Discovering Online Moments with Real Time Analytics

In our digital world where everything occurs and changes instantly, it’s really hard for us to wait, for anything, really. With eCommerce, we don’t join queues anymore (well, maybe just to get the new iPhone first). We skip sections in the videos we watch to get to the meme at the end. We even get impatient when it takes a few seconds to load a webpage. It seems justified for us to expect our analytics to be just as fast, providing results instantly.

Surely you’ve found yourself refreshing the open rate and CTR stats of an email campaign you just launched. Or better yet, you’ve been staring at the real time counter of Google Analytics to see the number of current visitors updating live as a result of a new campaign that was just launched. If these situations sound familiar, then you can probably appreciate the immediacy of fast answers to more complex questions such as:

Which campaign is bringing in the most deposits at this very moment?
Which article has been shared the most in the last 10 minutes?
Which bonus campaign is currently bringing in the most revenue?
Or which version of my landing page is gaining more in the current A/B test?

These type of real time reports, or counters that rely on raw user events flowing in real time, are a perfectly compliment our agile time-series behavioral analytics.

Real-time reports or time-series analysis, when do we need each?

Digital behavior is a series of actions or events, that happen over time. Time-series analysis examines the timeline of these series of actions for a deeper understanding of the customer journey, as well as the reason for the customer behavior. But when there’s a need to look at customer behavior in a shorter period of time, such as hours or minutes, that’s when  real-time reports come in handy.

To enable real time reports, we developed a parallel database (Using the Druid technology) that skips all the foundations of behavioral analytics such as smart user identification and unification, server-side sessionization and behavioral segmentation and makes the data available in real time.  This makes it possible to deliver instant insights as fast as the customer completes the action, enabling a fast reaction to an online scene that is always full of surprises.

Understanding virality and popularity when every minute counts

Digital publishers and broadcasters constantly monitor their content to see which articles or videos are most popular, or have the most views or shares. Popularity measures what happened with your content over time, till now. It doesn’t tell you what happened in the last 10 minutes, which can sometimes paint an entirely different picture, as well as give you clearer insight about the future.

Digital publishers and broadcasters that need to measure the immediate effect of their content need to be able measure virality, or the K-factor which reflects the number of invites and numbers of shares. They need answers to more complex questions than the number of content pageviews and shares, answers to questions such as: How many videos were viewed in the last three hours? Which videos were shared the most? On which channels did readers prefer to share the content? Did paid or free users share the most articles within the specified time frame?

Sometimes, though, a digital publisher or broadcaster needs to simply measure popularity of the content over a short period of time — say the last few minutes or hours.  This real time report from a video publisher shows the number of video views in the last three hours in relation to the number of users currently in the app. With real time insights, publishers can react quickly to increase the promotion for a specific video that has had many shares, views or conversions or change the placement, headlines, or suggested video content to optimize engagement.

This report shows that since there were an average of 1.2 to 1.4 video starts in the last three hours, that means that the suggested related videos for this app are successfully engaging users; i.e. most users are watching more videos than the one they initially came to watch.

Real time analytics

This goes to show that in an online world of constant information and interruption, video is a channel that successfully holds a visitor’s attention.

Measuring the impact of a marketing campaign, instantly

“I know that half of my advertising dollars are wasted … I just don’t know which half.” –John Wanamaker, pioneer of the concept of the department store

Marketing managers in gaming companies are obsessed with being able to monitor the click and open rates of players for the latest email marketing campaign. But what if instead of these traditional metrics, they were receiving more complex reports that told them in real time how many deposits or installs were being made as a result of a particular campaign?

real time analytics

This real time report shows the number of ad clicks in comparison with app installs in the last hour. Since ad clicks increased significantly around 12:39 pm at almost exactly the same amount as the app installations, drilling down further can show great insights. We might see that these players followed a particular series of actions before installing, or came from a particular geographic region, or ad search.

The knowledge of how successful a marketing campaign is (or isn’t) in real time is incredibly valuable. Gaming platforms spend valuable time and resources on these campaigns. Once they know the exact point at which the campaign is no longer effective, they could immediately switch gears to build and deliver more effective campaigns.

Run A/B testing in real time to optimize trading revenue

One of the most popular methods for acquiring new traders or reengaging dormant traders is the offer of a bonus campaign. Bonus campaigns require constant monitoring of trader behavior, however, since certain types of traders may take advantage of the offer without becoming real money traders.

