, and confusing. How can you “use” AI today? 

SiteFocus is excited to announce that 247 Insights is now available to organizations and individuals alike.

247 Insights is the world 1st web-based subscription service delivering AI curated predictive insights on streaming business and financial news.

Whether you're at home or in the office, thousands of signals are created every second. Oceans of information can move markets, define trends, and create new opportunities. How do you find the signals that matter the most - that have the highest impact?

247 Insights is powered by Communications in Focus (CIF) - a universal analytics engine that uses AI technologies to find insights humans can’t see. It works like an army of analysts to automatically seek out and tell users what the most important subject matter is - and why. 

247 Insights amplifies the user’s intellectual productivity. It presents insights in a  dynamic visual quadrant chart that ranks subject matter as hashtags by importance. Using this visual dashboard, users can explore insights, navigating supporting facts for validation and context, narrowing focus by filters (e.g. keywords) for specific themes. 247 Insights enables users to absorb more, learn faster,  and execute faster. This offering is now available as a Software-as-a-Service (SaaS) subscription service at www.sitefocus.com/signup.

To see how it works, visit www.sitefocus.com or drop us a line!

[sitefocus-blog1]

What does #Elections2016  Mean for Business and Big Data?

One week today, Donald Trump won the election and single-handedly rocked the foundations of data-driven analysis. Understanding “what went wrong” is critical to understanding the implication on business. Some of the questions this post tries to address are:

-       What went wrong?

-       Were the elections a rejection of Data?

-       What could we have done better?

 The Rise of Data – and what we got wrong

In going “mainstream”, data changed the way we do business. From movie recommendations to credit scores and auto loans, the advancement of big data technologies enabled organizations of all sizes to collect, analyze, and transform data into opportunities for monetization across all spectrums.

We saw data leveraged with increasing effectiveness and efficiency in the 2008 and 2012 elections with Nate Silver’s now-famous execution leading the charge. Up until results came in, most pundits figured the 2016 elections to be no different. Although the data-driven narrative was perfect, to borrow from Minority Report “…if there’s a flaw, it’s human. It always is.”

Crowd sourcing the post-mortem has revealed many theories on why so many polls & models got it wrong. Setting those aside, I believe #Elections2016 validated a major flaw in the modeling process: the data was incomplete; models completely left out qualitative unstructured data.

 Huh?

Unstructured/qualitative data (text) is the foil to structured/quantitative data (numbers). While numbers give us the big picture, unstructured data give us the context & details behind a figure or statistic. Stepping into the business world, our unstructured data can be internal (emails, reports, memos, product specifications, meeting notes) or external (customer support forums, call transcripts, emails, user reviews, social media conversations).

 Were the elections a rejection of Data?

If anything, #Elections2016 reaffirmed our need for more data – lots of it – big and small.

Models & polls attempted to simulate who people would vote for, but in doing so completely forgot about the most important variable in the equation: the humans! How many models considered the sentiment of the demographic? How many polls could divine what made a target segment choose one candidate over another (and why)?  

Case-in-point: The Washington Post ran a piece this week focusing on why readers voted for the president-elect.

Try finding insights like this in a quantitative model. As of the time I write this post, WaPo has received 1,200 responses. One of the theories posit that people lied about who they wanted to vote for. Could it be that people simply didn’t have the means to express their full sentiment? While binary questions are easy to graph, they’re hard to qualify.

If a candidate can key in on the pulse of a demographic – know what resonates and what doesn’t – they can present a more convincing platform or draw light to a misunderstood policy position. It’s not rocket science.

It doesn’t take a business 101 class to know we should listen to our customers because it gives us insight into things a business is doing right…or wrong. One of the reasons why social media has become an integral part of today’s modern business operations is because it exists as an excellent, non-intrusive channel for customer outreach. It fosters communication. Got a problem with that product? DM the company support Twitter account.

  What could we have done better?

The Rise of Data has been told largely from a quantitative narrative. #Elections2016 have shown us it’s time to rethink that approach – or at least the balance of the variables in play. When a human element is in the equation (with the power to vote with their voice, feet, or wallet), it behooves an organization to adopt a hybrid approach that combines qualitative and quantitative analysis as the situation demands it.

Telescoping this to business, you can see the benefit of unstructured data at virtually every level of an organization. From talent management to sales to customer support – unstructured data insights enriches our ability to understand our stakeholders – be they internal or external – and amplifies our collective ability to execute against business strategies and objectives.

If it’s so amazing, why doesn’t everyone hop on the unstructured bandwagon?

It’s not easy. For starters it can be time-consuming. How long does it take to read and analyze 10 comments? Alright, now how about WaPo’s 1,200 comments? Let’s scale that out to a township, a city, a state? There’s only so much time in the day and hiring more analysts can get expensive quite quick. Using technology can help, but implementation can be complicated and mileage will vary. There’s a significant resource investment to get a software system to understand thousands of idioms and expressions to parse the meaning behind a line of text. Even then, the results aren’t always ‘actionable’. The time to value is a huge reason why so many firms still turn to numeric rankings or binary Q&A – hence a vicious cycle. Still, there’s a silver lining in the clouds. New firms and technologies are emerging with innovations that weave unstructured data insights into approachable and accessible solutions. From new R&D startups like SiteFocus to more visible platforms like Siri and Alexa, if there’s one thing that’s certain, unstructured data will have a huge role to play in the near future.

