Alex Ballarin Latre

Este artículo es la segunda parte de otro donde hablaba de como integrar el trabajo de los especialistas de UX y los desarrolladores de software.

Tanto el desarrollo de software Agile como el diseño de experiencia de usuario (UX) tienen como objetivos comunes entregar antes sistemas más usados y valiosos para el cliente.

In this article I will stress again the utmost importance of understanding the context of the organization as a strategy to achieve a possible, evolutionary and deep application of Scrum within organizations.

In this post I will review what's a Proxy Product Owner, why you can find so often this role in organizations embracing Scrum, and some proposals to avoid this role and achieve a true Product Owner. What is a Proxy Product Owner? A Proxy Product Owner (Proxy PO) is a middleman role between the people taking decissions about a product and the people developing it. A Proxy PO usually perform activities that are usually done by a Product Owner, such as: Gather the customer needs. Define and order the Product Backlog. Plan how to realize the Backlog together with the team. Decide when the product increments can be released to the customers. However, a Proxy PO is an incomplete version of a Product Owner. This seriously undermines its efectiveness and makes the value-maximizer mission of a PO difficult to achieve. The main shortcommings of a Proxy are typically: It is not the owner of the product! So neither takes the main decissions about the product nor is truly accountable for its success. Does not control the product's budget. Does not define neither the product's vision nor its strategy. Does not have the last say over the Backlog and its items. So, although having a Proxy PO may enable a Development Team to have a stable demand management, this role does not optimize neither decission-taking nor value management of the product, specially in big and complex organizations. Functional organizations and the Proxy PO Once upon a time there was an organization grouping employees according to the type of work they did and to their capabilities. This business groups receive several names such as business areas, departments or teams, being the most frequent organizational model. This model is also a heritage of Scientific Management or Taylorism, introduced by Frederick Winslow Taylor in early twentieth century. A typical business organization divides their workers between the "business" and the "IT" people, subdividing those areas in smaller function-specific groups such as sales/marketing/HR and Dev/Ops/QA respectively. IT traditional roles such as Business Relationship Manager, or Business Partner, prioritize the business demand, and another IT role such as Project Manager leads their realization. When it comes to embrace Scrum, those organizations without a good understanding of the Product Owner role oftenly identify that role as belonging to the IT department which is set to maintain the communication between the business users and the Development Team. In addition, this role is expected to analyze the backlog items and even define their acceptance criteria. This set of responsibilities fits quite well those of a Project Manager, who also may see himself closer to a Product Owner than to a Developer or Scrum Master. For all this reasons, Proxy PO easily emerge from the IT department. Furthermore, business people which have the "MAN" (Money, Authority and Need) are usually too busy to also deal with time-consuming activities such as detailed backlog management and requirement definition. That may makes this new role not so attractive to those business roles, which could be the ideal Product Owners. So that, in Scrum adoptions we oftenly have all the ingredients to cook a good Proxy PO recipe. Proposals to achieve a true Product Owner First of all, it is essential that organizations those helping them to embrace Scrum understand deeply the Product Owner role. Even more difficult, they should understand the severe organizational redesign this role demands both in the business and the IT areas. IT should not be an "internal IT provider" any longer. Secondly, it is also important that everybody understands that a true Development Team should be self-sufficient so it may carry out large parts of elaborated backlog management activities, such as refining, item definition and acceptance criteria. This delegation from the Product Owner to the Development Team should be made on the principles of transparency and trust. The Product Owner should remain confident on being informed and responsible for taking the most important decissions about backlog management and item definition. Thirdly and lastly, finding a good fit for existing project managers within the new organization shouldn't encourage them to be Proxy POs. If they are experts in team management, who else may help a Development Team to be self-managed as another team member, but avoiding the micromanagement and working in another value-adding activities within the Sprint.