If brokerages had up-to-the-minute reports of how different bonus campaigns affected trader behavior, they could compare different ones to measure their impact on the number and amounts of deposits and optimize their revenue accordingly. They could also pinpoint the exact time that traders were no longer making real-money deposits and stop those ineffective campaigns in real-time.

Learn how the real time reports work Watch this video

Delivering results and change in real time

Cooladata’s ability to store data from the last 24 hours gives you up-to-the minute granularity of your customer behavior. Businesses now have real time insights on product engagement, content virality, and purchasing patterns, as well as a partner portal that enables users to embed real time dashboards and data widgets for each of their customers. Businesses can now tackle challenges as they happen and prevent them from growing bigger as time goes on. Sometimes it’s all about being in the moment.

Want to learn more about real time reports? Schedule your demo
Book a Demo

Already using Cooladata analytics? Contact your customer success manager to enable real time reports.

Download the Guide

A Data Story about a Trip That Begins with an Email

It may surprise you to discover that email campaigns remain the most effective marketing channel for many eCommerce businesses. According to McKinsey, email is 40 times more effective at acquiring new customers than both Facebook and Twitter combined, with customers three times as likely to make a purchase.

In the digital age, when everything is measured, we must ask ourselves: Are we measuring email campaign effectivity enough?

Marketing automation platforms such as Hubspot and Marketo measure email campaigns using traditional email marketing analytics metrics such as open rate, CTR (Click-through Rate), bounce rate, conversion, forwarding rates, and overall ROI. But while these metrics focus on the beginning of the customer journey, they fail to connect the dots to understand the bigger picture.

By integrating email campaign data with all other data sources, including events from web and mobile apps and CRM data, we gain a more complete understanding of the customer journey and deep insights into customer behavior from email campaigns — from the first click to their travel booking.

Finding the sweet spot: Optimizing the impact of each email

A leading business in the travel space concentrates its effort on email campaigns to effectively promote their offerings. Hubspot’s ability to collect traditional email data to measure clicks, open rates, and conversions was only the first step. To go beyond these traditional metrics, they needed to be able to answer the more complex business questions, such as:

– What times and regions are customers most likely to click and open their emails?
– What is the right number of emails to optimize my monthly engagement with customers?
– Which group of customers will potentially bring more value to my business?

Answering these questions require time-series behavioral analysis capabilities that gathers multiple data points into one. A more holistic analysis of the entire customer journey required integrating email campaign data with other data sources.

What is the optimal hour to send emails for best open rate?  

To answer the question of when is the optimal hour of the day in different countries to send an email, we integrated email campaign data from Hubspot with raw historical event data. By querying and filtering data of email campaigns as far as 6 months back, we get the preferred time of the day for customers in each country.

email marketing analytics

Not a difficult question to ask your data, (using the right tools) but quite a useful insight to have for all future campaigns.

Adding the time dimension to email analytics

Examining the customer activities in relation to emails, over time, using time-series analysis, reveals fluctuations in customer actions, or behavior. For instance, examining open and click trends showed that people continue to open/click emails in the days after the email was sent. We learn that emails that were sent in a particular week in the winter for a warm beach vacation received a high response from Europeans even 14 days after the email was sent.

email marketing analyticsThe segment of customers who opened the email 5 days after it was sent could be saved as a segment of “late openers.” This segment could be later targeted with optimally timed follow-up email campaigns.

These are the types of insights that provide great value in creating effective targeting and follow-up campaigns.

How many emails should be sent to customers to yield maximum engagement?

It’s a matter of hitting the sweet spot: too many emails and you’ll get customers to unsubscribe from your email campaigns; too few and you miss out on customer engagement. Does 7-10 emails sound too much? Well, the raw data tells a different story.

email maileting analyticsBy querying email campaign data over time, businesses can see the magic number of emails they should send at any given month to maximize their open and click rates.

A conversion funnel between data points

What is the most popular conversion funnel that starts with an email open? What is the conversion rate from email to booking that occurs during the same day? Answering such questions becomes easy when your analytics gather data from different data points.

With advanced funnels, you’re not limited to predetermined steps or predefined goals. You can easily build any conversion funnel by examining any step of every event or property in the database, including email_open, booking_confirm, or booking value over 10$, regardless of the data source.

email marketing conversion funnel

Such advanced funnel analysis leads to data-driven marketing actions by targeting any user behavior segment created from the funnel. For instance, you could target the particular segment of customers from a certain country who arrived at the payment page in that same session and dropped off at the payment page, the last point in the conversion funnel.

