The Essential Guide To Building Analytic Applications
The Essential Guide To Building Analytic Applications
Analytic Applications
Introduction................................................................................................................................................................................. 3
Chapter 1: Embedded Analytics Is Transforming Product Roadmaps........................................................................... 4
The Data Divide.. ......................................................................................................................................................................... 5
Data-Driven Services Are Taking the Lead. . ................................................................................................................................. 7
Best Practices for Data Storytelling.. ..........................................................................................................................................10
Chapter 2: Designing Dashboards......................................................................................................................................... 13
Dashboards as Guides for Decision Making............................................................................................................................... 14
Speaking the Language of Your Customer.................................................................................................................................18
Designing With a Purpose......................................................................................................................................................... 22
Chapter 3: Product Strategy for Analytic Applications................................................................................................... 25
Articulating Your Product Strategy........................................................................................................................................... 26
The Job Behind the Feature...................................................................................................................................................... 29
Data as an Accelerator.............................................................................................................................................................. 33
Mapping for the Persona.. ......................................................................................................................................................... 36
An Application Fit for Use......................................................................................................................................................... 39
Chapter 4: Predictive Analytics Is the Future. . ....................................................................................................................41
Improving the Value of Analytics.............................................................................................................................................. 42
Decision Support vs. Decision Automation............................................................................................................................... 44
Advanced Capabilities. . ............................................................................................................................................................. 48
Chapter 5: Security Is Crucial...................................................................................................................................................... 51
Preparing for Proper Deployment. . ........................................................................................................................................... 52
Identifying Risks and Vulnerabilities......................................................................................................................................... 54
About Logi Analytics.....................................................................................................................................................................57
INTRODUCTION
Embedding analytics in software presents some unique opportunities—and poses unique challenges—to
application teams. What are best practices when designing the UI and UX of embedded dashboards, reports,
and analytics? What should application teams know about implementing security that works with the rest of
their products? What should product managers keep in mind when adding an analytics project to their roadmap?
We talked to experts about what it takes to build a successful application with analytics at its core.
No matter where you are in your analytics journey, get advice from experts on:
When to add or enhance new features such as analytics
3
CHAPTER 1
As a result, the pace of innovation has vastly accelerated as companies seek the next great feature to differentiate
their software and drive customer value—leading to seismic shifts in analytics. We talked to industry leaders who
discuss emerging capabilities in the analytics space. What trends can we expect to see for 2020 and beyond?
How are data-driven services transforming business models?
4
CHAPTER 1
WAYNE ECKERSON
FOUNDER AND PRINCIPAL CONSULTANT, ECKERSON GROUP
What trends do you see for business The “Data Divide” has replaced the “Digital Divide” in
intelligence (BI) and analytics in 2020 determining winners and losers in the new economy.
and beyond? Organizations that fail to master data and analytics are
now falling behind competitors who do. The winners
use data to proactively capitalize on opportunities and
address problems while losers react after the fact, doing
too little too late until they can no longer compete cost-
effectively with fleet-footed rivals.
5
CHAPTER 1
How is buyer behavior evolving? It has moved up the chain to business executives.
Digital, data, and analytics is becoming an imperative—
it’s “transform or die,” basically. We are seeing the rise of
C-suite titles such as Chief Data Officer, Chief Analytics
Officer, etc.
What are some emerging capabilities We’re seeing AI in everything to facilitate self-service,
in the analytics/BI space? data prep, anomaly detection, root cause analysis,
remediation, and optimization. We are quickly moving
from human intelligence to augmented intelligence to
automated intelligence. Machines will make most of the
decisions in the future.
Should application teams develop a Yes, especially when trying to monetize data or
“product” vs a “project” mindset? analytics. To create a revenue- or customer-facing
product, you need to engage many other parts of the
business: sales, marketing, legal, support, etc. and have
a clear product plan and rollout strategy.
