Unforgettable 2 days at the 3rd Global LeSS Conference, at Angel Orensanz Foundation – the historical landmark in NYC.
Experience Report by Guest-Blogger Ram Srinivasan
Experience Report by Guest-Blogger Mark Uijen de Kleijn
LeSS Graphic Art
Personal Memorable Moments
Next LeSS conference (2019) – Munich, Germany
- There is a frequently seen confusion with respect to the definition of agile coaching: coaching focus (e.g. enterprise vs. team) is confused with coaching alignment (centralized vs. decentralized) within an organization
- Centralized coaching departments run the risk of turning into a single-specialty organizational silos that are locally optimized for their own expansion and personal success; they are also removed from real action. The reasoning behind: standardization – has its weaknesses.
- Centralized coaching is often limited to being “responsible for introducing KPIs, documentation of script-style-one-size-fits-all best practices and cookie-cutting approaches”. This leads to system gaming by other departments and organizational silos that must “meet numbers goals”
- Centralized Agile coaching makes sense only when it takes place within an organization that is small enough to be effectively managed front-to-back (including its all organizational layers) and is genuinely supportive of its own coaches, by providing them with “organizational immunity” and operational safety – to enable them perform their challenging duties
- The main advantage of decentralized coaching approach is that coaches are close to real action: deeply engaged with products/services, and are intimately engaged with senior leadership. Decentralized coaching is deep & narrow (as opposed to being broad and shallow) and takes time to cause meaningful and sustainable organizational changes.
The purpose of this post is to summarize two very important and independent topics and then integrate them together, into a joint discussion. The topics are:
- Moving from rigid annual budgets to rolling forecasts (super important! in agile/adaptive product development environments)
- Quality of scaling in agile product development, specifically Scrum
…and tying effective scaling of Scrum to dynamic financial forecasting.
Rigid Annual Budgets vs. Dynamic/Rolling-Wave Forecasting
Challenges presented by rigid annual budgets have been known for a long time. For people that are new to the topic, a great way to stay on top of most recent research and publications, is to follow what is going at BBRT.org (Beyond Budgeting Round Table). One of BBRT’s core team members – Bjarte Bogsnes, in his book “Implementing Beyond Budgeting: Unlocking the Performance Potential” (please, refer to the book’s highlights here), clearly summarizes the problems with conventional, end-of-year rigid budgets. They are as follows:
- Budgets represent a retrospective look at past situation and conditions that may not be applicable in a future
- Assumptions made as a part of a budgeting process, even if somewhere accurate at the beginning, get quickly outdated
- Budgeting, in general, is very time-consuming process, and it adds additional, financial overhead to organizations
- Rigid budgets, can prevent important, value adding-activities, and often lead to fear of experimenting, researching and innovating (crucial for incremental development)
- Budget reports are frequently based on subjective metrics, as they take on the form of RAG statuses, with the latter, introducing additional errors and omissions (for details, please refer to Red, Yellow, Green or RYG/RAG Reports: How They Hide the Truth, by M. Levison and The Fallacy of Red, Amber, Green Reporting, by G. Gendel)
- Budgets, when used as a yardstick to assess individual performance, often lead to unethical behaviors (e.g. “churning & burning cash”at year-end to get as much or more next year) or other system-gaming activities
…The list of adverse effects caused by traditional budgeting is long…
On contrary, a rolling-wave forecast, respects the fact that environmental conditions are almost never static, and recognizes that if too much reliance is placed on prior years’ financial situation, it may lead to miscalculations. Rolling-wave forecasts are based on frequent reassessment of a small handful of strong KPIs, as oppose to large number of weak KPIs, as frequently done in conventional budgeting.. The more frequently forecasts are being made, the higher chance that most relevant/reliable information will be used in assessments. One good way to decide on cadence of rolling forecasts is to align them with meaningful business-driven events (e.g. merchandise shipments, production code deployments, etc.). It is natural to assume that for incremental/iterative product development (e.g. Scrum), when production deployments are made frequently and in small batches, rolling-wave forecasting could be a concurrent financial process. Short cycle time of market feedback could provide good guidance to future funding decisions.
