The days of IT being fully in control of information technology are behind us – and that's a good thing. Who knows when the actual tipping point occurred? Was it mobile, BYOD, cloud development, shadow IT, SaaS adoption, the emergence of millennials, or the convergence of all of the above that caused the sea change? All we know is that we're now in a different era, and it represents a unique opportunity for IT to become a true business enabler.
Modern enterprise end user infrastructure is comprised of multitudinous people, devices, applications, tools and communications with millions of interactions and dependencies going on at any one time. This is the Unified Workspace environment and, with limited visibility, IT is only partially in control. In many cases IT organizations rely on their partners or end users to understand components they cannot observe directly, making them purely reactive in planning and delivery.
In conjunction with this, users and their expectations have also changed. Empowered users with the ear of the business want the same kind of productivity and choice at work that they enjoy with their personal devices and applications. They are tech savvy and always want to understand the technology at the core of how they earn their livelihood.
Given the dynamics of this world, how can personas help IT?
Why Use Personas?
Personas are abstract models of end users based on the work patterns, behaviors and tools of real people within an organization. As models, personas give us the ability to understand how end users do their work and how their work can be made more productive. Information technologists use personas to better understand their companies and help align the business with IT. Personas are used as tools to distill users down to a manageable number of user types, and to communicate and discuss their goals and needs. Personas also allow for a more holistic understanding of infrastructure, its components, and the business processes to which it is tied. Knowing what a persona requires from a hardware, software, mobility, and security perspective allows end user computing teams tasked with their support to properly provision hardware, software and services as well as direct support resources, including budget and personnel, to maximize the end user experience.
From a budgeting perspective, being able to negotiate hardware contracts, software licensing contracts, and the like based on real numbers of what is required (and, more to the point, what is NOT required), allows for optimal budgeting and cost savings in these negotiations, since the enterprise will not need to buy more compute power than is needed per desktop or laptop. IT and purchasing will know exactly how many licenses of what software product are needed, instead of the current semi-educated guessing that is used by so many IT organizations.
Personas have been around awhile in various forms. They have been used for years by software developers to better understand requirements and to develop use cases. They have also been used by consultants as a framework in the initial assessment of an IT environment. Personas allowed the IT consultant, for example, to segment a complex IT environment so they could better understand underlying issues and develop strategies for remedying those issues. However, there were problems with how these personas were constructed. Consultants often used interviews and day-in-the-life studies to gather information, invariably leaving holes in the information gathered. Unfortunately, one common result of the challenging collection and structuring process was the inclusion of fictitious information. This approach would, more times than not, ultimately yield either an incomplete picture or a completely wrong picture of the environment due, in large part, to human error. When these manual efforts are used, human error will eventually always set the effort on a path of diminished benefits and possible failure.
Even worse, after a company or a consulting group went through the process of defining personas and doing the initial groupings it was effectively impossible to manually keep the personas up-to-date; people changed jobs, new people were hired and others left the company. In addition, the attributes that made up a given persona could change as new devices and applications were added and PCs and laptops were replaced or refreshed.
This all culminates in a simple question: is there a process that will allow IT organizations to maintain visibility into their end user needs and manage them efficiently and accurately with Personas?
We believe that we can use Lakeside Software's SysTrack to overcome the biggest weakness in the use of Personas and that we can use SysTrack to define them with real world, continuously updated data. From there we can ensure that, based on these defining characteristics, users will remain in the proper Persona group, as defined by their needs and usage.
In the event that these characteristics change, dynamic updates can then move a user to another Persona group. Beyond that, alerts, notifications, and actual actions can be taken to ensure that this user is receiving all the proper device, software and mobility deemed necessary for that Persona. More importantly, we can also ensure that provisioning of items from the previous Persona are removed as well.
SysTrack is an end user analytics platform that gathers thousands of data points per second on end user and supporting systems. This data covers every detail about the end user experience, hardware, performance, applications used, operating system, boot and login times, power consumption, mobility, user security, virtualization infrastructure, storage and more. Leveraging a patented distributed database architecture, stored locally on the desktop, laptop, VDI session or SBC host, data is then summarized and uploaded on a daily basis to a master DataMine. Lakeside has built visualization and reporting tools that allow the data to be analyzed and trended so that patterns and anomalies in the environment can be seen. One of the tools, SysTrack Site Visualizer for example, facilitates user segmentation. It is capable of visualizing specific personas and then continuously populating the data that make up those personas with real time information.
Figure 1 – Approaches to Persona Assessment
Our SysTrack End User Analytics platform can:
All of the above points help organizations optimize their resources and save money. The potential savings directly correlates to how much automation can replace manual steps in the processes involved in creating, managing and maintaining Personas in a given environment.