[sitefocus-blog1]

Small Data, Big Insights

Big Insights, Small Footprint

When most people think “big data”, they think of data in terms of gigabytes, terabytes, and petabytes. When they think of “insights”, they think of racks of servers processing/crunching away. These images are the message that conventional analytical solutions have preached since the big data revolution began. 

With CIF, SiteFocus shows that

1. It’s possible to acquire big insights from small data

2. It’s possible to get big data from a small footprint

Many Small and Medium size Enterprises (SMEs) have yet to reap the benefit of “big data” because they don’t have that much data to begin with. A 2015 survey found that “80% of businesses are dealing with less than two petabytes of data, and 60% say they just want to do better analytics with the data they already have.” Conventional text analytics require significant upfront investment (personnel, hardware acquisition, software integration, time to learn/train) that make it hard to realize value.

CIF’s unique approach enables enterprise of all sizes to realize the shortest time to value. Analytics are designed to be accessible and approachable. In other words, business users of all backgrounds can quickly deploy/setup CIF and extract actionable insights from data to take action and measure impact over time. 

All this is made possible because of CIF’s unique approach. We recognized the barriers to adoption facing the industry and designed CIF to break through them. 

CIF is domain agnostic -  unstructured data analytics can be deployed to any business domain without data training or supervised learning. 

Small data, big insights - CIF can extract insights irregardless of dataset size - from a single document to hundreds of millions. 

Insights are approachable - anyone can look at a CIF analysis and take away insight

Use-cases are context-driven - CIF can be deployed to every level of an Enterprise to optimize operations and identify operational risk

Analytics are accessible - turnkey deployment eliminates the headache typically associated with deploying enterprise analytic solutions

This approach is made possible by disruptive high-density architecture that enables CIF to quickly scale up or down as required. For more information or to see it in action, check out our website: www.sitefocus.com

[sitefocus-blog1]

Presidential Debates Retrospective: a Time Series

With the last presidential debate in the books, we wanted to close with a retrospective. A time series is a great way to put disparate information into a narrative. 

Below you’ll see snapshots of the sentiment quadrant’s transposition of debate message by resonance. As a general rule of thumb, elements spatially mapped on the left (NW, SW) quadrants resonated negatively while elements on the right (NE, SE) resonated positively. CIF’s analyses are context-driven. It’s use-case changes based on (1) data that’s loaded into the platform and (2) the analytical objective.

CIF deconstructed each “response” into thematic trends. Remember, the hashtag/trend is only the messenger. Clicking through on each of the trends opens up a table of supporting facts - excerpts pulled from the candidates debate response to a question. 

First up, we’ve got Donald Trump’s first debate performance on 9/26

followed by debate #2 on 10/9...

and his last debate of the series yesterday (10/19)...

The time series offers interesting perspective into the message that Trump has been trying to communicate at each of the debate performances

While the questions can dictate the subject matter from a high-level, the debates have shown that candidates have a way of working their message in

Now for for Hillary Clinton’s retrospective, starting with the first debate on 9/26...

and 10/9

...and the grand finale on 10/19!

By and large, the analysis of Hillary Clinton’s messaging was the more diverse/balanced of the two presidential candidates

This is evident in the spread of HRC’s responses and the subj. matter covered (national security, healthcare, foreign policy, economy) and confirmed with a deep-dive into the excerpts of the responses

While the debates weren’t designed to host detailed policy discussion, they’re one of the most effective ways for a candidate to paint a picture of who they are & what their campaign represents

[sitefocus-blog1]

Presidential Elections

Transcripts are in!  Come November 8, we’ll know who “won” the debates! In the meantime, take a look at an analysis of debate responses by candidate below!

Our team followed the same procedure from previous debates. The entire debate transcript is first split up by speaker and then loaded as separate data topics in CIF for instant analysis.

[sitefocus-blog1]

Presidential Debates: Round 2...FIGHT!

EDIT: correcting for analysis link to Debate 1

Tonight’s debates were lively to say the least. The most tweeted debate of 2016, tonight’s 90-minute debate created more than 17-million debate-related tweets.

The real question is, what were people talking about? What resonated with the American public? 

After getting the full transcript, we ran it into CIF and the results are in! From our initial review, it looks like Donald Trump narrowly “won” - his messages resonated most strongly with the simulated American majority. High-level reports below. 

Donald Trump

Hillary Clinton

Some quick points below:

Hillary Clinton has a surprising dearth in positive elements compared to the first debate’s message analysis

“American” and “Americans” - after reviewing supporting facts - should be collapsed / considered as one comprehensive element

[sitefocus-blog1]

Predictive Analytics on  #VPDebate Part 1

Edited for Links to #PresidentialDebate Parts 1 & 2.

