Measuring our Digital Transformation using Splunk
First, an introduction: For those who do not know me, my name is Gerry D'Costa, and I am a Transformational Architect, Splunk enthusiast and Founder of PiiVOT Inc.
These past few years, I have helped organizations assess and build their digital transformation journey. In that time, I have come to realize three (3) main challenges:
Transformation paralysis. Technology publications, evangelists and software vendors alike have all unconsciously muddied the waters on what digital transformation means - which has, in many instances, paralyzed I.T. departments and organizations.
Where to start. Organizations that have overcome this paralysis still grapple with trying to understand where to begin and how to define what business problems they are trying to solve. (Is it to increase end-user productivity? Improve innovation for competitive advantage to prevent market disruption?)
What defines success. If an organization has worked hard to tackle the previous two challenges, they then inevitably struggle with getting transformational initiatives off the ground and establishing metrics for success.
These items are challenging in their own right, however, in my experience, the hardest to achieve is the last point: To launch digital transformation initiatives.
Getting transformational initiatives off the ground requires delivering on many tasks, least of which are the clear communication of objectives and the definition of timelines/budgets to executive leadership.
But most importantly, we need a process to define success criteria through key performance indicators (KPIs) and to use a mechanism to measure/visualize these KPIs, allowing leadership teams to measure progress and determine the return on investment. (ROI)
In this blog post, I would like to convey some of my insights and education on the following:
The concept of digital transformation and what it truly means;
How to define transformational problems into use cases to build a roadmap towards achieving them; and
Most importantly, my solution for helping organizations act on and monitor transformational progress using the Splunk platform for data aggregation, transformation, analysis and visualization.
Early Digital Transformation
To those old enough to remember, I harken back to the year 1983 when Lotus software (later owned by IBM) came out with the Lotus 1-2-3 spreadsheet software for the recently released 2-year old IBM Personal Computer.
Lotus 1-2-3 killed off its Apple II competitor, VisiCalc, seemingly overnight and changed microcomputers from just being a hobby for computer enthusiasts, to being a serious business tool.
The IBM PC, together with Lotus 1-2-3, was considered a new innovation that changed the way organizations performed all financial operations, such as testing economic forecasts and the creation of departmental budgets.
Having an IBM PC on everyone's desk meant that number-crunching could now be democratized and accessible to the masses, instead of being manually calculated. (changing a single number resulted in the manual recalculation of the entire physical worksheet!)
Any company that did not understand the value of owning this hardware and software meant that they were likely going to be disrupted and were now at a competitive disadvantage.
This adoption of technology by businesses to enable their end-users to become more productive was one early example of digital transformation.
However, if I just stopped there, I would mistakenly lead you to believe that all you needed to do is to purchase some kind of hardware or software product and viola - here is your digital transformation!
This is where some of today's technology vendors typically end their engagement with the business and I.T. by selling them technology that they say will allow them to digitally transform. While this may be true, it is not the whole story.
What does Digital Transformation mean?
If we once again go back to 1983...
I was my school's computer nerd. This "proud" distinction allowed me the privilege to stay inside during breaks (I grew up in freezing cold central Alberta) copy and play computer games during lunch and learn educational prescribed software to demonstrate to teachers and students on how to use the Apple II.
This is where I was first introduced to VisiCalc. (check out this humorous video)
I was genuinely fascinated by how you could automatically alter an entire series of numeric results in VisiCalc by just changing a single number in a cell!
The problem was, I didn't really understand all the cryptic commands and functions of VisiCalc (my young age - notwithstanding) and did not know how I would incorporate it in my every life - so I never used it.
In other words, without having appropriate product training and an understanding of how I needed to change my lifestyle (or daily processes) to effectively use spreadsheet technology meant that it was not going to be adopted.
Real digital transformation is not just about purchasing or developing something innovative. It must account for the ability to understand the business's cultural, procedural and human changes that are necessary for the effective adoption of technology.
How can we start our Digital Transformation?
Here is where it gets tricky.
Achieving digital transformation is unique and depends entirely on what your organization defines as success. It also depends on your use case: A Public Cloud transformation can be utterly different from the goal of achieving a Modern Workplace or Digital Marketing use case.
What I recommend is to focus on the most impactful use cases. For most organizations, it is the transformation of their I.T. departments so they can better support other business units with their own technology initiatives. This support usually manifests as a way to securely and optimally operationalizing public cloud services.