To ensure that only the most valuable customers are being targeted, businesses could focus only on sending certain last-minute promotions to customers who have previously booked a trip 12 hours before their flight.

Email marketing data combined with web and mobile user behavior data

One of the best ways of having an edge over your competition in eCommerce is by better understanding your customer journey and behavior. Traditional email marketing and even marketing automation systems give you some insight, but take you only so far. Businesses who want to understand what happens to their customers after they leave the email ecosystem should consider integrating their existing system with Cooladata to gain these vital insights into their customer behavior over time. All of these insights are available on a single dashboard so you can react quickly to any sudden changes in your customer email engagement. For businesses focusing on customer acquisition, conversion and growth mainly through email, the types of insights above are priceless.

Download the Guide

Want to gain insights from all your data? Learn more about our data integrations.

Book a Demo

Behavioral Segments – The Secret Sauce of Analytics

“The past is never dead. It’s not even past,” said William Faulkner back in 1951. What Faulkner may have meant is that the present is built from moments in the past and it’s our past experiences that influence our present-day actions.

That’s even truer when applied to our digital behavior. When analyzing user behavior, time is of the essence.  We need to look at behavior as a series of user actions over time, starting from their very first online action to the present. Past user behavior reveals patterns that can provide great insights on the present (and even shed light on future predictions, but that’s a post for another time).

In traditional analytics, user statuses are static, single states or events frozen in time. For instance, a user’s demographics, referring campaign, device they use, or even just the initial session duration are all points in the user’s history that don’t give you much of an understanding of the present.
Yet in analytics,  these are how we normally define user segments, profiles and personas to differentiate between users.

We’re thrilled to announce our new advanced behavioral segments, with a segment builder that can transform a group of users that demonstrated a particular behavior into a behavioral segment that can be used in further analysis. For instance, you can now query segments, compare them using either cohort or funnel analysis, or simply filter the entire dashboard according to a single segment.

Learn How Behavioral Segmentation Works
Schedule Demo

Here are just a few examples to captivate your analytical imagination:

Building Segments From Any Identified Behavior

Our segment builder allows you to create segments by combining different user states and their associated behaviors over time. A segment could include active/inactive users during a specific period as well as those who completed a series of specific events in a defined time period. Behavior also includes users whose action had a particular value during a specific time period (e.g their purchase was over a given amount).

Here we’ve taken users who visited an eCommerce site in the last 7 days who also searched and bought an item over $10 in the last 30 days. We’ve then filtered the report to include only users from the United States.

Behavioral analytics segments builder
From here, you can drill down further to analyze the behavior of this segment. You might find, for example, that these users prefer brand-name or luxury items, and you’ll begin to understand why they responded to different marketing campaigns as they did. You’ll also gain insight into their general engagement with the site.

Watch our 3 minutes video on how to build behavioral segments

Create a Segment From a Cohort Report

You can also create a segment directly from a cohort report. Take the retention analysis report below of players from different marketing campaigns of a gaming operator.


In the first group of players, “bonus campaign #1”, players engaged with that campaign during the particular week it was offered. Suddenly, four days later, approximately a quarter of them churned. Why?
We need to analyze past user behavior to understand their present behavior more deeply. Did they only engage when they were offered a bonus in the past? Maybe the bonus campaign gave players a reduced incentive to continue to deposit money, and when these players realized they wouldn’t be receiving any more bonus coins, they left the site. Instead of offering many bonus campaigns to this segment of players to re-engage them, gaming operators might want to try creating bonus campaigns that expire after a certain period and offer them directly to this player segment.

A Tale of Two Funnels: Analyzing Different Segments

You can also compare two segments using cohort or funnel analysis to highlight the differences in their behavior.

For example, you might want to compare the behavior of different segments as they move through the conversion funnel throughout their entire life cycle. When comparing the actions of one-time customers, whales and “others,” we discover that whales have a much lower drop-off rate after both searching and viewing an item, as well as a higher global conversion rate.

To discover even deeper insights, we can drill down further into the behavior of each user segment, for example, by analyzing the exact terms they searched for or limiting the analysis to a more specific time period.

The Past Can Reveal Future User Patterns

The segmentation examples I provided are powerful in that they allow businesses to drill down more deeply to analyze the behavior of that particular segment. Static segmentation doesn’t necessarily provide these insights into future user behavior.