As part of our on-going series on the 2016 Elections, tonight SiteFocus analyzed the Vice Presidential Debate (one and only!). Remember to tune-in tomorrow for our 24h compare & contrast!

Tonight’s debate between the two presidential candidates was certainly different in structure and tone from last week’s debate. How did the substance fare? 

As the team watched on, some of us at SiteFocus ran a transcript of the debate into Communications in Focus (CIF) - our unstructured data analytics platform - to see what insight we’d glean from one-liners we’ve seen tonight. 

How’d we do it?

Click on the hashtag to deep-dive context and details

CIF is a fully automated analytics platform that does not consider political affiliation.

CIF does not use pre-fed parameters nor does it use supervised learning on problem domains

CIF analyzed the raw debate transcript from the Washington Post

Using multi-factor sentiment analytics and technology that emulates a focus group’s objectivity, CIF predicted and validated winning and losing candidate messages

CIF discovers and maps results as hashtags on a quadrant graph - “Sentiment Insights”

Winning and losing hashtags are spatially gravitated on the quadrant graph

CIF provides supporting facts driving each hashtag

To the verdict!

Splitting up the responses by respondent (Kaine, Pence respectively), we’ve run it up onto our Sentiment Quadrant below:

Kaine Screenshot Analysis

Pence Screenshot Analysis

The Takeaway

Following Mike Pence’s response, CIF concluded that messages in the left quadrant failed ( #security, #policy, #spending, #shooting, #taxes). While messages touching on #criminal, #economy, #reform, and #university resonated well.

A quick excerpt on #spending below

Pence: Look, what -- what you all just heard out there is more taxes, $2 trillion in more spending, more deficits, more debt, more government. And if you think that s all working, then you look at the other side of the table. I mean, the truth of the matter is, the policies of this administration, which Hillary Clinton and Senator Kaine want to continue, have run this economy into a ditch. We re in the...

Taking a quick look at Tim Kaine’s Sentiment Insights graph, CIF believes messages on #school, #taxes, #security, #economy, #wage resonated negatively while subj./themes on #iraq, #women, #hillary, #clinton, #wife, #troops resonated positively w/ a simulated American public. A quick excerpt below on #security 

Kaine we re going to protect Social Security, which is one of the greatest programs that the American government has ever done. It happened at a time when you would work your whole life, your whole life, raising your kids, working, being a Little League coach or a Sunday school teacher, and then you would retire into poverty. And Social Security has enabled people to retire with dignity and overwhelmingly not be in poverty.

Looking Deeper

It’s worth mentioning again that CIF identified hashtag/trends. CIF aggregates the supporting facts (extracts from the transcript) supporting each of the flagged hashtags. If elements are on the left, they’re resonating negatively. If they’re on the right, they’re resonating positively.

Lack of Substance: messages that resonated negatively were usually heavy with subjective attacks and light on actual arguments/policy - case-in-point: discussion on #debt, #taxes, and (social) #security.

Tag-Team: Kaine was a definitive winner, praising Hillary with strong substantial messages

Style v. Substance: Strictly by content/substance, CIF appears to suggest that Kaine “won” (if you can call it “winning”).

Staying on Topic: both candidates had trouble answering the actual questions (e.g. questions on homegrown terrorism). Poor answers are reflected very effectively by the lack of resonance in the CIF analysis results set.

In Conclusion

All in all, not the most substantial debate -- as CIF can attest. The team’s ready to log this one into the books with a 24hr-news-cycle-later comparison to see how the media has weighed in. Going on a limb - without CIF - to predict that everyone is looking forward to a more lively debate with Donald & Hillary. 

Curious to learn more? For full access to an interactive reporting account, drop us a line at [email protected]

Where there’s data, there’s a story to tell

Unstructured data is the voice of communication. Comments, emails, reviews, memos, documents are how people and organizations communicate. Being able to understand how people receive and distill a message is critical to all facets of life - whether it’s on the homefront, in the company, or in society at large. Unstructured Data can help companies better understand their customers, build better employee relationships, monitor risk, or provide critical insights into the competition, but getting insights isn't easy. While text analytics isn't new, emerging technologies like CIF can finally help people and organizations cut through unstructured data with the same efficiency and efficacy as we do with numbers.

For more, check out SiteFocus today!

[sitefocus-blog1]

Unstructured Data Analytics show Deutsche Bank’s crisis-in-the-making started 12+ months ago

Communication in Focus (CIF) is an unstructured textual analytics cloud platform. One of CIF’s key features is its ability to analyze massive amounts of textual data, identify critical issues by context, and extract supporting facts driving trends. Since 2014, CIF has analyzed and monitored global news from major media channels.

While the week’s headlines were dominated with Congressional hearings on Wells Fargo’s cross-selling scandal and the Presidential Debates, we’ve turned to CIF to monitor and track the development of the Deutsche Bank crisis.

This post walks through a summary of our findings - insights driven by CIF’s unstructured data analytics as an alternative to traditional quantitative analysis which abounds in the blogosphere.