Figure 2 – Persona Distribution
In May 2015 Gartner analysts Federica Troni, Ken Dulaney, and Richard Marshall published an excellent paper entitled, "Segment Users by Workspace to Allocate Physical Devices, Digital Tools, Support and Services." In that paper the authors came up with four essential workspaces:
They defined each of the workspaces in detail including what devices are used in each workspace and what roles might be included in each workspace. They then provided a methodology for how each workspace might be populated with the right set of computing and communication devices. In our experience this is a very valuable approach, and using Lakeside's SysTrack, we can determine whether a persona is Deskbound, Non-Deskbound, or Shared. "Industrial", however, is trickier because a client's hardened devices might not be included in a public list of known hardened devices. In that case we'd need the client to provide a list of what they consider to be their Industrial devices and SysTrack can take the discovery from there.
Once we determine whether a persona is Deskbound, Non-Deskbound, Shared or Industrial, we then propose to take that Persona to the next level by determining whether the Persona is a Knowledge Worker, a Task Worker, or a Power User. Using the vast wealth of data that the SysTrack agent collects, Lakeside has developed an automated discovery process that analyzes usage patterns to determine Knowledge Workers, Task Workers, and Power Users.
In the example below, we determined that a user could be Deskbound using any device and that he would be considered Deskbound if he was on the same subnet for 70% of the time. We would not consider a user Deskbound if they used a VM or published desktop or if they were mobile. For the Non-Deskbound, a user could be using any endpoint and would be considered mobile if they were on a different subnet 70% or more of the time. A Shared user could share any device and be either fixed on one subnet or move across multiple subnets. If they were using a VM or a Published Desktop, we considered them a Shared user. As mentioned above, customers would have to make a determination about which devices they considered Industrial, and once we have that list we can segment the associated users.
Once we segment users into Deskbound, Non-Deskbound, Shared and Industrial, we can further segment into Knowledge Worker, Task Worker, and Power User. Our experience has been that for Task Workers you need to normalize the data per customer. Again in the example below the customer determined that a Task Worker used 10 or fewer applications which is probably at the high end of the number of applications for a Task Worker. We developed a score for determining a power user based on the user's consumption of CPU, Memory and IO. The Power User data is pulled from our SysTrack Site Visualizer "People" dataset. Knowledge Workers would be all those users in neither the Task Worker nor the Power User groupings.
Figure 3 – User Segmentation Guidelines
The example below is from an actual Financial Services customer. We were able to go into a unit within that company and break out the number of Deskbound, Non-Deskbound, and Shared Users. The customer did not have any Industrial Users. We further segmented each of those groups into Knowledge Workers, Task Workers and Power Users.
Figure 4 – User Persona Segmentation Study
At this point we are going to be able to satisfy the Persona needs of many customers, systems integrators and consulting firms. As we have seen, at the high level, and using the Gartner personas, we can determine Deskbound, Non-Deskbound, Shared and Industrial and, if needed, we can go a level deeper and determine whether for each of those segments the Persona is a Knowledge Worker, a Task Worker, a Mobile Worker or a Power User.
However, we find that in many cases clients want to define personas even more deeply and they want to take it down to yet another level. Let's call this level Role-based Personas. They're still abstract models, but another step closer to the work of actual users. For example, a large healthcare provider might decide that one of their personas is "doctor" and doctors might include General Practitioners, Nurse Practitioners, Internists, Surgeons, Dermatologists, Oncologists and all of the other specialties. Again we find that this level of persona resonates with our clients. We think that over time Systems Integrators and consulting organizations will not only utilize personas at the more abstract level like those suggested by Gartner, but also engage with clients to help in the development and delivery of personas that are closer to the actual roles in a given industry. As such, they will begin building libraries of personas for each of the industries they serve. Thus, when they first engage with a
client they will already have a set of role-based personas that they will offer as a starting point, and clients will then work with their integrator or consultant to further refine and customize per their particular business.
Figure 5 – Persona: Doctors and Nurse Practitioners
In the rest of this paper we're going to take a look at how Role-based Personas might be developed and used at a bank and how SysTrack could help populate those personas, as well as continuously refresh them.
In this very large "Big" bank a project team was charged to develop a set of personas that would work for the entire bank. Representatives from across the bank were chosen to participate on the project team. The project team consisted of a project manager and two stakeholders from each of the verticals, as well as representatives from global engineering and global operations. This project team worked for just under a year developing their personas.
From the outset the project team decided that they needed a succinct problem statement that would identify the issues that needed to be addressed, to set the direction for the project team and to help guide the planning process. Their problem statement addressed why they saw the need for personas. The problem statement stated that the sheer size of the bank added a level of complexity that made it difficult to properly support the various user types within the organization and the advent of unified computing further exacerbated the problem. By knowing what each persona required from a hardware, software, security, mobility and support perspective, it would allow the bank to properly direct resources including budget and personnel so that their personas could be effectively and efficiently provisioned and supported.
The project team conducted numerous meetings across the various business verticals and at varying levels within each organization. At each meeting the team explained the goals of the project which were to:
The intent was to include the input from all of the verticals in order to make final agreement easier because every vertical and all key groups had ownership and input throughout the process.
The project team determined a persona based on user CPU, memory, and disk usage, what applications were used beyond core applications in the image, how mobile the user was, as well as their encryption requirements. As the team worked, they identified gaps and problems that needed to be resolved.
The final persona plan recommended nine personas, and the plan was sent up to EUC senior management for approval and implementation. Implementing the persona plan included hardware renegotiation with desktop, laptop vendors, application package suite restructuring and virtual image changes to reflect different personas to– VDI Image requirements for CPU/mem/disk and application usage.
Nothing they planned included anything remotely like an automated process. Everything was done manually. This meant that while on day one everything looked and functioned fine, without proper maintenance, the management of users to persona quickly became disjointed. There was no automated process to keep the personnel-to–persona map correct. For example, if a user currently in a remote sales site (persona: Mobile Worker) changed jobs and moved to an administrative assistant inside the sales office she reported through (persona: Office Worker), she still continued to use the mobile application stack on the mobile hardware platform. Accordingly, she didn't have the proper applications or hardware needed for an Office Worker. The bank saw 9,000 changes like this per month or 108,000 in a year. Without the ability to keep up with these changes, the bank was wasting millions in over provisioned hardware, and excessive application licenses.
Another problem was that the bank decided to keep the number of images per region to just one for both traditional and virtual desktops. This mandate required every user to have the same core application stack that came with the build or in the image. This wasted licenses for products that users simply didn't use (Adobe Writer, Full license, WinZip, Office Pro instead of base, etc.). Using the personas to identify optimal asset utilization would have made the program a huge success.
Finally, there were no steps taken to ensure data remained valid and not stale. Some testing was done to validate users were getting the right Virtual Machine based on their need and persona, but again no automated process was put in place to move users as their jobs changed. The result of all of the time and effort put into this manual process was a snapshot in time listing every employee at the time of the snapshot and their associative persona identification. No real cost savings were identified because decision makers were not comfortable enough with the stale data to:
At Lakeside Software we have undertaken a project to automate the persona discovery, incorporation and management processes. To do this, we deployed our SysTrack software onto many disparate systems inside our company. End-user behavior was monitored by the Lakeside SysTrack Agent and compiled on the SysTrack Master Server. After thirty days of data collection, patterns of usage were analyzed and compiled. The goal of the project was to determine how those involved in the study worked by answering the following questions:
We then developed an automated report that addresses and populates each of those questions. In the report we look at all of the key attributes of the persona including devices used, resource consumption per CPU, memory, and storage, printer use, what domain they use, USB usage, GPU usage, and mobile device usage. We also looked at their business critical applications and what public and private websites they visited. We mapped the day in the life of each user including when they were working remotely. We were able to determine who all of the users were, what their system usage was, which users used multiple systems, how much they used each system, and what domains they used. We took a look at their resource consumption by CPU, by memory and by IOPS, as well as their storage usage and what storage devices they were using. We mapped their drive usage and their printer usage.
SysTrack's dashboarding capability allows us to display the results of our persona segmentation in a variety of graphs. Below is an example where we are showing the number of users broken out by workspace:
In the world of Unified Computing, IT absolutely needs to understand what users are mobile and how much they are mobile. Using SysTrack we were able to determine end user mobility by looking at two behaviors:
For Observed Mobility we determined the following classes:
In the graph below, we segment Knowledge Workers, Task Workers and Mobile Workers based on their mobility:
With Exchange Synchronization we can determine the number of mobile devices attributed to the user by determining how many of that user's mobile devices are synchronizing with the exchange server. We can also determine the mobile device type – iPhone, iPad, Android, Blackberry down to Samsung SMN920A
Application Usage is also key in determining which users belong to which persona. SysTrack not only tracks which applications are used, but we also track applications by Focus Time. We not only know what applications are open at any one time, but also which applications are getting the most actual usage. This allows us to determine which applications are the highest priority for a user, and a persona and the applications that are the critical business applications for the enterprise. In fact, after determining the critical applications, we can determine their typical weekly use. For a given persona and even for a given user, there may be a critical application that falls outside the business's critical applications. This is important to understand as well.
Other areas that we track include VPN usage, the number of concurrent users, and website usage including public and private website usage.
The net of the multiple dimensions of data gathering and consolidation made possible with SysTrack is a fully formed, continuously updated picture of existing personas and the users associated with them.
By deploying personas populated and continuously refreshed with SysTrack data, we see numerous benefits:
Going forward, Lakeside is looking to work with their SI partners to help them build out persona libraries for their customers. It may be that a given SI wants a more abstract level of persona – for example, Knowledge Worker, Task Worker, Power User, Mobile Worker. We can help them do that. There are others who will want to build a library of personas by industry and we've provided some banking examples in this document demonstrating what those might look like. The key for us will be automating the cultivation and usage of these personas to the fullest extent possible. For example, we want to create custom triggers based on hardware utilization realities. Accordingly, we would like to dynamically move a user to a different persona based on their real world hardware utilization, notify the correct support team, update the correctly maintained persona listing, and potentially (in the VDI space) move that user to a new VDI pool, reducing their hardware reservations and freeing up resources for other users. In a large VDI plant this could save thousands of dollars and many calls for "system slowness" to the Help Desk.
Our contention is that the value that can be derived by businesses using personas is enormous. We see SysTrackbased personas as a catalyst for innovation across numerous disciplines within an enterprise. This journey has just begun and we couldn't be more excited about its potential.