Shift Left in the SDLC (Software Development Life Cycle)
Laia López22:032023
Wallapop cut definition bugs 55% after making quality the whole team's job
Developers at Wallapop once had a safety net: they knew a tester would check everything afterwards, so edge cases slipped and fixes arrived late and expensive. Frontend engineer Laia López and three Wallapop colleagues explain why they moved testing to the start of the cycle and turned QA into risk assessment and requirements work. What does each role actually do differently when quality stops belonging to one team?
See how product, QA and developers each change their daily work when testing shifts left
Laia López and her Wallapop colleagues Agustina Santos, Anna Ruiz and Rebecca Porras explain how they moved from a traditional, phase-based software development life cycle to a shift left approach where quality is owned by the whole team. They describe the problems that prompted the change, how product, QA and development roles each adapted, and the benefits and challenges they found.
“Developers had a safety net, an insurance plan, because they knew that a tester would review the implemented product afterwards.”
“We have stopped seeing testing as a stage and we are now seeing it as a series of interventions that can and should be applied during the whole software development lifecycle.”
What you will learn
Distinguish a traditional, phase-based testing stage from shift left testing as a series of interventions across the whole cycle.
Spot the warning signs that call for shift left: QA bottlenecks, late feedback, low automation coverage and quality owned by one role.
Plan how product, QA and developers each change their responsibilities, from three amigos requirement sessions to developers testing their own code.
Measure the effect of shift left by classifying bugs by root cause and tracking definition bugs over time.
Anticipate the main adoption risks, such as blurry role boundaries and teams that need time to leave their comfort zone.
Key takeaways from the speaker
Shift left mindset
experimentation culture
SDLC
analytics
software developing
Details
Level
Intermediate
Format
Panel
Technical depth
Introductory technical
For
Early career, Mid-career, Senior individual contributor
Length
22:03
Published
June 2023
Questions this talk answers
What is the shift left approach in software development?
It is an approach where quality and testing practices happen earlier in the development life cycle, in planning, design and build. Testing becomes a series of interventions across the cycle and quality is owned by the whole team rather than a single role.
Why did Wallapop decide to adopt shift left?
They faced waits, bottlenecks and dependencies because developers waited for QA before merging, feedback came late and automation coverage was low. They also wanted to release faster, run more experiments and stop treating QA as the only owner of quality.
How does the QA engineer's role change under shift left?
Manual validation of each feature drops, and QA engineers focus on acceptance criteria, test plans, strategic thinking, requirements definition and risk assessment before coding starts. Remaining manual testing is mostly exploratory, and they spend more time analyzing quality metrics.
What results did Wallapop see from shift left?
The team said bugs caused by a lack of definition fell by 55% compared with the previous year, based on its internal classification of bugs by root cause.
What are the downsides of shift left?
Role boundaries can become blurry, teams used to working in a specific role need time to adapt, and test documentation has to be more exhaustive because the whole team now performs it.
About this talk
We are a group of 4 engineers working at Wallapop and we would like to discuss how has been our journey in applying the Shift Left methodology in our daily basis.
We would like to cover different ways of delivering, such as feature launches, A/B tests, smoke tests, releases and how that affects all different roles involved in the process.
55% fewer definition bugs, and developers beyond code (52 s)
Speaker 1: Okay, we'll start. Hello, everybody. Hello, dear friends from Women in Tech Network. We are very happy to begin a talk here among such amazing lineup of speakers. Today, we will be speaking about ShiftLeft in the software development lifecycle, an approach that we have recently implemented in our company. Speaker 1: First of all, we'd like to introduce ourselves. We are a group of engineers currently working at Wallapop. We have a crazy good frontend engineer in the group, Laia López, a rockstar data analyst, Agus Santos, and two unreal QA engineers, Rebecca Porras and myself, Ana Ruiz.
Speaker 1: In this index, I'm going to give you an overview of what we will discuss today. Speaker 1: First of all, we will review the traditional model of the software development lifecycle because to go forward, you need to know where you come from. Speaker 1: Then we will get to know the definition of the ShiftLeft model. After that, we will give you a sneak peek of what Wallapop is about as context.
Speaker 1: Following that, we will see how we implement the ShiftLeft model in our current organization. And lastly, as a way to wrap this presentation, Speaker 1: we will see an overview of which were the ups and downs we got from this approach. Speaker 1: So let's get into it. And now August will tell us all about the traditional approach of the lifecycle.
Speaker 1: Hi, everyone. Speaker 1: So when we talk about traditional lifecycle, we talk about the software development lifecycle. That is a methodology that provides a framework for software creation. Speaker 1: We devise the software development process into a series of steps that can be assigned, completed and measured. Speaker 1: If we go through each of the steps, we first have the strategy phase that involves project planning, where discoveries and analysis are done in order to define goals.
Speaker 1: Also, the requirements needed to achieve the plans are gathered and defined, together with the estimation of cost, timing and resources. Speaker 1: In step two, we have the design phase. This phase involves analyzing requirements and identifying the best solutions in order to develop the software from the UI and engineering point of view. Speaker 1: Usually, product and tech refinements are done in this phase, where teams look for the best ways of integrating features within the current infrastructure.
Speaker 1: In step three, we have the development phase, where engineering teams code and develop all the requirements defined in previous phases through sprints. Speaker 1: In step four, we have the testing phase, that this phase is also known as the regression phase, where QAs are involved in testing for quality analysis, where they check the correct functionality of the feature and if customer requirements are met.
Speaker 1: This is partly combined with the development phase, where developers also test automatically their work. Speaker 1: In step five, we have the deployment phase that involves the launching and production of the feature and monitoring its correct functionality. Speaker 1: And last but not least, we have the maintenance phase, which is a continuous phase where monitoring takes place.