Where Structured Data analytics mostly deals with processing data from records and fields commonly found in systems of record, Unstructured Data analytics is about processing textual data using text analysis such as Natural Language Processing (NLP) in conjunction with Machine Learning models, contexts and semantic inferences.

Procedural Background

3 million data articles strong and growing, CIF processes, analyzes, and stores analyses in real-time using unstructured data stores - our Data Galaxy. Using CIF, we created real-time data streams for any and all articles relating to Deutsche Bank. These streams have created several datasets, breaking down results on a monthly basis.

CIF processes the results for sentiment, identifying critical issues, and extracting supporting facts. CIF represents these critical issues using hashtags that are tagged by sentiment (positive, negative, common). These hashtags are then trended and spatially plotted on a quadrant graph. The quadrant graph - Sentiment Insights - is made up of 4 principal sectors:

- The Southeast (SE) quadrant represents extreme positive sentiments

- The Southwest (SW) quadrant represents extreme negative sentiments

- The Northwest (NW) quadrant represents common hashtags with extreme negative sentiment bias that are shared with extreme positive sentiment

- The Northeast (NE) quadrant represents common hashtags with extreme positive sentiment bias that are shared with extreme negative sentiment

CIF uses the quadrant metaphor to make unstructured insights approachable and actionable. Results are charted by month. Each hashtag (or node) contains supporting facts (reasons driving CIF’s assessment). Clicking through the hashtag will open a new window and provide users with a chronological index of supporting facts and source. Clicking on each supporting fact opens a sentiment snapshot - a detailed article-centric analysis.

The Goodies:

Using public domain content, CIF’s data streams helped us to deconstruct the Deutsche Bank story with insights into the following:

With the German Government poised to help Deutsche Bank with this crisis, why is investor confidence diminishing?

When were the earliest tell-tale signs of this crisis first visible?

What are the causes leading to this crisis?

What are the side-effects of this crisis?

What should be expected in the next term?

We have included the following sentiment quadrant charts to depict the evolution of facts leading up to this crises. We have also included a small news excerpts in context to support the hash-tags in these quadrant charts.

Month-by-Month

Excerpt:

#global, #cryan  > Deutsche, hit by a string of scandals and fines, is currently the lowest-ranking globalinvestment bank when measured by its price to book ratio of around 0...  > He was hired by Chris Yoshida, who joined Deutsche Bank in 2014 from Morgan Stanley as global head of rates sales...  > (nytimesbusiness)Deutsche Bank to Shrink Investment Banking Unit in Russia ; LONDON Deutsche Bank said on Friday that it was scaling back its investment banking operations in Russia as the bank's new co-chief executive, John Cryan, looks to reshape its business...  > The board has played an active role, replacing Jain this summer with John Cryan and announcing that Fitschen will step down as co-CEO next year...  > Co-Chief Executive Officer John Cryan, 54, who replaced Anshu Jain in July, is pressing ahead with the bank's plan to bolster profitability by reducing expenses and cutting back businesses...

Excerpt:

#cryan  > New Chief Executive John Cryan, who took control of the bank in July with the promise to cut costs, is readying plans to cut roughly 23,000 jobs, or about one quarter of total staff, sources told Reuters last month 6 billion) quarterly loss is Deutsche Bank s largest in at least a decade ; Deutsche Bank AG said trading income rose 7 percent in the third quarter as co-Chief Executive Officer John Cryan's clean-up of the bank's books leads to the largest three-month loss in at least a decade...  > At Deutsche Bank AG, co-CEO John Cryan said on Thursday he will shrink the workforce by about 26,000 people by 2018 Germany's biggest bank, Deutsche Bank announced an overhaul on Wednesday, saying it would sacrifice its 2015 and 2016 dividends as new Chief Executive John Cryan seeks to bolster the bank's capital and retain money to pay for sins of the past...  > (business)Deutsche Bank plan leaves investors puzzled over growth, with workforce to shrink by 26,000 ; For investors, John Cryan's plan to revive profit at Deutsche Bank AG looks too much like the strategy that contributed to his predecessor's departure: thin on details...  > John Cryan, Deutsche Bank AG's co-chief executive officer since July, said on Thursday the next two years will be tough as plans to shed workers and businesses and revamp technology hurt results...  > In detailing the restructuring plan, Deutsche Bank 's new co-chief executive, John Cryan, was in effect bringing a failed experiment to a close...