It is worth noting that one of the key challenges that Scrum teams face today, is the “iron triangle” of conventional project management, with all three of its corners (time, scope, budget) being rigidly locked. And while the most common approach in Scrum is to make scope flexible, ‘clipping’ the budget corner brings additional advantage to teams. Above all other benefits, rolling-wave forecasts address the problem described in #4 above, as they provide safety to those teams that want to innovate and experiment.
But what if there not one but many Scrum teams, each working on their own initiatives, running under different cadences (asynchronized sprints) and servicing different customers? How many independent rolling-wave forecasts can one organization or department adopt before things become too complicated? What is too much and where to draw a line?
Before we try to answer this question, let’s review what is frequently seen, when organizations attempt to scale scrum.
Proper Scaling vs. “Copy-Paste” Scaling
Let’s look at the following two situations: (1) more than one Scrum team, independently, doing their own Scrum and (2) more than one Scrum team, working synchronously, on the same product, for the same customer, sharing the same product backlog and domain knowledge. The former case, is referred to as “Copy-Paste” Scrum, clearly described by Cesario Ramos. The latter case can be seen in skillful Large Scale Scrum (LeSS) adoptions. Here are some of the most classic characteristics of both scaling approaches:
|(1) – “Copy-Paste” Scrum||(2) – Large Scale Scrum (LeSS)|
Note: Please refer to Scaling Organizational Adaptiveness (a.k.a. “Agility”) with Large Scale Scrum (LeSS) for additional graphic illustration.
Based on the above, the following also becomes apparent:
In “copy-paste” Scrum, development efforts, marketing strategies and sales (ROI) are not treated as constituents of the same unified ecosystem. In this scenario, it is almost impossible to fund teams by means of funding real, customer-centric products. Why? There are too many independent ad-hoc activities that take place and artifacts that are created. There is no uniformed understanding of work size and complexity that is shared by all teams. Estimation and forecasting made by each individual team is not understood by other teams. Team stability (and subsequently, cost-per-team member) is low, as individuals are moved around from project to project and shared across many projects. Further, with multiple teams reporting into different lines of management, there is a much higher chance of internal competition for budget. By the same token, there is a low chance that a real paying customer would be able to step in and influence funding decisions for any given team: too many independent and competing requests are going on at the same time.
In organizations, where “copy-paste” Scrum is seen (and is often, mistakenly taken for scaled scrum, due to lack of education and expert-leadership), there is still strong preference for fake programs and fake portfolio management. Under such conditions, unrelated activities and, subsequently, data/metrics (often fudged and RAG-ed) are collected from all over the organization and “stapled” together. All this information rolls up to senior leadership, customers and sponsors. Subsequently, what rolls down, is not dynamic funding of well-defined customer-centric, revenue-generating products, but rather rigid budgets for large portfolios and programs that are composed of loosely coupled working initiatives, performed by unrelated Scrum teams (secondary, to conventional departmental budgeting). As rigid budgets cascade down from top, onto individual teams, they further solidify the “iron triangle” of conventional project management and hinder teams’ ability to do research, experimentation and adaptive planning.
On the other hand, in Large Scale Scrum, things are different:
- When up-to-eight LeSS teams work synchronously, together (side-by-side), on the same widely-defined product (real), their shared understanding of work type and complexity (having certain scrum events together really helps!) is significantly better. As a result, when it comes to forecasting a completion of certain work (features), eight LeSS teams will do a better job than eight loosely coupled teams that work completely independently, on unrelated initiatives.
- Since all LeSS teams work for the same customer (Product Owner), there is a much higher chance that they will develop a shared understanding of product vision and strategy, since they are getting it from an authentic source – and therefore will be able to do planning more effectively.
- Having more direct correlation between development efforts LeSS teams (output, in the form of shared PSPI) and business impact (outcome, in the form of overall ROI), makes strategic decisions about funding much more thoughtful. When real customers can directly sponsor product-centric development efforts, by getting real-time feedback from a market place and deciding on future strategy, they (customers) become much more interested in dynamic forecasting, as it allows them to invest into what makes most sense. Dynamic forecasting of LeSS, allows to increase/decrease number of scrum teams involved in product development flexibly, by responding to increased/decreased market demands and/or product expansion/contraction.