6
CHAPTER 1
DOUG HENSCHEN
VICE PRESIDENT AND PRINCIPAL ANALYST, CONSTELLATION RESEARCH
How is embedded analytics different Business intelligence has typically been in the descriptive
from business intelligence? analytics vein, like a report that is very rearview mirror.
We have moved to the analytics part
of that and we’re getting much more active, exploratory,
and diagnostic. Not just what happened, but why did
it happen? If you add the embedded, that means it’s
available in the context of work. It’s not a separate
analytic interface that you go off to but right there
within the context of your transactional or working
interface. You’re seeing much more concise, granular,
analytic information in the context of something you’re
doing very specifically.
7
CHAPTER 1
What are the benefits of embedding The BI and analytics world of vendors has focused a lot
analytics for the end user? on this idea of democratization. Not everybody wants to
be a data analyst and go off into a separate place—they
want to get their work done. Embedding is a route to
democratization and a route to getting to more business
users because they don’t have to go off to a separate
interface and interpret. They can get much more
concise information that supports the decisions they’re
trying to make and what they’re trying to understand in
the context of their work. They’re getting information
that supports a deeper understanding and supports
driving towards actions.
How important is it that users can act Business people don’t want to sit there and interpret.
immediately based on the insights? They want to know what to do. Embedding is a way to
be more outcome based, not only because you have
more concise information. People are looking to cross
that divide from descriptive and diagnostic into more
predictive and prescriptive. What’s required to get to
predictive is a more concise context.
Where do you see the market currently? There are the pioneers, fast followers, laggards, and
Is there a desire to move more towards cautious adopters. Pioneers and fast followers are
outcome or are people still trying to the software companies and analytics is a big part of
level-set with the rearview mirror? their service offering. Banks are looking at aggregated
information and providing benchmark information to
businesses and credit card companies offer insight
on an aggregated basis on local spending patterns to
businesses like retailers and restaurants.
8
CHAPTER 1
pressured on costs because you’re not just like every other provider.
9
CHAPTER 1
MICO YUK
FOUNDER, BI BRAINZ
Mico is the founder of BI Brainz and the BI Dashboard Formula (BIDF) methodology, where she
has trained thousands globally how to strategically use the power of data visualization to enhance
the decision-making process. Her proprietary BI Dashboard Formula Methodology (BIDF) fuses
the art of data storytelling and visualization to drive users to take action. Mico’s ability to help
companies gain ROI from their business intelligence investments has been sought out by Fortune
500 companies such as Shell, FedEx, Nestle, Qatargas, Ericsson, Procter & Gamble, Kimberly-
Clark, FedEx and more. Rated as the #7 most influential entrepreneur in BI by Onalytics, she also
authored Data Visualization for Dummies (Wiley 2014).
What are the best practices for data When telling the stories, there are three key things
storytelling in the context of to keep in mind:
embedded analytics?
• The message has to be really focused. Identify your
top two avatars and really understand what they are
seeking to do. With embedded analytics, it’s easy to
want to capture a huge net of audience. But with
that, you end up providing something to no one.
10
CHAPTER 1
How should application teams think With embedded analytics, there are more opportunities
about data stories when they’re for testing. A lot of tools coming out today use multiple
embedding dashboards and reports heat maps and confetti maps. They show you exactly
as compared to a more standard BI where people click, where they came from, what’s being
instance? Is it different? used, and how often. Specifically with dashboards or
reports embedded into an interface, that level of detail
is important. If you have analytics, it’s very visual and
if there’s an area in your story that’s not receiving any
attention, you need to make note of it and remove
it. Don’t just let it sit there because you’re building
something out. You can evolve faster because you
don’t necessarily have to check with a user on every
turn. Tracking can be automated and you can make
adjustments and push it out and see how people will
respond. It’s slightly different than the standard BI. When
analytics are embedded, you can track usage and figure
out what’s being used and what’s not being used.