Excerpt:

#employees, #cryan  > Cryan, who took over as co-CEO of Germany's biggest bank from Anshu Jain in July, elaborated on his philosophy for paying employees Monday -- even taking aim at his own compensation package -- just weeks after broadly warning that staff bonuses will have to reflect the cost of the firm's fines for past misconduct  > To lift profitability, Co-Chief Executive Officer John Cryan said the company would pull out of businesses in 10 countries from Argentina to New Zealand and shrink headcount by about a quarter  > In the words of John Cryan , DB's Co-Chief Executive Officer (from the conference call ):

Excerpt:

#equity, #london, #russian  > Deutsche Bank, which declined to comment on the size of the trades, said it is investigating certain equity trades in Moscow and London, adding the total volume of the transactions under review is "significant"...  > The sale will help boost the German bank's common equity tier 1 capital ratio as of Sept...  > The fraud office announced in November that it expected to bring conspiracy to defraud charges against 10 current and former employees of Barclays and Deutsche Bank when they make their first court appearance at Westminster Magistrates' Court in London in January... > Deutsche Bank AG has identified as much as $4 billion in suspicious transactions related to its Russian operations, in addition to $6 billion in so-called mirror trades it is examining...  > (business)More details: Deutsche Bank flags $10 billion of trades that may not have been vetted for money laundering ; Deutsche Bank AG has identified as much as $4 billion in suspicious transactions related to its Russian operations, in addition to $6 billion in so-called mirror trades it is examining, said several people with knowledge of the bank's review of the matter...

Excerpt:

#china, #profit, #cryan  > While Deutsche Bank AG sees a minor retreat in volatility, it says more turmoil may come from China and rising U...  > Like other investment banks, Deutsche Bank struggled with near-zero interest rates, a slump in oil prices and investor caution due to worries about slowing growth in China, but analysts have said Deutsche's revenue performance appeared weaker than that of its U...  > Insider trading case looks at Deutsche Bank emails: 'This looks like turning into a horror show' - Four men accused of insider trading split profits of almost 1 million ($1...  > (business)Deutsche Bank drops by most in 5 months as concerns mount over revenue ; Deutsche Bank AG, which runs Europe's largest investment bank, slumped the most in five months in Frankfurt trading after saying the securities unit drove a decline in fourth-quarter revenue, casting doubt on the company's ability to meet profit targets...  > Cryan is shrinking assets and cutting costs to boost capital levels and profitability to reverse a share slump that has made Deutsche Bank the worst-valued global bank...  > Deutsche Bank's securitization business has faced additional pressures internally, including plans by Co-Chief Executive Officer John Cryan to cut severely the lender's most capital-intensive businesses...  > In his place came John Cryan, a cerebral Englishman and, to some insiders, an unlikely choice to restore Deutsche Bank to its former glory...

Excerpt:

#stocks, #credit  > The cost of protecting Deutsche Bank's debt against default has more than doubled this year, while its stock trades at about one-third of the company's liquidation value...   > Unfortunately, Deutsche Bank credit default swaps are now telling us that there is deep turmoil at the bank and that a complete implosion may be imminent...

Excerpt:

#global  > Deutsche Bank for example said last year that it would cut the number of clients that itsglobal markets and corporate and investment-banking units deal with in half...

Excerpt:

#investor  > The chief executive officers of Deutsche Bank AG and Credit Suisse Group AG are squeezed between investors pushing them to boost profitability and employees demoralized by job cuts, lower pay and their bosses' blunt criticism

Excerpt

#dixon  > A $541 million loss haunts Deutsche Bank and its former trader Troy Dixon ; Troy Dixon had hit the big time: he'd gone, as they say in his old neighborhood, from Hollis to Hollywood...

Excerpt:

#debt  > Investors said Deutsche Bank should have warned them in offering materials how it had taken on significant exposure to subprime markets through residential mortgage-backed securities and collateralized debt obligations...

Excerpt:

#global, #debt  > The bank suffered another massive blow after the International Monetary Fund (IMF) branded Deutsche Bank one of the riskiest banks globally...  > Investors said Deutsche Bank should have warned them in offering materials how it had taken on significant exposure to subprime markets through residential mortgage-backed securities and collateralized debt obligations...

Excerpt:

#banks  > "It really is a disaster with these banks...

Excerpt:

#trading, #settlement  > Justice Department is seeking a $14 billion civil settlement with Deutsche Bank over the German financial institution's alleged role in artificially propping up the...  > The German lender is now barely worth more than the $14 billion settlement...  > Deutsche Bank also faces inquiries into legal issues including precious metals tradingand billions of dollars in transfers out of Russia...  > Amid mounting concern about Deutsche Bank AG’s ability to withstand pending legal penalties, about 10 hedge funds that do business with the German lender have moved to reduce their financial exposure. The shares slumped...

This is just the tip of the iceberg...

To learn more, visit www.sitefocus.com!  Curious to explore the full breakdown? Wonder what kind of insights you can telescope to your business? Drop us a line at [email protected]

[sitefocus-blog1]

Part 2: Automated analytics predicts winning and losing issues of the Presidential Debates

Earlier today, we published predictions made by our Communications in Focus (CIF) platform. These predictions were the result of an automated analysis run on the Clinton-Trump debate this past Monday. CIF analyzed candidate messages and predicted which themes would have most resonated with a simulated majority of the American public. 

In recent years, many organizations have engaged in social media analytics (e.g. tracking Twitter, Facebook mentions) or standard sentiment analysis (tracking & counting emotional responses/reactions). While insightful, we couldn’t help but wonder:

are these numbers ‘good’ or ‘bad’?

what problems do these data points try to answer/describe? 