Noteworthy that in LeSS Huge cases, when product breadth has outgrown capacity of a single Product Owner and requires work by more than eight teams, dynamic forecasting can still be a great approach for Product (overall) Owner and Area Product owners (APO): they can strategize funding of different product areas and make necessary timely adjustments to each area size/grown, as market conditions change.
All of the above, as described in LeSS scenario, will decrease organizational dependency on fixed budgets, as there will be less interest in outdated financial information, in favor of flexibility, provided by rolling-wave forecasting that brings much closer together “the concept” (where value is built – teams) and “cash” (where, value is consumed – customers).
Another Large-Scale Scrum Training (CLP), taught by Craig Larman in NYC, is in the CompuBox.
More than thirty people from all-around the globe (North America, South America, Europe) came together for this brain-jelling learning experience! The group consisted of product owners/managers, software engineers, managers and organizational design consultants (scrum masters, coaches and trainers) – people coming from different backgrounds and with a focus on different aspects of organizational agility. What has united them all, however, was their eagerness to learn in-depth about principles of organizational design and implications of Scrum adoption at scale in complex organizational settings.
With exception of a few rare questions/clarifications, the class spent NO time discussing basic Scrum. It was implicit (assumed) that everyone in class had strong knowledge and hands-on experience with the basic framework. On occasions, the topics discussed would bump into “…oh this is not even LeSS-specific; this is just basic Scrum…” but those cases were rare.
Not until day three,is when the class took a deeper dive into LeSS Framework and LeSS-specific events, artifacts, roles…. Why was not it done sooner? Well…
- LeSS is Scrum. It is the same very Scrum described by Ken Schwaber and Jeff Sutherland in the Scrum Guide, but done by multiple teams, as they are working together, on the same product, for the same product owner. LeSS is not “…something that IT does, that is buried in a company’s basement, under many layers of organizational complexity…”. LeSS is an organizational design that uses Scrum (team) as a building block. Understanding basic Scrum made understanding of LeSS very easy for everyone.
- The class was made of people that have completed all assigned homework (self-study), before attending. People knew what LeSS picture looks like 😉, when coming in. Everyone in class was an educated customer. Importantly: there were no attempts to change LeSS (or change training content 😊 of LeSS), to make it better fit conditions of organizations, where people came from.
- Spending the first two days on understanding system modelling techniques, differences between causation and correlation (as well as other dynamics) among many system variables, made full understanding of LeSS on day three, come more naturally.
The class learned how to see ‘the whole’/full picture of organizational ecosystem and learned to appreciate why Organizational Design is the first-order Variable that defines System Dynamics (followed by everything else: culture, policies, norms, processes, etc.)
One of my (Gene) biggest take-away points (on the top of an excellent LeSS refresher, from Craig himself), that I plan on using immediately, was the fact from history that was discussed at the beginning of the course (and, sadly, forgotten or known known by many). And it goes as follows:
…Back in 2001, at Snowbird, UT, where the group of seventeen entrepreneurs-product-developers have met and came up with what is known today as ‘Agile Manifesto’, the two contending terms to-be-used were adaptive (suggested by Jim Highsmith, the author of Adaptive Software Development) and agile (suggested by Mike Beedle). ‘Agile’ won because of the reasons that are described here. Truth be told, because the English meaning of ‘agile’ is not as intuitive is the meaning of ‘adaptive’, today, there is a huge number of fads and terminology overloading/misuse that make the original meaning of agile so diluted and abused…. As it was meant to be: Agile == Adaptive ==Flexible. We all have to be careful with the meaning of words we use, to avoid this painful irony😉.
Here are some Kodak moments from the event:
Nowadays, for too many organizations, Agile Maturity Metrics (AMM) have become a trusted way to measure improvements of agility at personal (individuals), team and organizational level.