If you want a better understanding of the present, you need to measure user behavior over time. These behavioral segments can be created, saved, and compared to empower you to reveal patterns in past behavior. These patterns can give you insights that drive actions to guide user behavior in the future.

Going above and beyond traditional analytics, behavioral segmentation will allow you to segment behaviors of groups of users who have taken specific actions over time. THIS is the secret sauce that will empower you to gain insight for optimizing your product for future user engagement.

Behavioral segments are available in our starter plan
Start FREE now


When Your Content Isn’t Just Popular – It’s Viral!

As a publisher or content creator, if you had a nickel for every time you heard: “This is going to go viral — I just know it!” well, you’d probably have a lot of spare change.

Whether that content ended up going viral or not each time, you were probably at a loss to explain why.

Most media analysts are still tied to the traditional content performance measurement – which means measuring popularity. What’s popular? Say a major breaking news story in the New York Times gets viewed and shared many times, thus making it a popular article.

The thing is, measuring popularity is not enough. It’s old news by now. What if you want to know the effects of your content now, or at least in the very near future?

For instance, maybe as a publisher, you want to know what articles will be popular for the rest of the day or tomorrow. Or the rest of the week? What if you want to know which segment and what medium to promote these types of articles in order to monetize more effectively?

If a publisher wants to measure a boost of popularity over a short period of time, they’ll want to measure virality. The question is — how?

Virality can be measured through analyzing audience behavior over time. Time-series behavioral analytics enables us to dive deeper, slicing and dicing the data and segmenting the characteristics of both the viral content and the behavioral profile of the audience that made it viral.

The result is a clearer picture of how your audience responds to content over a specific period of time and a greater understanding of the type of content that is likely to be shared in the future.

Download the Guide

Popularity is a Never-ending Series of Snapshots

Successful content is content that attracts an audience. The larger the audience, the more successful the content.

One method of measuring this ability to attract an audience is according to popularity. Content popularity can be measured by the number of views, the average time spent on the article, or the number of views or shares. Unlike virality, however, it isn’t specific to short timeframes.

For example, these are the most popular content pieces from the New York Times in 2015 measured in total combined time readers spent reading them:

The Top New York Times Stories of 2015, by Total Time Spent
The Top New York Times Stories of 2015, by Total Time on Page

That’s all fine if all you want are continuous snapshots of your user journey and don’t care about how long it will take for your content to become popular. But if you want better insights, you’ll have to dig a bit deeper.

Virality Measures the Immediate Popularity of Your Content

If you want to gain a better understanding of the immediate effect of your content on your audience, you’ll want to start measuring virality. Virality is much more than just counting the number of pageviews and shares of your content; it takes into consideration the time that elapses between the publication of that content and the time the content was shared, as well as the clicks, pageviews, shares and conversions that were a direct result of those shares.

Virality is commonly measured using the K-factor. In a nutshell, the K-factor is measured by the number of invites and invite conversions. A K-factor greater than one indicates exponential growth.
The K-factor should be part of any publisher’s dashboard, along with other must-have metrics. The K-factor of each article over time, since it’s important for monetization and for setting the ad price by the best parameter.

Content Performance analytics dashboard
Content Performance Dashboard containing the K-Factor

Since virality is measured in short timeframes, it is best visualized using the cohort report. Like in this example below that shows the percentage of shares over a five-day period after publication. The segment of users who shared the most are those who first came to the site on February 8th, 2016.

Cohort Analysis of Content Performance

If we’d like to learn more about these users who shared the most, we’d have to drill down further to see that paying readers share the most:

Content Performance Cohort Analysis by User

Now, if you want to increase the number of shares among this segment of users, you’ll want to understand the type of content these users are sharing the most:

Virality Content AnalysisThese are the types of categories and articles you’ll want to make sure are suggested content for paid users after they complete articles. (And you’ll probably want to pay more attention to articles with a K-factor higher than 1).

Predicting Virality

Measuring content success and its influence on users is certainly challenging. Different metrics and trends are constantly evolving in the world of digital media, including attempts at predicting whether a particular piece of content will get viral. These types of predictions include various studies into the psychology of sharing and the type of emotional triggers that can increase the probability that a piece of content will be shared.

While these types of studies are important, time series behavioral analytics that measure content performance is a more accurate way to predict content virality. By analyzing past behavior and segmenting our audience  according to user type, category and article preference, we gain a clearer understanding of the path users might take in the future. From there, it’s a quick path to optimization and capitalizing on user attention.