I have read in several Lean and DevOps sources that the Sprint events imply a kind of waste, so teams are supposed to move to a Continuous Delivery or Kanban lifecycle as they become mature. This post is to discuss that is simply not true, regardless what lifecyle each team chooses as their favourite one. Scrum theory roots the Sprint design As you can read in the Scrum Guide and in many posts the Scrum.org blog, Scrum is based on the empirical process control theory, or empiricism. Complex problems are better solved by self-managing and self-sufficient teams in short feedback loop iterations, where those teams can inspect and adapt whatever needed as they discover information about the problem being solved. Events should not be judged just by estimating their efficiency in terms of how many team member's time they take. Their outcome is much more than that. Sprints and their events are not just "meetings to plan and track work", but opportunities to share points of view, to remove erroneuous assumptions (and hence wasteful work later) and to come up with the best possible decissions (because they are rooted in the collective intelligence and not the individual's capacity). The House of Scrum by Gunther Verheyen. Beside the decission optimization, Sprint events also contribute to the growth of the team by the co-learning and with the development of a true team conciousness and sound relationships among its members. The Aristotle Project by Google showed that a team where the decission-taking and communication is balanced works better than other more talented but less participative. Focusing on individual team member performance or backlog items throughput (e.g. velicity) may be counterproductive. A good team's outcome is more than the sum of the outcomes of their members. The product's outcome is more than its output (e.g. lines of code, number of resolved tickets). Waste according to Lean In Lean thinking, waste means uneeded work which adds not value to the product. Tom and Mary Poppendieck identified these seven sources of waste in their book Implementing Lean Software Development. Inventory: Unfinished goods (also called as "work in progress," or WIP) Overproduction: Producing more than the demand requires Extra processing: Additional steps in the process that aren't really needed Transportation: Shipping the goods from one place to the other Waiting: Lag between process steps Motion: Moving around within the process Defects: Flaws in the deliverables that impact their features/functionality Let's use these sources as a definition of waste and find out how could the Sprint events become wasteful. Why could events produce waste? If a team does not understand correctly the Scrum or fails to take advantage of the continuous improvement provided by retrospectives, events could become sources of waste. Some examples according to Poppendieck's sources of waste, underlining possible flawed events, are: Inventory: teams do not track sufficiently the Work In Progress during the Daily Scrum and end the Sprint with "undone" work. Overproduction: Development Team and Product Owner fail to forge a good Sprint Goal during the Sprint Planning and to undestand together what are the priorities for a Sprint. Extra processing: teams fail to use Retrospectives to identify uneeded steps, or steps that could be automated. Transportation: teams may not be self-sufficient and PBI may have dependencies from external sources. This may not be indentified or tackled enuogh during Retrospectives. Waiting: teams use the Daily Scrum as their only Synchronization mechanism and hence, they wait to them to take decissions. Also item's dependencies may not be identified previous to the Sprint during refinement, or during the Sprint Planning, so items could be blocked within the Sprint. Motion: during Sprint Planning, team members may not understand sufficienly which is the team's plan to achieve the Sprint Goal and Backlog, and later they may fail as well to optimize the Sprint activities during the Daily Scrum. Defects: teams do not use the Definition of Done to identify items that are not ready to be released into the Sprint's increment, and that may not be identified during the Sprint Review and fixed within the Sprint Retrospective. Agile it's not all about speed but value The goal of this post was not to compare in detail neither Scrum and Lean nor Scrum and Kanban. It was not also to say which is better. This goal aims to remove the mistaken idea that Sprint events are wasteful and getting rid of them increases team productivity. The term agility too often is confused with speed. In my opinion, the ultimate goal of Agile is not to produce more (output) but to produce more (outcome).

Most of the organizational change initiatives focus on the "hard" aspects such as organizational design, process or tools. Although those are critical, often other "soft" aspects such as values, principles or culture are overlooked. As the iceberg metaphor: consider them or they may sink yout Scrum adoption!

This is a very common myth, frequent on people used to develop software only within the context of a closed scope (traditional project). The Scrum framework is agnostic when it comes to set the context of software development; it just talks about “complex product development”. In general, agile software development avoids this concept of “project”. The iron triangle Software development is an activity that is frequently outsourced to IT companies with technical and domain know-how. Most of the times those collaborations are shaped by a “project contract” based on the iron triangle model. This model identifies four aspects of the software development endeavor that are tightly related: the product scope, the delivery time, the contract cost, and last and frequently overlooked, the product quality. It happens so often than later those constraints become impediments in case the contract was not accurate (the bigger the more likely) and changes should be agreed to succeed in the project. Why is this? The main reasons to continue using fixed-scope projects include: Distrust between the customer and the software supplier Corporate aversion to risk “We always did it this way” The last available edition (2015) of the famous Chaos Report by Standish Group shows than less of 1/3 of projects is considered successful, so people buying software projects would do well in considering why to continue doing it. Trying to maximize all the four aspects of the iron triangle at the same time usually entails losing out in all. Agile software development generally changes this “fixed scope project” approach, fixing resources, time and quality, and estimating scope. Due to the transparency and prioritization of the features provided by the product backlog in Scrum, the probability of having a troubled project decreases significantly. Unused software Another source of criticism for fixed scope projects it that they produce software with many unused features. The classical “ROI, it’s your job” study by Jim Johnson reveals that 64% of features delivered by projects are rarely or never used. Although the sample of projects is very small, my gut feeling tells me that reality may not be very different from that, which is a big waste of money and people time. Planning a project using Scrum? Projects can be planned with Scrum. From the point of view of product management, this is usually called “release management”. The scope management tool is the product backlog, and it can be estimated before the development starts. Scrum calls for empirical process control to avoid waste due to requirements identified too early, so the lightest scope definition should be made which creates trust with the product buying part. The “transparency leg” of empiricism entails being radically honest with the customer about the risk of not delivering the full scope in the expected timeframe and eventually allocating a reasonable contingency, e.g. in the form of extra sprints. The Product Owner is responsible for involving the stakeholders in the progress tracking, and helping them to shift from a “original scope mindset” to a “value delivered mindset”. Closing Scrum can be used to plan and track development with a fixed scope, although some other common beliefs about the traditional project management such as the concept of a fixed scope project should be revisited to help customers to evolve into a healthier product management mindset.

One common consequence of teams that do not deeply understand Scrum and the nature of its events is that they believe it is possible to run sprints which do not produce a Done and releasable increment of the product. This belief typically leads to dangerous consequences so it’s important to caution about them and to review the basics of what is a Sprint. A Sprint Goal is always to produce a Done increment. The Scrum Guide starts defining a Sprint with this sentence “The heart of Scrum is a Sprint, a time-box of one month or less during which a ‘Done’, useable, and potentially releasable product Increment is created”. Why is that? A Sprint is based on the empirical process control theory to optimize the predictability and the risk control of the development activities. The three pillars of this theory are transparency, inspection and adaption. Producing a Done increment it’s the best way to create transparency about the ability of the team, within its context, to bring value to the customer by delivering that usable and valuable increment of the product. Inspecting the outcome of the Sprint unveils possible improvements to both the product and how the team worked. That inspection allows adapting the course of the development regarding the product, the team and its environment. Therefore, Sprints not designed to produce a Done increment undermine the empiricism of the development and do not meet the basics of Scrum. Let’s review some popular “undone” Sprint types. Sprint 0 A Sprint 0 is the name often given to a short effort to create a vision and a rough product backlog which allows creating an estimation of a product release. There is nothing bad in doing that as long as everyone concerned has clear that the plan is subject to change as more is known as a result of the inspection of the sprints outcome. To sum up, that activity does not meet the definition of a Sprint in Scrum, so it is better not to call it so. Design Sprint A Design Sprint is the name given to a “Sprint” dedicated to create a functional design or a technical architecture to guide the rest of the release. This is a “worse” case than the previous because, on top of not meeting the definition of a sprint, it narrows or even impedes the needed inspection and adaption of the following sprints. Again, doing a lightweight functional and architectural design as a part of the product vision can be helpful to avoid risks and put all stakeholders on the same page, but do not create a detailed and final design. Hardening Sprint A Hardening Sprint is to me the most worrying interpretation of what a Sprint could be. This concept was widespread by a popular scaling framework that partially changed their mind about this in its last version. A Hardening Sprint goal is achieving a releasable and integrated increment that essentially could not be achieved before. That means that it is accepted that sprints can create undone increments, so the ability of teams to create transparency, inspect and adapt are seriously jeopardized, and therefore, no real sprints are being performed. Closing In this post we saw why a Sprint must always produce a Done increment, some popular myths about “special sprints” and some consequences of ignoring the true nature of a Sprint.

I would like to kick off a series of posts in this blog trying to debunk some common myths about Scrum. Many of them arise sometimes from a poor understanding of the Scrum Guide, and even more often, from not having read it at all. What is velocity? According to the Scrum.org glossary, Velocity is “an optional, but often used, indication of the average amount of Product Backlog turned into an Increment of product during a Sprint by a Scrum Team, tracked by the Development Team for use within the Scrum Team.” A couple of posts in this blog, by Derek Davidson and Hiren Doshi, show velocity as a useful metric to forecast how many items can be delivered in a sprint or how many sprints may be needed to achieve a release. Gunther Verheyen makes a couple of interesting points about velocity in this post. The first about its transparency: if the increment is really undone, velocity is not a valid indicator. The second is about trying to see velocity as a commitment: in an empirical process control, we accept that what will happen is unknown. What is value? Value is a more general concept. According to the Oxford Dictionary is “The regard that something is held to deserve; the importance, worth, or usefulness of something”. Let’s consider that money is a proxy of value in most circumstances, and therefore a valuable product should increase revenue or decrease costs for the organization that uses it. In the context of a Scrum Team, value is only created when the product (increment) reaches the customers, it is useful to them and they start obtaining benefits from using it. Measuring value is not a trivial task. Value indicators need be identified and they should be aligned with the product vision and strategy. Good places to learn further are the Professional Scrum Product Owner course or the Evidence Based Management approach. Once value indicators have been defined, they can be used to estimate the value of individual Product Backlog Items, and therefore the value created by a product increment could be calculated as the sum of the value of the PBIs included in the increment. Tools such as Value Burn-up are useful ways for the product owner to estimate the value delivered. However, the real value can only be measured once the product has been released and is being used. So, does velocity = value? As previously described, velocity and value are very different things. The first is a tool that can help the Scrum Team to self-manage during the product development, while the latter is a way for the customer to evaluate the usefulness of the product already being used. Closing: a worrying interpretation A worrying interpretation that derives from confusing velocity and value could be “the more software is released, the more value is created”. A product backlog item that is not useful to the customer, even if their designers think so, does not create value. A product backlog item released with defects or not meeting the Definition of Done can even create negative value, because it can undermine the trust of the customer in the product, disrupt their operations and also produce unexpected work in the future, such as bug fixing or extra work to build something on top of a bad quality product. Making these concepts clear to all the stakeholders of the product, and especially to the product owner, could help avoid falling into the vicious thinking of “the more is always the better”. As Jeff Patton puts it in his book User Story Mapping: Minimize the output (code size), maximize the outcome (value).