Are there common data storytelling There are a few things. As mentioned earlier, the
mistakes you see application teams number one mistake is building far too many avatars.
making, and what can they do to You have a large audience. With a bank, for instance,
avoid this? you’re talking about thousands and thousands of users.
It’s very tempting to just build a story that captures
everybody so you need to be focused.
11
CHAPTER 1
12
CHAPTER 2
Designing Dashboards
A dashboard serves as a visual display of the most important information needed to achieve an objective.
They transform raw data into critical information, rich stories, and easily understandable data visualizations
for your users, who can comprehend and act on it in a fraction of the time it would otherwise take them.
Dashboard design can mean the difference between users excitedly embracing your product or ignoring it
altogether. Great dashboards lead to richer user experiences and significant return on investment (ROI), while
poorly designed dashboards distract users, suppress adoption, and can even tarnish your project or brand.
13
CHAPTER 2
LAURA KLEIN
PRODUCT MANAGER AND AUTHOR
Laura is a product manager and the author of Build Better Products and UX for Lean Startups.
She has worked as an engineer, UX designer, and product manager at startups and large
companies in Silicon Valley. She blogs about UX, research, product management, and lean
startups at www.UsersKnow.com.
What are the risks of bad UX design for Generally speaking, dashboards should be guides for
embedded dashboards/reports people who want to make decisions. In most products,
in applications? they exist to give users a quick overview of what’s going
on so that they can choose what needs their attention.
For example, a dashboard might tell the head of sales
which sales reps are missing their targets so that they
can make early changes rather than be surprised by bad
numbers at the end of the quarter.
14
CHAPTER 2
How can product teams tell if their No product is good or bad in a vacuum. A dashboard is
dashboard UX is good or bad? What only good if it makes the people using it more effective
are the signs of good/bad UX design without hurting other people. You can tell you’ve got
for analytics/data visualizations? a great dashboard the same way you’d find out if you
had a great product. First, you need to make sure that
you know what the goals are for the dashboard. Then
you need to figure out if the dashboard is helping users
reach those goals. You can do that with qualitative and
quantitative user research.
Are there common mistakes you I see three common mistakes when designing dashboards:
see product teams make when they
design dashboards/reports for their • No follow up. A lot of dashboards are very static
applications? things that don’t give users a next step to take.
When a dashboard is integrated into the rest of the
15
CHAPTER 2
16
CHAPTER 2
What are the most important steps Building useful dashboards requires the same skills
or best practices application teams as building any great product. Spend time observing
can take to improve their analytics users and understanding their ultimate goals. Don’t just
experiences? ask what they want to see. Figure out what they’re trying
to do, and then build things that help them do it.
17
CHAPTER 2
TRISTAN KROMER
INNOVATION COACH AND FOUNDER OF KROMATIC
ANTEO QUIROZ
INNOVATION COACH AND HEAD OF SALES KROMATIC
Tristan Kromer is a lean startup coach and founder of Kromatic. He works with innovation
teams and leaders to create amazing products and build startup ecosystems. Anteo Quiroz is an
innovation coach at Kromatic. He works with innovation teams and leaders to create amazing
products and build startup ecosystems.
What are some of the most common Dashboards have a few common uses:
ways you have seen dashboards used?
• Warnings: When your oil pressure gauge indicates
a dangerous situation, you’d better pull over and
take action. If you don’t, your car will pull over on
your behalf.
18
CHAPTER 2
How can products team ensure that You speak a very specific language than your
customers understand the dashboard? customers—that of your product. Probably very well.
But you also need to speak the language of your
customer who uses the product.
20
CHAPTER 2
How can product teams know which Many high-level business dashboards are relatively
features to include? static since the data isn’t changing moment by moment,
so interaction and usability are often secondary
considerations. Even when it comes to the analytics use
case, there is often a design assumption from engineers
that the analysis is being done by a power user who will
want a ton of features and doesn’t mind struggling with
a ton of dropdowns.