To summarize –
First, when coming to measure content performance, don’t mix popularity and virality  – it’s not the same. Popularity is a metric that stands alone, without any relation to time. In the digital publishing world, where days, hours, and minutes makes a difference in content success, virality is the more meaningful metric.

Secondly, once you’ve analyzed your past data and are able to predict virality, you can focus your marketing efforts accordingly. Understanding timebound past sharing patterns of a specific audience or segment as well as their individual characteristics (their age, country of origin, preferred method of content consumption, etc) gives us the ability to predict future consumption in short timeframes among similar audience segments and to focus marketing in short timeframes to maximize their potential virality.

Download the Guide

Digital Intelligence Rules in the Age of Insights

If you have bought into the concept that we are living in the Age of Big Data, you are wrong (or, at least…too late). In fact, big data as a revolutionary disruption is history.

Big Data is now a commodity: access to it no longer differentiates companies; the ability to create it, compile it, or process it no longer crowns victors.

Instead, the differentiator of our times is not data science, but rather the insights gleaned from those endless datasets. Today is not the Age of Big Data, the Age of Information, or the age of Data Science. Today, we are living in the Age of Insights.

The value of a business can be measured by the significance of the insights it possesses.

Said another way (quite brilliantly, actually) “Businesses are drowning in data but starving for insights.” In our customer-centric times, we cannot comb through vast datasets searching for the proverbial needle in the haystack. Instead, we need Systems of Insights that will empower us to focus on the customer and take astute action to touch people through personalized engagement and customized experiences.

Insights for Digital Disruption

Online business or mobile apps have an endless selection of customer experience or web/mobile analytics tools to choose from. Studying the 2016 Marketing Technology Landscape Supergraphic will leave any marketer or product leader with a thirst for a solution that will analyze that big dataset and provide an insight as to which tool is right for which need.

But even then . . . having the tools (and a growing budget for analytics) is proving ineffective. According to Datawatch, satisfaction with analytics is declining as investments in analytics rise.

We are spending more and getting less.

The underlying reason for the disparity is that common analytics tools relying on sampled data and limited data sources do not provide those action defining insights. Those web and mobile analytics looking at a snapshot of the data, fail in the attempt to make sense of incongruous data points in time, result in fragments of insights not valuable for action.

The ultimate analytics is the System of Insights that fuels the digital business.

Product leaders, marketers and business executives in insights-driven business cannot rely anymore on analytics that aggregate pageviews, clicks, and conversions. The omni-channel dynamic digital business needs more than just analytics – digital intelligence that yields fast meaningful insights to feed the customer engagement.

Time series analysis wins over snapshots

Snapshots look at a segment of data at a moment of time. To truly understand customers’ behavior, we need to add dimensions of context and time. The golden insights on customer behavior must  derive from Time series analysis examining the full sequence of actions over time, analyzing the main goals within the context of events that happened before and after.

In many cases, drilling down into the anomalies in the customer journey (Shown as spikes on the path analysis sunburst) gives us business insights about innovative approaches to grow, and differentiate our business.

Smart Digital Intelligence

Traditional Business intelligence is often not very shrewd. Sounds like an oxymoron, but it is the reality faced by many data scientists and data-driven marketers and product managers. To get insights from BI systems that are not designed for the digital world, a full-time analyst will query a 600 lines of SQL, perform numerous joins of tables, and spend hours waiting for the query results, which, by the time they arrive, might be irrelevant to the fast and dynamic digital business.

Instead, by using the new generation of true digital intelligence solutions, equipped with data analysis tools such as reporting wizards and query shortcuts, everyone can customize reports to get the right insights and fast.

A single source of truth

Complex businesses need the ability to unify data from many sources and channels, inspect it as a whole, and provide insights based on the holistic view. By accommodating server side sessionization and sophisticated unified identities mechanism you achieve cross-platform, cross-device, cross-channel user-centricity. With this granular focus on user behavior, companies gain actionable insights that enable them to optimize the customer experience to thrill their users.

Big Data is only the beginning

Big Data is by no means a bygone, at least as an issue to discuss. Big data is here and every digital business has it. It is here and it is here for good. But the fallacy that Big Data alone will prevail must be put to rest. Big Data, in an of itself is just piles of data that needs to be handled and analyzed to provide insights you need to excel. However, by adding Systems of Insights that can make sense of the Big Data, your business can find its differentiator and carve its path towards dominance.

Schedule a Demo