However, it is not always apparent to everyone that AMMs are different from Agile Check-Lists (e.g. classic example of Scrum Check list by H. Kniberg) and this can often lead to problems and dysfunctions:
Check-Lists are just a set of attributes that are usually viewed on-par with one another; they are not bucketed into states of maturity (other logical grouping could be applied though)
On contrary, AMMs place attributes in buckets that represent different states of maturity, with one state, following another, sequentially.
With very rare exceptions (favorably designed organizational ecosystems), there are three potential challenges that companies face, when relying on bucketed AMMs:
1 – System Gaming: If achieving a higher degree of agile maturity is coupled with monetary incentives/perks or other political gains (for many companies that are driven by scorecards and metrics, this is the case), there is will be always attempts by individuals/teams to claim successes/achievements by ‘playing the system’, in pursuit of recognition and a prize.
Note: Translation of the text in red: “(Пере)выполним годовой план за три квартала!!!” = “Will meet/exceed the annual plan in three quarters!!!”
2 – Attribute-to-Maturity Level relationship is conditional, at most: Placing agile attributes in maturity buckets implies that attributes in higher-maturity buckets have more weight than attributes in lower-maturity buckets. However, this is not always a fair assumption: weight/importance that every organization/team places on any given attribute, while defining its own maturity, is unique to that organization/team. For example, for one team, “…being fully co-located and cross-functional…” could be much more important than “…having Product Owner collocated with a team…” For another team, it could be the other way around.
3 – Correlation between attributes is not linear, at system-level: Regardless of buckets they are placed in, many agile attributes are interrelated systemically and impact one another in ways that is not apparent, to a naked eye. For example, placing “Scrum Master is effective in resolving impediments”attribute in a maturity bucket that comes before the maturity bucket with “…Organization provides strong support, recognition and career path to Scrum Master role…” attribute, dismisses the real cause-and-effect relationship between these two variables, misleads and sets false expectations.
To avoid the issues described above, it would be more advisable to treat every identified agile attribute as a system variable, that is on-par with other system variables, while assuming that it has upstream and downstream relationship. In many situations, instead of spending a lot time and resources on trying to improve a downstream variable (e.g. trying to understand why it is so difficult to prioritize a backlog) it is more practical to fix an upstream variable that has much deeper systemic roots (e.g. finding an empowered and engaged product owner who has as the right to set priorities).
Below, is the list of agile attributes (a.k.a. system variables) that are logically grouped (check-list) but are not pre-assigned to levels of maturity (all flat). Some examples of suggested system-level correlation between different attributes are provided (cells are pre-populated).
Please, click on the image to download the matrix to your desktop, amend the list of attributes if you feel that your situation calls for modification, and then use “Dependency on Other Attributes?” column to better visualize system-level correlation between the attributes are of interest to you and other related attributes (some examples are provided).
Tonight -a great presentation by Malik Graves-Pryor of Natoma Consulting, as he shared how his company leveraged LeSS to achieve stunning results, while facing challenges and learning lessons.
At the Thursday October 12, 2017 NYC Large Scale Scrum Meetup, Malik Graves-Pryor shared his company’s LeSS case study, “Web and Mobile Applications Agile Transformation”. He covered the extensive issues the company faced at the beginning ranging from only 1-2 releases a year with hundreds of defects, and how they transformed over the course of several months to an organization that released monthly, and then continuously, with low defects and high customer satisfaction and engagement.
The discussion covered the merger of the Sales and Product Management Pipelines, adoption of technical practices leading to a DevOps-focused culture, how to take the necessary steps to build trust and cooperation within the organization, as well as the road-map they used to iteratively migrate the organization to continuous integration and deployment.
The interactive discussion spanned two hours with attendees raising questions and issues about the case study, as well as correlating them with their own challenges and aspirations.
Presentation deck is available at Natoma Consulting website for download.