To answer these questions, this experiment took a step in a decidedly different direction. Instead of sampling responses from social media, we relied on CIF’s core engine and its predictive functionality to be our crystal ball.

One of CIF’s key features is its novel ability to emulate focus group objectivity to find the what and the why from any text it reads. It does this, in an industry first, without using *any* supervised training or predefined business domain parameters. 

Putting this into perspective, it’s like having a ‘universal consultant’ that can  - on a moment’s notice - take on engagement from any industry, walk into any office, read business documents like a seasoned industry analyst without any preparation. It also figures out what’s important, why, spatially map its priority, and identify supporting evidence to justify it’s analysis.

So, for good measure and validation, we had CIF analyze the 24h news cycle immediately proceeding the debate. The results are pretty awesome!

CIF analyzed approximately 1,000 news stories from various media sources that were published in the 24h news cycle (starting with the debate)

Ignoring names and common words (e.g. used in the subject area), CIF’s original predictions on messages that resonated w/ the simulated majority of the American public aligned with 7 out of 9 trending issues.

FYI: color coordinated against CIF’s predictions: Red = Donald Trump, Blue = Hillary Clinton, Yellow = common words

Sources span major media publications such as Bloomberg, YahooFinance, CNN, BBC, Reuters, ABC, the New York Times, CBS, the Guardian, Washington Post and more.

So...how is this ‘validation’? 

Media coverage of the debates are meant to cover the 5Ws (who, what, where, when, why). Their main objective is to inform their respective audience on the things that matter - the key themes that resonate with their base. CIF does not use any supervised training, pre-defined templates or business domain parameters. This exercise validates the predictive analytics powering the CIF platform. 

Future blog posts will show how CIF can be easily deployed across various business domains. Stay tuned for more. Until next time, tally-ho!

[sitefocus-blog1]

Part 1: Automated analytics predicts winning and losing issues of the Presidential Debates

The first presidential debate of the 2016 election cycle has come and gone.  Sure, this might’ve been the “most tweeted debate” ever, but what resonated? SiteFocus uses predictive insights to take a look.

To do this, SiteFocus deployed Communications in Focus (CIF) - a disruptive cloud analytics platform for unstructured data that launched this Fall. Using machine learning, big data, and novel technology, CIF analyzed candidate responses to predict which of the candidates’ messages resonated most with the critical majority and why.

Elections are about winning the hearts and minds. Every email campaign, sound bite, and stump speech is about selling a message to an audience. With the ability to reach people from all walks of life, the Presidential Debates are a huge opportunity to understand the impact of each candidate’s message on the U.S. electorate.

SiteFocus analyzed the debate transcript to predict which debate messages resonated most and why - answers you can’t find in social media mentions and traditional sentiment analytics.

How’d we do it?

Click on the hashtag to deep-dive context and details

CIF is a fully automated analytics platform that does not consider political affiliation.

CIF does not use pre-fed parameters nor does it use supervised learning on problem domains

CIF analyzed the raw debate transcript from the Washington Post

Using multi-factor sentiment analytics and technology that emulates a focus group’s objectivity, CIF predicted and validated winning and losing candidate messages

CIF discovers and maps results as hashtags on a quadrant graph - “Sentiment Insights”

Winning and losing hashtags are spatially gravitated on the quadrant graph

CIF provides supporting facts driving each hashtag

So what’s the verdict?

Take a look below. Reports are broken down by candidate.

Donald Trump

Hillary Clinton

The Takeaway

Following Donald Trump’s Sentiment Insights graph, CIF concluded that messages in the left quadrant failed ( #audit, #mexico, #mayor, #politician). CIF tells us why.

Taking a quick look at #audit, CIF will create a report of supporting facts why this message would not resonate with the critical majority. A quick excerpt provided below:

Trump: I will release my tax returns. And that’s against -- my lawyers, they say, Don’t do it. I will tell you this. No -- in fact, watching shows, they’re reading the papers. Almost every lawyer says, you don’t release your returns until the audit’s complete. When the audit’s complete, I’ll do it. But I would go against them if she releases her e-mails.

Taking a quick look at Hillary Clinton’s Sentiment Insights graph, CIF believes messages on #american, #taxes, #putin, #muslim, #iraq, and #terrorism failed. These hashtags don’t seem that bad at first blush. WIth CIF’s aforementioned supporting facts model, we can take a quick deep-dive into the “why”.

Clinton: And the Russians need to understand that. I think they’ve been treating it as almost a probing, how far would we go, how much would we do. And that s why I was so -- I was so shocked when Donald publicly invited Putin to hack into Americans. That is just unacceptable. It s one of the reasons why 50 national security officials who served in Republican information -- in administrations...

For full access to an interactive reporting account, drop us a line at [email protected]

Where there’s data, there’s a story to tell

Unstructured data is the voice of communication. Comments, emails, reviews, memos, documents are how people and organizations communicate. Being able to understand how people receive and distill a message is critical to all facets of life - whether it’s on the homefront, in the company, or in society at large. Unstructured Data can help companies better understand their customers, build better employee relationships, monitor risk, or provide critical insights into the competition, but getting insights isn't easy. While text analytics isn't new, emerging technologies like CIF can finally help people and organizations cut through unstructured data with the same efficiency and efficacy as we do with numbers.