With this use case in mind, and our understanding of what digital transformation means, we must:
Identify the organization's business outcomes and goals for I.T.;
Assess and measure I.T.'s maturity, using a People, Process and Technology approach; and
Create a roadmap (aligned to our business outcomes) from where we are to where we need to be with timeline estimates and relative levels of effort required.
This may sound difficult, but if you use a framework that is pragmatic and delivered within 3 - 4 weeks, such as the custom-developed automated framework we have developed at PiiVOT, then you can overcome these significant hurdles.
There is much I could say about this topic alone; however, for more information on PiiVOT's transformational framework, please stay tuned to a future blog. But in the meantime, please visit PiiVOT's services website for some additional details.
Yet, what are we missing?
In the past, I would successfully run an organization through our transformational framework, provide them with some critical insights and benchmarks, then leave recommendations in their hands to execute.
Unfortunately, in most cases, this typically resulted in nothing getting done with the deliverable collecting proverbial dust on some file share or cloud share folder.
I came to realize that I was breaking my most crucial attribute for digital transformation; By only providing a roadmap, I was not addressing the necessary educational and behaviour changes to help organizations proceed with their transformational initiatives.
Acting on our Transformational initiatives
What was needed was to help executive leadership and I.T. to identify success by defining key performance indicators (KPIs) for each of our recommendations and measure their transformational progress using an automated dashboard. This would lead to increased support and understanding of the value they were bringing to the organization.
This is where Splunk comes in.
What is Splunk?
Splunk is a software company founded in October 2003 by Michael Baum, Rob Das and Erik Swan. Their goal was to build a search engine tool for real-time flows of machine data to produce visualizations for actionable insights.
Splunk has spent years building its service to include amazing premium service applications that integrate into their larger enterprise data platform. Some of the most popular service applications include:
Splunk I.T. Service Intelligence: Monitoring and performance analytics using A.I. of many I.T. service infrastructure and other business services.
Splunk Enterprise Security: Collects event information from security appliances like firewalls and server logs to provide insights for quick detection and response.
Splunk User Behaviour Analytics: A gathering of user statistics to better understand anomalous user behaviour events in the organization.
But what has really impressed me is their work with their customer base to build an impressive array of templates for specific use cases, such as:
Real-time DevOps developer analysis
Financial wealth advisement
Hospitality guest experience monitoring
Internet satellite beam capacity monitoring
and many more...
(Check out the "12 days of Splunk Use Cases")
A Splunk use case: Measuring transformational success
However, once you start using the platform, you will begin to see that the real power of Splunk is that it can ingest, transform, analyze and visualize any form of structured or unstructured data from any service, server or appliance.
This is an immensely powerful and flexible tool. Which makes it the ideal platform for our specific use case: To help measure the success of digital transformational initiatives through defined KPIs.
Using Splunk to measure organizational culture
Before we talk about how we use Splunk to measure KPIs, I would like to first briefly review PiiVOT's transformational framework:
If you have taken the time to visit PiiVOT's website, you would have noticed that we use eight (8) dimensions when assessing an I.T. department's transformational maturity:
Organizational;
Operational;
Governance;
Platform Selection;
Identity & Access;
Network;
Security; and
Data Management
As part of our People, Process and Technology methodology, each dimension will have each have associated recommendations to be actioned.
If we look at the Organizational dimension, one typical recommendation involves understanding and improving I.T.'s organizational culture to help employees embrace change, be more innovative and welcome new ways of thinking. Ways we address this can be varied, but a common area of focus comes from changing behaviours of the leadership group.
The Employee Net Promoter Scores (eNPS)
As we enact these changes, one way we can gauge our progress to improve organizational culture is by regularly issuing anonymous surveys to I.T. department staff to measure our employee net promoter scores (eNPS) over time. Questions in our eNPS survey are rated from 0 (low) to 10. (high)
The way net promoter scores are calculated is by respectively classifying all our responses, between 0 - 6, 7-8 and 9-10 as being Detractors, Passives and Promoters. We determine our eNPS score by subtracting Promoters from Detractors, dividing the result by the total number of responses and multiplying by 100. This means we can have a range of outcomes from -100 or 100, with the threshold value for a positive eNPS score being 70 - this becomes our KPI.
For more a great explanation of employee net promoter scores and their benefits, please click the link.