Agile frameworks (e.g. Scrum, Kanban, XP), individuals’ roles & responsibilities, processes & tools, metrics & reporting, burn-up charts, estimation techniques, backlog prioritization, agile engineering practices, agile maturity models etc. – all of them are important attributes of a typical agile transformation. However, NONE of them are first-degree-of-importance system variables that are responsible for transformation success. Most of them, are good superficial lagging indicators of agility but they are all corollary (secondary or tertiary) to another much more important system variable.
What is the most important system variable that defines a company’s agility? It is Organizational Design – the most critical element of organizational ecosystem that defines most of its dynamics.
When organizational leadership decides to take an organization through an agile transformation journey (it could take years, sometimes), it [leadership] needs to acknowledge that real, sustainable agile changes are only possible if deep, systemic organizational improvements are being made. For that, leadership needs to be prepared to provide to its organization much more than just support in spirit, accompanied by organizational messages of encouragement and statements of vision. Leadership must be prepared to intimately engage with the rest of an organization, by doing a lot of real “gemba” (genchi genbutsu (現地現物)) and change/challenge things that for decades, and sometimes for centuries, have been treated as de-facto.
What does it really mean for leadership to engage at System Level? First, it is important to identify what a system is: what are a system’s outer boundaries? For example, one of the most commonly seen mistakes that companies make when they decide on “scope of agile transformation” is limiting its efforts to a stand-alone organizational vertical, e.g. Technology – and just focusing there. Although this could bring a lot of local (to IT) success, it may also create unforeseen and undesirable friction between the part of an organization that has decided to change (IT, in this case) and parts of an organization that decided to remain ‘as is’ (e.g. Operations, Marketing, HR, vendor management). For example, if Scrum teams successfully adopt CI/CD, TDD or other effective engineering practices that enable them deliver PSPI (potentially shippable product increment) at the end of every sprint, but business is not able to keep up with consumption of deliverable (too many approvals, sign offs, hand-overs red tape) then the whole purpose of delivering early and often gets defeated. Then, instead of delivering to customers soon, in exchange for timely feedback, teams end up delivering in large batches and too far apart on a time scale.
A successful Agile Leader must treat an organization, that he wants to transform, as a sushi roll. Just like seaweed alone does not provide a full spectrum of flavors and does not represent a complete, tasty meal, one single department (e.g. IT) is not sufficient enough to participate in agile transformation efforts. Other organizational layers need to be included as well in a transformation journey. On-point!: a slice does not have be too thick. In fact, if organizational slice is too thick, it might be too big to “swallow and digest”. But still, even when sliced thinly, an organization must include enough layers that constitute a complete ‘sushi meal’.
Note: A great example of treating an organization as a sushi role, while making it more agile, is Large Scale Scrum (LeSS) adoption.
So, what are some key focus areas that every Agile Leader must keep in mind, while setting an organization on agile transformation course?
- Location strategies. Geographic locations.
- HR policies (e.g. career growth opportunities, compensation, promotions)
- Budgeting & Finance
- Intra-departmental internal boundaries and spheres of influence
- Organizational Leadership Style
- And some other areas that historically have been considered as …untouchable…
All the above listed areas are defined by Organizational Design and can be better understood through self-assessment, done by organizational leaders at all levels.
When people go to the doctor’s office, they often complain of superficial manifestations of a problem (i.e., “chief complaints”) that don’t present any serious concerns, but in reality, are indicators of a much more serious systemic illness.
I’ve come to the conclusion that we in the Scrum field are also dealing with a similar type of scenario.
For those of us who have been coaching for a while, it’s probably a common experience to have a team member or a manager come to us with what appears to them as a “key problem,” but in reality, turns out to be a symptom of a much more serious underlying dysfunction.
On June 1, 2015…(read more from the original post)
When we ask an experienced Scrum developer what defines a good User Story, the answer we hear almost immediately is that “…every user story must be INVEST-able…(taken from B. Hartman’s post)”.
When we further elaborate on “INVEST” part, we hear that splitting user stories should be done Vertically (along features), not Horizontally (along components or applications layers). Why is the latter condition so important? Because when split vertically, and cross-cutting through multiple components, a user story has a much higher chance of representing a potentially shippable product increment (PSPI). Delivering, UI/UX alone, or business layer alone, or database alone, does not present value to a buying customer.