21
CHAPTER 2
MIDORI NEDIGER
PRODUCT DESIGNER AT VENNGAGE
What are some common mistakes One of the most common mistakes I see in data
to avoid when presenting data? visualization is the misuse of color. Because color
has such an impact on the aesthetics of the design,
designers often choose color schemes based on their
own personal preferences or the requirements of the
brand they’re designing for. But color naturally carries
meaning in data visualization—it communicates
information about relationships within the data and
can be used to guide the viewer’s eye within the
design. Color choices should be made with intention
with presenting data, especially when designing
complex dashboards.
22
CHAPTER 2
What are some best practices 1. Design with a purpose: Your first step in any
for dashboard design? dashboard design project should be to clearly define
the purpose of the dashboard, because you’ll never
be able to design a dashboard that is all things for
all people. Each dashboard should be designed for a
specific purpose, and cater to the needs of a specific
end user.
23
CHAPTER 2
What are steps that you can take to Like I mentioned earlier, every dashboard should be
ensure that an analytics dashboard designed with a specific end user in mind. The best
or report is useful and usable? way to ensure the dashboard is useful and usable is to
involve those end users in the design process. Just like
you would when designing a product, talk to your users
to learn about their workflows, their pain points, their
needs, and their wants. Before you start designing, make
sure you have a thorough understanding of how the
dashboard is going to be used. And as soon as you have
a design in mind, show it to those users and get start
getting feedback!
24
CHAPTER 3
Whether you’re adding new analytics capabilities to your product or another feature entirely, you need to
consider many of the same steps and roadblocks. In this chapter, our experts provide detailed advice on
conducting customer interviews, navigating changing user requirements, and distinguishing a need from a want.
When it comes to your product roadmap, it’s all about the journey.
25
CHAPTER 3
JANNA BASTOW
CO-FOUNDER, PRODPAD
Janna is the co-founder of ProdPad, a software that helps product managers plan and deliver
better products. Janna also organizes ProductTank events around the world, including Mind
the Product, a global community of product managers. She likes to inspire great product
conversations by asking: “What problem are you trying to solve?”
How do user requirements change When you first start building a product, you attract a
over time? certain type of user: The early adopter. This type of user
tends to appreciate cutting edge technologies and can
stomach being part of a beta program. If done right,
they make great advocates and will help you immensely
with reporting early bugs and pointing out any areas
where your first versions of the product fall short.
26
CHAPTER 3
have little tolerance for the fast and furious releases that
set you apart in your early MVP days.
Any advice/tips for PMs who want You can stay ahead of changing user requirements by
to stay ahead of changing user building in time for validation and learning. Many teams
requirements, and/or make sure simply focus on building feature after feature, which is
they’re aware of them early? great for your cadence, but really ineffective for making
sure that you’re continually solving the right problems.
The best product teams make room for learning,
and make structured validation a key part of their
process, rather than an afterthought. This way, if user
requirements change, the product managers are clued
into it as soon as possible and can update their roadmap
to reflect the new needs of the market. Validation
at every step of a feature build ensures that the final
feature actually solves the problem it set out to solve,
rather than, quite commonly, diverging from the original
need into a spec of its own making.
27
CHAPTER 3
How can product teams be strategic A good product strategy ties in both top down and
when adding new features and bottom up thinking. Top down means looking at the
planning their roadmaps? company vision and company level objectives, and
outlining the obstacles and problems you’ll need to
tackle along the way. Bottom up means listening to the
market and your customers, and understanding all of the
opportunities in the field. The best product managers
combine these approaches, and stitch together a plan
that involves finding the best opportunities that the
market presents which also tie back to the company
level goals. This then culminates into a product roadmap
which helps to articulate the product strategy.