For more, check out SiteFocus today! 

[sitefocus-blog1]

Get your Zen on

Absolutely stoked to announce that our integration with Zendesk is officially live!

So what’s it do?

Instantly analyze *every* ticket in your desk in seconds

Identify high-priority issues by sentiment

Highlight key sentences driving sentiment per ticket

Monitor & track client sentiment over time - who needs a little triage?

Account services & support is fire fighting. What happens when you’ve only got so many firefighters?

The NOW Every customer’s request is ‘high priority’. SiteFocus reads through your tickets and renders its analytics into visual guides on trouble spots - the real burning fires that require urgent response. 

The BIG PICTURE Is there a bigger problem? Is the client relationship in need of immediate triage? SiteFocus analytics enable Enterprise to visualize client sentiment over time. Identify at-risk accounts and take preventive action before it’s too late. 

To see it on the marketplace, check it out here! To take a test drive, check it out here

[sitefocus-blog1]

Disney and the Importance of Unstructured User Feedback in the Age of Big Data

Analytics: - the discovery, interpretation, and communication of meaningful patterns in data.

Whether you’re looking to optimize your workflow or out-strategize the competition, analytics can add value to almost every business function. As Big Data and analytics tools have matured, organizations are devising new ways to leverage data to increase their competitive advantage and identify new business opportunities. While Enterprise have benefited greatly from Big Data driven quantitative analysis, a huge dimension of insight has been left untouched - unstructured user feedback.

Analytics are the means to a singular purpose: to help Enterprise (better) understand their customer/user. Understanding the user enables product teams to design products & services that appeal to their core base and customers-to-be -- folks who are sitting on the fence or using a competing service. Customer-first organizations that iterate for their user base will build proven winners. Products that don't, quickly find themselves on the wrong side of history. The consequences can be unforgiving.

Nowhere has this paradigm been more visible, as of late, than at Disney.

In recent years, Disney has steadily churned out an unprecedented string of box office gold. As of 2015 alone, Disney served up not one, but two multi-billion dollar movies. Already 2016 is shaping up to be another banner year with The Jungle Book, Zootopia, and Captain America: Civil War. Disney is cranking out movies that are appealing to people and building on them with appealing iterations (Captain America, Avengers: Age of Ultron, and soon-to-come Finding Dory). People are responding in kind, voting with their feet (and wallets).

Yet, in a footnote to its second-quarter earnings, the Mouse announced they were pulling the plug on the Disney Infinity franchise and getting out of the video-game business altogether. (Disney Infinity combines onscreen MMOPORG with collectible toys that can be ‘brought to life’ in-game)

It’s no secret the MMO space is worth a pretty penny. Time has proven that with hits from Activision Blizzard (World of WarCraft), SuperCell (Clash of Clans, Boom Beach), and the slew of copycats that have followed since. Where did Disney go wrong?

All that glitters is not gold

To be sure, it is not the lack of material. Disney assets as of late have been blessed by the Midas touch with big hits like Star Wars, Toy Story, and a pantheon of successful Marvel superheroes (backed by highly visible box office gold). Indeed, Disney Infinity helped lift Disney’s interactive unit to its first annual profit of $116 million in 2014. Collectible, highly sought-after and recognizable assets that can ‘come to life’ - what’s there not to love?

Yet, a quick look at the first few pages of user reviews paint a very different picture.

Disney’s CEO, Bob Iger, argued that making and distributing toys and video-game equipment proved costly and the risk of getting stuck with unsold inventory was high.

A Missed Opportunity

Where does the rift come from? Customer feedback enables an organization to see which problems are most visceral and pressing to their user base, but only if you can get to it in time. Feedback encapsulates user sentiment and, like the wick on a bomb, has a time to live (TTL) before users pull the trigger (terminate service). In Disney’s case, people just weren’t buying.

If Disney product and business teams were able to qualify and quantify user feedback and sentiment, they might have been able to identify critical features enhancements (e.g. highly desired usability like localized multiplayer and co-op, offline play) and business strategies (e.g. rethinking embedded pay-walls in the demo) necessary to retain and engage an increasingly disenfranchised user base.

The Devil is in the Details

“Just read the reviews,” you might say. While the solution sounds simple, ask yourself how easy is it to do your job and find time to read, analyze, process, categorize, and distill insight from 1,000 reviews? 10,000?

What if…?

What then might have happened if business teams were running SiteFocus analytics? Well, they might have been able to systematically get the big picture without losing sight of the details. Business teams might have parsed user feedback into real-time streaming trends and easily juxtaposed user sentiment against quantitative data points they already had access to (e.g. purchases, activations, in-game engagement metrics). By adding the user’s voice back into the equation, they might have been able to more quickly isolate, identify, and qualify obstacles to adoption before they got out of hand. This in turn would have enabled business and product teams to focus their energies on solving problematic features while nurturing a potential game-changing revenue channel.