Loading eNPS data in Splunk
So, how do we get the data into Splunk? I will not go through all the details here, but here are the high-level steps on how we do it a PiiVOT:
We create a survey form through Google Forms and capture all response information into a Google Sheet. Response results are rated from 0 - 10. The following is a sample survey build in Google Forms.
We can then go to our Splunk server and install the Google Import/Export App by browsing/searching the available apps.
Once installed, we will need to create the required API keys in the specific account to allow Splunk to access the Google Sheet.
Once we start collecting responses, we need to configure Splunk to add the data by monitoring the Google Sheet.
Survey Result Data will then be loaded into Splunk at a set frequency.
Since PiiVOT builds digital transformation assessments and eNPS surveys with multiple organizations for a broad range of Industry Sectors, we can use this information to make predictions based on the Industry Sector. (This will be discussed in a later section)
NOTE: The data and visualizations displayed here have been randomly generated for the purposes of this blog and do not reflect any previous or current customers.
Gaining insights from eNPS survey data through Splunk
Once data is being collected by Splunk, we are now free to start analyzing the data using the Splunk Search Processing Language (SPL) to see trends and progress through Splunk Dashboards.
Dashboards are areas where data analysis results, charts and graphs can be displayed for quick and easy consumption. In the following example, we have configured our Dashboard with dropdown options (populated by SPL queries) to automatically regenerate our visualizations based on our selections.
The next two charts show us in the six (6) months after having started our transformational initiatives and activities and the improvement trend of corporate culture, based on the Industry Sector against our KPI.
However, we can also use the dropdown boxes to drill down to any particular organization if required.
Great! We now can see our progress to improve organizational culture against a defined KPI, but it would be incredibly helpful if we could predict when we might reach our goal of 70.
Generate predictions with Splunk's Machine Learning
With Splunk's Machine Learning Toolkit, we can quickly build a linear regression model by segmenting out data into testing and training subsets to yield a prediction of how long it might take to reach our defined KPI.
No Python or R programming required!
We begin by using the Predict Numeric Fields Assistant in the Splunk Machine Learning Toolkit and pasting in our SPL query we used to generate our dashboards.
We can then pick any ML algorithm to generate our model. In this case, we'll use a simple linear regression technique to try and predict our eNPS score based on information in four (4) different fields.
We will then need to determine how much of our data will be allocated to train our model and how much will be to test it. A 70/30 split is typical.
By defining a name for our model, we can then apply it to any other data to yield predictions.
Clicking on the "Fit Model" button builds our model and gives us statistics as to how accurate it is based on the data we trained it on and the test data we used for validation.
We can now create a data scenario to feed into our model to predict when we will achieve our eNPS score of 70. The following charts show our prediction that Company 7 will obtain an eNPS score of 70 in between the seventh and eighth month of our transformation initiatives!
By publishing this analysis into the same Splunk dashboard, I.T. leaders can visualize and predict their progress, which can be reported back to executive leadership.
NOTE: As with any Machine Learning algorithm, predictions are only as good as the data we run our models on. The more volume and accuracy of data, the better the predictions of our model.
Other digital transformational metrics
What I have just outlined is only the tip of the iceberg when it comes to using Splunk to monitor your transformational progress. Other potential metrics include:
Automatically measuring the level of Cloud automation over time;
Assigning KPI and ROI analysis to visualize the real-time business cost savings;
Automatic measurement if we are protecting Cloud resources and how we are improving over time;
Measure how we are progressing with the development of Governance and design documentation which is actively published to a private GitHub repository;
and many, many more!
Conclusion
Digital transformation can not merely be achieved by purchasing a piece of hardware, software or SaaS service. We must first learn that it requires a change in our behaviours AND our technology to fully realize innovative transformation.
Once understood, we need to identify what problems are we trying to solve (or business use case), and employ a methodology that assesses people, process and technology aspects of the organization. This assessment needs to identify our current transformational maturity and provide recommendations for improvement.
However, for the business to have the confidence to support and enact these recommendations, we need a mechanism measure/visualize progress through defined success KPIs.
Using a flexible data engine tool like Splunk to ingest/process data, create app add-ons and use the Machine Learning toolkit allows us to easily measure our digital transformation KPIs, visualize our progress and predict future events.
Putting all these pieces together enables organizations the ability to assign a tangible value to our initiatives and achieve their business goals.
If you have any questions or want to talk further about how PiiVOT can help your organization with digital transformation or have questions about what Splunk can do for you, please reach out.