28
CHAPTER 3
DAVID BLAND
FOUNDER, PRECOIL
David is an author and founder based in Silicon Valley. In 2015, he founded Precoil to help
companies find product market fit using lean startup, design thinking and business model
innovation. David has helped validate new products and services at companies such as GE,
Toyota, Adobe, HP and Behr. Prior to advising, David spent over 10 years of his career scaling
technology startups.
How do you know if you should add a One of my all-time favorite quotes is from Theodore
new analytics feature to your product? Levitt. “People don’t want to buy a quarter-inch drill.
They want a quarter-inch hole.” This mindset not only
applies to drills, but it also applies to your software.
Behind every tool or feature request you receive is a
customer job to be done. It’s a task they are trying to
accomplish but unfortunately, they may not be able to
clearly communicate that to you.
Most likely you only see the tip of the iceberg as the
customer requests come in through your sales team,
customer support, email and social media. It’s your
responsibility to get to the job behind that feature
29
CHAPTER 3
What are some best practices for Customer interviews are deceptively hard to do well, but
conducting customer interviews to get by following a few guidelines, you’ll become an expert
to the job behind the feature request? in no time.
30
CHAPTER 3
31
CHAPTER 3
Do you then build the feature? Customer interviews are one way to get to the job
behind the feature, but it doesn’t necessarily mean you
should rush off and immediately build the feature. Sites
like Google often incrementally invest in new features
by first running more experiments. They may paper
prototype features with customers, show them clickable
dashboards without any of the backend infrastructure
in place or release a feature stub on the live site. The
feature stub is essentially a link to a feature, but when
clicked shows a “We’re not ready yet” message.
32
CHAPTER 3
Data as an Accelerator
JOHN CUTLER
PRODUCT EVANGELIST, AMPLITUDE
What are some best practices for People converge too quickly in their mind on a solution.
holding user interviews and/or And that guides their question asking. Allow your teams
gathering user requirements? to be a lot more divergent in the beginning. Some
people say, “Okay, we’re going to go out and interview
five users. And then, we’re going to know what’s going
on.” But by the third interview, because of the pet
solution they had in their mind, all their questions are
biased to that solution. Create a learning backlog and be
pretty rigorous about not pulling everything in progress
at once. Start with very specific questions.
33
CHAPTER 3
How do you know what capabilities The customer is always right and they’re always wrong.
your users need versus what they Especially in B2B, people know and understand the job
just want? they need to do, deeply. A lot of times, their vision of their
job might be constrained by how that job has been done
for the last 20 years or so. For instance, an accountant is
always used to manually going through and reconciling
everything. So the job is perfect reconciliation and
reducing risk. It’s not going through all their individual
items. If you go in with this idea, you might have some
innovation on how to do that. Customers are not skilled
UX researchers or business analysts. It’s not to discount
the customer, but you go in and really respect their
knowledge of their job. With this perspective, it becomes
much more empowering.
How can product teams validate their It boils down to getting it in front of them as quickly
users’ requirements throughout the as humanly possible. Some things are more important
development phase? to de-risk than others. The word “requirements” is
very loaded. Who is requiring what? We want a Boeing
airplane to stay in the air. So there’s requirements when
it comes to building a guidance system for Boeing. But
with a lot of B2B things, what is a requirement? What it
boils down to is prioritizing areas of risk. And that risk
might be technical. We don’t know if this thing will stand
up and if we’ve picked the right tools.
34
CHAPTER 3
How does a product team know it’s You need dashboards as early as humanly possible.
time to add new dashboards, reports, You need a way to communicate, and it’s often through
or data visualizations to their product means related to data. Any time you need to make sense
roadmap? of a problem, or understand an opportunity, you will
need a lot of qualitative and quantitative data. The earlier
you can introduce that to build a shared language about
work and share an understanding, the better.
How can product teams establish that Often, if you have a fancy dashboard and a user catches
there’s a demand for their dashboard? a hint of it, they’ll think it’s amazing or great. The eye
Should they be adding a new analytics candy is powerful. But when it comes to the UX side, the
feature entirely? product side of things, you really need to think about
the job that your customer is trying to do and think
about presenting data to them as just an accelerator
to get that job done. Because the risk that you run
is putting generalized analytics into your product.