Thinking outside the box and beyond

Looking for validation, business teams could have pointed the SiteFocus engine outwards - processing thousands of public domain content (PDC) such as user reviews and forums from competing products by Lego and Activision. Using SiteFocus, business users could have processed this unstructured data into sentiment trends of the competition’s user base. This real-time market intelligence could’ve had implications for product (What’s working? What’s not?) and sales (prospective defectors lying in wait) teams.

In 2005, Google Analytics was launched. It had a profound change on the way people consume and produce content on the Internet. Using quantitative data, Google Analytics made it possible for website publishers to easily understand their user base, divining insight from traffic patterns and interaction metrics.

With technology finally reaching maturity, analytics engines like SiteFocus are finally here to help change the way people build for and service their customers.

[sitefocus-blog1]

The SiteFocus Manifesto

sentiment

- a view of or attitude toward a situation or event; an opinion - a feeling or emotion

Sentiment is woven into the fabric of our day-to-day lives. Whether it’s the people we interact with, the restaurants we frequent, the clothes we buy, or the services we subscribe to, sentiment is a powerful conduit of our taste and preferences - the mental connection that causes us to ‘pull the trigger’ - or in some cases refrain.

Knowledge is Power

Businesses have been trying to understand their customers since the dawn of commerce. The more you know, the better (products & services) you sell.  The proliferation of the Web 1.0 and web analytics empowered enterprises to to see their customers in even greater detail. Anything and everything could be tracked - each link visited, every click logged. The trail only went cold when users left their homes. That changed with the rise of Web 2.0, mobile, social media, and Internet-enabled mobile devices. Putting the Internet into your pocket quickly ushered a new age: the Dawn of Personalization. The sky was the limit. Big data made it possible to create never-ending profiles. Enterprise could track and study transactional history and behavior ad infinitum.

The Missing Puzzle Piece

While this is fine and dandy, enterprise have left a huge dimension unsurfaced. A mother lode of insight lies buried in plain sight: unstructured user feedback. Technology has made it easier to acquire and store quantitative data points (clicks, conversions, unique, etc.), but nothing’s really moved the dial when it comes to unstructured data. Sentiment analysis (#tbt: “text mining”) and natural language processing have been around for decades. Yet, many enterprise analyses still heavily skew in favor of qualitative data points (Likert scale, NPS, etc.) because it’s a struggle to process unstructured feedback at scale.

After all, how do you graph an email? How do you visualize an IM log?

Say you launched a product and you’re running surveys to a target demographic. You’ve got some Likert scale questions mixed in with a few open-ended prompts (free-form text box - 1,000 characters). You’re staring down the barrel of 5,000 responses. How do you ‘get the big picture’? How do you qualify response? Where do you even begin to distill meaningful insight? What customer insights and nuances are lost to the flames because they’ve never been looked at?

In a recent example, the Telecom Regulatory Authority of India (a U.S. equivalent to the FCC) purportedly received 16 million responses to a call for comment on net neutrality. While the TRAI alleged many to be template messages, can you even begin to imagine what that might look like?

There’s only so many hours in a day

Every enterprise is sitting on a gold mine of unstructured data. Mining and refining it into actionable insight takes time and resources. What are the options? Hiring more people to read – let alone process – unstructured data is not a scalable solution. Comprehension and critical analysis skills will vary. Assuming that an enterprise can solve for that, let’s not forget about the time thing. After all, there’s only so many hours in the day to eat, sleep, work, and do everything in-between.

Is it a question of having the right tools? Tapping into text mining technologies is an expensive endeavor. Today’s standard bearers will require extensive data training, significant upfront investment in personnel and hardware to acquire results that don’t easily translate into actionable insight. How much A/B testing do you have to go through to see results?

  A Better Way Forward

SiteFocus pioneers a unique approach that closely mimics the human model of learning and comprehension. SiteFocus Sentiment Insights enables enterprise to realize productivity gains and derive critical insight without sacrificing consistency or quality of analysis.

SiteFocus Sentiment Insights (SSI) is an analytics engine - built specifically for unstructured data - that helps enterprise extract critical insights from any unstructured textual data.

 Context-driven: Our unique approach mirrors the grunt work an analyst does today. SSI automatically reads the entire document to (1) calculate a 5-factor sentiment composite and (2) identify key sentences driving sentiment.

 Objective analysis: Sentiment is subjective by nature. The SSI engine enables an enterprise to compare & contrast the believability of each document using a proprietary “objective analysis.” SiteFocus core technology simulates the critical mass to acquire an objective sentiment measurement – like getting a second opinion from a doctor.

Actionable insight: SSI analyses – when aggregated – create a real-time “sentiment EKG.” Enterprise can get the macro view of user sentiment or deep-dive into detailed reports for micro analysis.

 --and this is just the beginning! With the tools to understand, model, and process unstructured customer data in real-time, Enterprise can fundamentally redefine how they understand the customer experience.

 In the next few blog posts, we’ll take a deeper look into how this approach translates across different problem domains.

[sitefocus-blog1]

Just getting started

Here’s our first post!