Frequently, we hear the analogy of a sushi roll, when describing story slicing: “…every User Story must represent a thinly sliced sushi roll that provides taste and flavor of multiple layers (caviar, seaweed, rice, tuna, avocado, etc)…”
…Now, imagine an organization that undergoes agile transformation. Predominantly (statistically), transformation efforts stem from Technology, where agile improvements come in the form of introducing good engineering practices, such CI/CD, TDD, unit testing, test automation, automatic deployments, etc. But technology alone, is just one layer of an organizational sushi roll. Sure, just like seaweed alone, it may be tasty enough for starters, but without adding additional flavors, it is not a complete, tasty meal.
Other organizational layers need to be included as well, when identifying a slice for agile transformation experiment. A slice does not have be too thick. If organizational slice is too thick, it might be too big to “swallow and digest”. But still, even when sliced thinly, an organization must include enough layers, to be considered as complete meal.
What are some of those layers? Let’s consider a few:
Business & Operations: it is imperative to have real customers intimately involved in agile transformation efforts and make them closely aligned with technology partners. This requires identifying and providing a strong support for some key agile roles, such as product owner, stakeholders and SMEs. Often, this requires organizational realignment and changes to sphere of influence.
HR: Subjective monetary rewards that are based on individual performance assessments and fueled by invisible internal competition among employees, must be discontinued, in favor of incentives that promote teams’ collective ownership and performance. An organizational slice that wants to become more agile will have a much higher chance for success if HR policies were genuinely supportive of the initiative of the efforts.
Budgeting & Finance: For agile teams (Scrum, Kanban) that use adaptive planning and iterative development, and continuously have their work re-prioritized by business, it is imperative to have flexible budgets (“flexible spending”). By unlocking a rigid budget corner of what is known as Project Management Iron Triangle (budget, scope, timeline), technology dramatically increases their chances to do research or conduct new experiments, if an opportunity presents itself (often, unexpectedly). There is also a higher chance that more Scrum teams would be built out faster, if budgets were flexible and not done as ‘budgets against the wall’ (‘the world ends on December 31st’) – more flexibility to acquire new talent.
Vendor Management: Relying on same-old vendors-partners that are just convenient to engage because they are listed at the very top in a vendor management system, may no longer be viable. Engaging vendor companies on “fixed-everything” SOWs, while expecting them to work in an agile fashion may create tension and friction. Considering vendor-workers and a part of an agile team (e.g. Scrum, Kanban, XP), yet still treating them as ‘external workers’ and structuring communication with them, by going through engagement managers and ‘on-site leads’ would defeat the purpose of teaming, diminish transparency, collaboration and knowledge sharing, as well as would be un-supportive of healthy dynamics and interaction, expected of agile teams. Organizations should apply a different set of standards and benchmarks, when selecting vendors that are expected to engage in agile work (e.g., development, installation, configuration, etc.). Contracts and SLAs should be supportive of agile (adaptive) ways of working.
Real Estate & Facilities: Having agile teams co-located under the same roof is a huge win: e.g. Scrum team dispersed around the globe is not as good as Scrum team placed on the same floor. (Note: Please, do not confuse a single Scrum team spread thin across multiple locations with multi-site Scrum product development, when many, whole teams are collocated but separated from peers-teams by geography).
However, putting a team on the same floor is not sufficient. Interior design must be supportive of team collaboration and dynamics: ‘caves & common’ (read A. Cockburn about XP), information radiation techniques (lots of whiteboard space, flip-charts), breakout areas, extra space to accommodate extended user community during sprint reviews, etc. – is all required. It is unfortunate but it is pretty common to see so-called Scrum team members sitting in a long single row, next to each other, spear-headed by a line manager (usually, a window seat), all joining a daily stand-up call by phone (while sitting!) and ‘reporting on status’, while staring at an electronic story board on their respective monitors. Agile teams don’t want that.
As a summary, please consider the following quote that describes sushi-roll-like organizational design in Large Scale Scrum (LeSS), by C. Larman: