This post is from the Numerify blog and has not been updated since the original publish date.
Top Five Challenges to Change Management in 2020
We are in the midst of a bold new transition in IT operations: away from rigid processes and toward agility. Organizations are seeking the benefits that a DevOps or similarly agile operating model can offer, which include faster delivery time and more responsiveness to market demands.
Technologies like Artificial Intelligence (AI) and Machine Learning (ML) are not only aiding in this transition, but they're also absolutely necessary. One review of DevOps literature even concluded that automation was one of the "two pillars of DevOps," alongside human collaboration.
Pressure to adapt and adopt to improve IT operations has introduced some key challenges for change management. Some of these challenges have existed for years, and they're now making their presence known more than ever. Others are just now emerging.
To help IT leaders prepare for these challenges and meet them head on, here are five change management priorities to focus on in 2020.
1. Continuous Development
The DevOps challenges conversation must start with the unique pressures of continuous development. Most organizations are deploying changes at an increased pace today. Amazon, famously, went from deploying a production change once every 11.6 seconds in 2013 to one deployment a second, on average, in 2015. Organizations much smaller than Amazon can still introduce thousands of changes a week.
Using legacy practices make it difficult to keep pace with the sheer velocity of change. The use of 100% human (non-automated) change overseers operating under a traditional development pipeline is no longer feasible. Plan/design/build no longer happens along a single track, for starters. Instead, each platform feature will have a separate development team working independently.
Continuous development and continuous integration prioritize zero downtime above all else. Should a major change-related incident occur, there is enormous pressure to avoid service disruption at all costs. Every minute of downtime can cost $5,600 – $9,000.
Making matters more complicated, the velocity of changes can mean that the real root cause of an incident can be related to a change made days or weeks ago. Organizations that aren't proactively monitoring for change-related performance threats will be unable to sift through backlogs of changes to track down the culprit.
This agile-focused environment has led to a dire need for more efficient processes that don't require gatekeepers and that can autonomously manage risk with minimal service disruptions.
2. Break Up ITIL Hierarchy
ITIL methodology prides itself on using managers and supervisors to mitigate risk. Between every step, including deployed changes, there is someone there to act as a checkpoint and gatekeeper.
One consequence of agility is that modern DevOps teams cannot reasonably be asked to stop. Previously, work was performed in stages with checks in between. It's now considered infeasible to have an oversight body like a Change Advisory Board (CAB) review every single decision.
As a result, organizations will phase out the use of CABs as a gate between every single change. In its place, many organizations are turning to CABs that only step in for certain changes or emergencies.
Internal DevOps teams typically work independently. They are rarely accountable for any consistent performance metric, save for areas that have the potential to directly affect the value stream. Each team may have its own working process, such as a Kanban board or project management software. Rarely are all of these separate systems consolidated, nor do they commonly have their data compiled and aggregated.
The main challenge with this lack of organizational hierarchy lies in consolidating risk management tactics across diverse teams, all without hampering their independence and agility. CABs and other operations managers need tools that can help them minimize the gating of work, accelerate the decision-making process, and automate lower-level tasks that cannot reasonably be performed manually at the current pace.
3. Break Down Information Silos
In a "Harvard Business Review" of DevOps organizations, half said that organizational silos represented their biggest challenge to value creation. Nearly a third (32%) said that problematic informational silos existed within IT.
Informational silos can be difficult to break through in IT since organizations tend to rely on tools that were not meant to talk to each other, such as service desk software and monitoring systems, like Automated Performance Management (APM).
However, IT leaders need data from all these systems of record in order to be able to perform actions like investigating root cause or grouping similar incidents. Without a robust, 360-degree data view, identifying risk factors and predicting incidents is extremely challenging. This challenge is compounded by the fact that many independent dev or ops teams tend to use their own preferred project toolsets.
In the context of change management, the fastest way to break down these siloes is to adopt a pre-built analytics solution that comes pre-integrated with your systems of record, including development tools upstream of the change, and monitoring tools downstream of the change.
4. Identifying Change Risks Before They Have an Impact
Major incidents are not the only contributor to service disruptions; minor problems can have a cumulative effect on technical performance, user productivity, financial performance, and other primary value-indicating metrics.
Isolating individual change risks from a human perspective is often subjective. CABs acting on minimal or siloed data lack the information to properly identify, quantify, and respond to a change-related risk. Objective metrics are needed to identify which factors contribute to change risk and what their suspected future impact will be if ignored.
Having a ML-enhanced change risk scoring algorithm can allow risk management leaders to rapidly identify and assess risks based on actual performance data. More advanced AI/ML capabilities can allow for prescriptive scoring, which streamlines decision-making by indicating the most optimal paths to take in light of a risk.
5. Reconciling the Role of AI in the IT Organization
IT operations have begun to embrace AI with less and less fear. While 16% of surveyed organizations saw AI as a "serious job killer" in 2017, just 9% had this opinion in 2019.
Despite enthusiasm for AI, many IT leaders are unsure of the best tactic for implementing AI in change management processes. If automation systems aren't properly calibrated and monitored, there is a risk that problems could go unchecked until they turn into an incident, if IT organizations over-rely on AI as the sole gatekeeping measure.
On the other hand, under-reliance on AI means manually intensive work, like performing analytics across hundreds of data sets across dozens of sources. The prospect is simply unmanageable for many organizations, which again means that risks could go unchecked.
The solution is an optimal "sweet spot" of automation that can enhance existing teams' capabilities without replacing key talent and expertise. This "right" level of automation will differ from organization to organization, but those concerned about finding it should focus on capabilities that produce immediate value. E.g.: Using change risk prediction and scoring analytics to suggest actions for CAB approval, or flagging problems with common root causes so that repeat mistakes are not introduced in new code deployments.
The goal is not to replace workers, but to add efficiency and remove tasks that have low value-add compared to the amount of manual work they require. IT leaders need to be sure that they not only acquire the right AI-augmented tools but also train employees on their proper use.
Embrace AI and Agility in 2020 for Ops' Ease of Mind
Every organization will have to grapple with all of the innovations to change management in 2020.
Using ML-driven analytics models and AI analysis can help further the goals of agility without sacrificing risk prevention potential. The new agile-focused structure of IT operations offers layers of protection against risk while allowing feature teams to operate according to their own pace and idiomatic process.
The collective benefit of these adaptations is that organizations can embrace agility while keeping risks low, thanks to the benefits of AI along with human supervision and expertise.
Learn more about how AI can enhance change management and make it more agile in our recent webinar produced in collaboration with Forrester: "Make Innovative IT Change Management Process Smarter and Faster with Artificial Intelligence (AI)"