Everyone says it’s helpful to them but it doesn’t really
help your customers be better at what they do.
35
CHAPTER 3
ANNETTE FRANZ
CCXP, FOUNDER + CEO, CX JOURNEY INC.
Are there common mistakes you see Three most most-common mistakes I see are as follows.
product teams make when collecting I think they are all intertwined.
customer feedback?
1. The first mistake I see is to not collect customer
feedback or bring feedback into product design.
Allowing the product to be designed by software
engineers based on internal requirements without
first understanding who the customer is, use cases,
problems to solve, etc. happens more often than
anyone cares to admit.
36
CHAPTER 3
What are some tips for creating • Map the journey from the customer’s viewpoint with
a customer journey map? the customer. The customer must be involved in the
mapping process.
37
CHAPTER 3
Why is CX design important when an The customer experience is important to consider when
application team is building a new designing or building any type of product, including a
dashboard/report for their product? dashboard or report. You have to remember that the
customer is the one buying and using the product.
So, understand who the customer is and what problems
she’s trying to solve or jobs she’s trying to do—and
develop the product with that in mind.
38
CHAPTER 3
BEN LINDERS
INDEPENDENT CONSULTANT IN AGILE, LEAN, QUALITY,
AND CONTINUOUS IMPROVEMENT, BEN LINDERS CONSULTING
Ben is the author of several publications including What Drives Quality. As an adviser,
trainer, and coach, he helps organizations with effectively deploying software development
and management practices. He focuses on continuous improvement, collaboration and
communication, and professional development, to deliver business value to customers.
Product teams want to build quality My definition of the quality of an app is how much it
applications. But what does that mean? satisfies the needs of the people who use it and the
value that it delivers to them.
39
CHAPTER 3
How do you know that your application To deliver value, an app has to fit the purpose. It has to
is providing value? do what users need and expects it to do.
What are some key considerations for Fitness is a relative thing, it’s not binary. Usually, apps
designing an app that is “fit for use”? don’t cover all needs, one app might have a better fit for
you than another one.
40
CHAPTER 4
Companies that succeed say predictive analytics makes their applications infinitely more valuable, separates
software from competitors, and offers new revenue streams. We interviewed predictive analytics experts who
share insights on new opportunities, common challenges, and ways your end users can benefit from this
capability today.
41
CHAPTER 4
MARCUS BORBA
FOUNDER, BORBA CONSULTING
Marcus has more than three decades of in-depth information technology experience, and more
than 20 years of experience developing data-driven solutions for companies.
What are the benefits of embedding Embedding predictive analytics helps improve the value
predictive analytics into an existing of analytics in existing business applications, enabling
business application? users to access all the data they need at the right time
and context, more efficiently, enabling valuable insights,
helping streamlining internal business processes,
monitoring latent risks, identifying trends, and also
creating mechanisms for continuous improvement.
42
CHAPTER 4
How can end users use predictive Predictive analytics can help you to upsell customers
analytics to upsell customers? in many ways: enabling qualified lead generation,
more information about your ideal customer profile,
information to adjust services or products according
to customer tastes, better information for marketing
campaign targeting, and finding the best channels to
reach your target audience.
For companies interested in predictive In my opinion, a good way to start applying predictive
analytics, where should they begin analytics to a business is in its marketing and sales
their journey? strategy. All customer relationship actions can benefit
from more accurate information—from engaging
audiences to customer loyalty. As generated leads
become paying customers, data collected from new
customers influences the next actions taken, improving
the whole marketing and sales process.
43
CHAPTER 4
Eric is the founder of the Predictive Analytics World and Deep Learning World conference series,
and executive editor of The Predictive Analytics Times. He is the author of the award-winning
Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die, a former Columbia
University professor who used to sing to his students, the host of The Dr. Data Show, and a
renowned speaker and educator
How do you define predictive analytics? Predictive analytics is basically a synonym of machine
learning. Although, it’s a bit of a subset: learning from
data in order to make predictions for each individual in
order to render all large scale operations or activities
undertaken by organizations more effectively. Each
individual prediction, such as whether someone would
respond to an upsell offer or whether they’re going to
cancel their ongoing subscription, directly informs the
operation.
44
CHAPTER 4
What are the benefits of embedding When you’re talking about embedding analytics, it refers
predictive analytics? to a case where the analytics are delivered to a human.
From the end user perspective, it might be a customer
service agent on the phone with a customer who might
be a prospect for upsell. If we’re talking about delivering
analytics to humans, then in the case of predictive
analytics and machine learning, we’re talking about
decision to support rather than decision automation.
45
CHAPTER 4
How would you recommend someone You want to assess the greatest opportunities which
start their predictive analytics journey? are basically anywhere there’s large scale operations,
such as mass marketing, and you think that could be
done more effectively. With the probabilities output by
pick the model, pick model is the thing generated by
machine learning. It’s the thing, it’s the new intelligence
gained from historical data. The thing it learns and
that thing, that predictive model, which is because it’s
a pattern, rules or formulas that has been determined
over historical data, now can be applied to current
data and make predictive probabilities or predictions
for each individual. These predictions aren’t generally
accurate in the conventional sense of the work up.
They’re significantly better than guessing. It’s a huge win
to apply it systematical over large numbers or decisions.
The lowest hanging fruit for an organization is often
targeting customer retention by predicting who is most
at risk of departure or defection is.
46
CHAPTER 4
What is the business value of Sales has to start with the value proposition first and
predictive analytics? How can it focus on the technology second. The value proposition
affect a company’s bottom line? is often a very simple arithmetic forecast. For targeting
marketing, it’s simple arithmetic to show that if we only
expend the cost of contact to the top 25 percent of
customers most likely to actually make the purchase
as a result of the marketing treatment. First, we cut
the marketing cost by 75 percent we also lose some
business.
47
CHAPTER 4
Advanced Capabilities
Ronald helps data-driven companies generate business value. He has been recognized as one
of the top 10 Global Big Data, IoT, Data Science, Business Intelligence Influencer by Onalytica,
Data Science Central and Klout and top 10 Predictive Analytics influencer by Dataconomy,
is author for leading Big Data sites like The Economist, Datafloq, Data Science Central
Dataconomy and is public speaker at leading Big Data, Data Science and IoT events.
How can embedded predictive analytics Embedded predictive analytics can work perfectly
increase value in existing applications? within existing applications, as it can increase the value
associated with them.
48
CHAPTER 4
What are other benefits of embedding Predictive analytics presents a complex challenge, which
predictive analytics? when mastered can help multiple industries around us.
49
CHAPTER 4
What are some industry-specific The industry-specific use cases for embedded predictive
benefits of embedding predictive analytics include usage in healthcare for proper
analytics? documentation and analysis of patient records, retail
for inventory check, and eCommerce for tracking.
Other industries include hospitality, finance, and
manufacturing.
50
CHAPTER 5
Security Is Crucial
For application teams, security scenarios can be complex and have very precise requirements. If you don’t start
with security, nothing else matters.
Embedding a third-party product—such as a platform for embedded dashboards, reports, and analytics—can raise
new security challenges. Other products have their own security frameworks and may not integrate seamlessly with
your setup, or they may even introduce new risks or gaps in security. This can lead to delays in your release cycle
and negatively impact your ability to grow and maintain your product over time.
In this chapter, application security experts outline key considerations for choosing and embedding a third-
party product in your applications. What are the risks of not asking the right questions early on?
51
CHAPTER 5
EDWIN KWAN
APPLICATION AND SOFTWARE SECURITY TEAM LEAD, TYRO PAYMENTS
Edwin is the Application and Software Security Team Lead for Tyro Payments. His approach
toward application and software security is to raise security awareness, provide light touch
controls to the software development life cycle to increase visibility of security issues and work
closely with engineering teams to quickly develop secure applications. He has presented at
a number of events, including RSA, All Day DevOps, AppSec Day and DevSecOps Leadership
Forum. Edwin started out as a software engineer and transitioned into the application security
role to lead a range of security initiatives when the company was working towards obtaining an
unrestricted banking license.
When embedding a third-party It ultimately boils down to CIA. CIA is the Confidentiality,
solution in your software, what are Integrity and Availability of the data/service.
some key considerations that you
need to think about? Some questions to ask about them are:
52
CHAPTER 5
What are the risks of not asking these Not asking these questions early might result in the
questions from the very beginning? business investing in rolling out a third party product
that might not be the right fit or properly deployed. The
consequences of that is delays in the project due to
additional work that needs to be done.
What are some common security I have seen people integrate platforms with multiple
mistakes people make when authentication systems. This usually results in users
integrating a third-party platform having multiple credentials, which is a nightmare
(such as embedded analytics) with to manage especially when adding new users or
their application? removing them as a part of the exit process when
they leave the company.
53
CHAPTER 5
TAL MELAMED
HEAD OF SECURITY RESEARCH, PROTEGO LABS
Tal has 15 years’ experience in the information security field, specializing in security research
and vulnerability assessment. Prior to being the Head of Security Research at Protego, Tal was
a tech leader at AppSec Labs, leading and executing a variety of security projects for serverless,
IoT, mobile, web, and client applications, as well as working for leading security organizations,
such as Synack, CheckPoint, and RSA. Tal is also a keen speaker; training DevOps and hackers
around the world, as well as lecturing at major security conferences; and a neat developer,
experimenting daily with offensive and defensive security.
What are some common security The main problem is new vulnerabilities being
mistakes people make when introduced into the system and there is always this
integrating a third-party platform risk when integrating a third-party. Even after going
with their application in order to add through due-diligence, and achieving compliance with
a new capability (such as dashboards most standards, there is a chance that the third party
and reports)? has a known or unknown vulnerability in the system
that will now become your problem as well. As for
vulnerabilities that are introduced through common
libraries, it is important to integrate security for your
applications, either in the cloud as a SaaS solution
or even in the local stations for developers that will
analyze the application code in search for these known
vulnerabilities prior to deployment. They then should
54
CHAPTER 5
What are security requirements to look In terms of security, the most important requirements
for when you’re evaluating platforms are regulations and standards. You do not want
or vendors to add new features to your to integrate a solution that will interfere with your
application? compliance requirements (e.g. HIPAA, GDPR) and
security posture. For example, what are the vendor’s
capabilities and policies for protecting data? If you can,
makes sure to ask the third party to provide you with a
penetration-testing report that shows that they have put
efforts into making their system secure.
55
CHAPTER 5
When integrating third party When introducing new features and technologies to
technology into your application’s your application, you first need to make sure you have
existing tech stack, what are key a have well defined and documented requirements.
considerations you need to think What are the business processes will you be supporting?
about from the very beginning? What standards and regulations are you subject to?
Do your research, a few hours in Google is not enough.
You want to be sure that you maintain your
security posture.
56
ABOUT LOGI
About Logi
Delivering compelling applications with analytics at their core has never been more crucial—or more complex.
Logi is the only developer-grade analytics platform focused exclusively on embedding analytics in commercial
and enterprise applications. Logi leverages your existing tech stack and supports unlimited customization and
white-labeling, so you can quickly build a completely unique analytics experience.
Over 2,100 applications have trusted the Logi platform to deliver sophisticated
analytics capabilities and power their businesses. The company is headquartered
in McLean, Virginia, with offices in Ireland and England.
57