Speaker 1: But it's not only for resolving customer issues or fixing bugs, but it's also, it's an opportunity to find new ways to improve existing features. Speaker 1: To conclude with this approach, we wanted to share some pros and cons of this methodology. Speaker 1: So from the pros point of view, this methodology provides a framework where a problem or a goal is defined.
Speaker 1: Together with proper roles definitions, as well as inputs or outputs of each step in order to have a clear perspective of what is needed to reach the final launch of the feature. Speaker 1: But there are also some cons, such as not being a cost effective or time efficient process, where risks are not minimized through forward planning, which usually ends up being reactive to bugs.
Speaker 1: So now that we have introduced the traditional approach, I will hand it over to Rebe to tell us about the shift left. Speaker 1: Thanks. Speaker 1: Let's jump right into seeing what the shift left approach is. Speaker 1: The shift left is an approach where quality and testing practices happen earlier in the development life cycle.
Speaker 1: If we see the different phases, we will see that the shift left occurs in the planning and design stage and also in the develop and build stages. Speaker 1: Whereas the traditional approach that I was just mentioning has a testing stage and it's mainly focused on the deploy and release and monitor and analyze. Speaker 1: It is important to mention that the traditional approach has a testing phase, whereas in the shift left testing are a series of interventions along the whole cycle.
Speaker 1: If we see the attention of detail that both different models have, we see that the shift left has a higher attention to quality because quality is not owned by a single role anymore is owned by the whole team. Speaker 1: So the whole team is accountable for quality. Speaker 1: Therefore, it enhances the attention. Speaker 1: Then, if we see the cost factor of error correct correction, we see that in the shift left, it's way less costly to fix the errors in advance, rather than in the traditional test approach where issues are spotted at the end of the cycle and the cost of fixing them.
Speaker 1: It's much higher. Speaker 1: Now, this being said, let's see which are the attributes of the shift left approach. Speaker 1: If we see the focus on continuous quality and whole team ownership, this is because now quality is owned by the whole team. Speaker 1: Then it happens earlier and shifting left reduces risks and many issues are addressed before they are committed to production.
Speaker 1: It enhances the continuous collaboration of the team. Speaker 1: It encourages everybody in the team to be communicative and to be collaborative. Speaker 1: We understand the product and understand the behavior of the feature that is about to develop much deeper and also much earlier. Speaker 1: We continuously test from the start with this approach, and this approach invests in issue prevention over issue resolution.
Speaker 1: There are different dynamics put into an assessing risks and also mitigating these risks earlier in the in the life cycle. Speaker 1: If we go to compare the different models. Speaker 1: It is important now to sit and give an overview on what's the shift right approach and the or the traditional approach, as I was was mentioning.
Speaker 1: And this approach focuses quality at the end and also every face is very staged, but it can be a really good complement for the shift left approach that we are giving because shift right allows us to be involved in monitoring a user's behavior. Speaker 1: We can analyze business metrics. Speaker 1: We can also use it as a stage for running some tests such as performance and security metric, or we can actually test some deliberate failures.
Speaker 1: So it can be very good complementary to the shift left approach. Speaker 1: Now let's see the shift left approach and which in here, as we have stated and testing is an activity is not a phase anymore. Speaker 1: It's a series of interventions, and these allow the team to be testing early and often, and we need a really open minded team since the communication go beyond the road.
Speaker 1: The objective is to detect problems earlier and we avoid problems before they happen. Speaker 1: And we we can say then that the pros and the of these model at the communication is improved. Speaker 1: There is constant feedback. Speaker 1: We have risks mitigation. Speaker 1: We boost the software efficiency. Speaker 1: We produce faster.
Speaker 1: And we catch back much earlier. Speaker 1: On the cons and there might be some blurry limits of the of the roles, since we keep encouraging everybody to go beyond their scope. Speaker 1: At the beginning, some lines might be blurry and we might need to put some efforts on this. Speaker 1: There is a drastic culture in in the team narrative, and it might not be easy.
Speaker 1: Teams are comfortable working in the in the way they have been working traditionally, so it might require some time to accommodate to this new shift. Speaker 1: And in the another point to to state is that the test documentation needs to be a bit more exhaustive, since it won't be performed by single roles anymore. Speaker 1: It will be a documentation performed by the whole team, so it needs to be much clear.
Speaker 1: And this being said now is going to give us an overview on what what a pop is. Speaker 1: Okay, thank you. Speaker 1: Hello, everyone. Speaker 1: So now that we saw what the shift left is, we are going to talk about what is well above and how we are organized inside the company. Speaker 1: Here we have an example and to explain a bit what is well above.
Speaker 1: We are a platform on that here in Spain for buying and selling unique goods to leave them a second life, and you will be able to do so from our apps for iOS and Android. Speaker 1: But we also have a website where you can use the shipping method with our purpose. Speaker 1: We want to encourage and promote sustainable consumption and encourage also a more dynamic and circular economy.
Speaker 1: As a summary, our users earn money and we take care of the planet. Speaker 1: And now that we saw it, let's see what the organizational model and how we are organized inside the teams. Speaker 1: Basically, we are organized by topic teams, and maybe you are wondering right now what is a topic team. Speaker 1: So it is a multidisciplinary team that have a ownership of a part of the company or the product.
Speaker 1: And you can see on the drawing that we have on the left that inside every topic, we have engineers, analysts, marketing people, and so many other roles involved. Speaker 1: It will depending on the necessities that we have in every team. Speaker 1: It's really important also to note that the teams do not work alone, or it's related because we collaborate with each others on our current basis.
Speaker 1: And now, Anna will introduce us how we indicate the shift left inside the company. Speaker 1: Thanks, Leah. Speaker 1: So before applying shift left in Wallapop, we had several issues that we wanted to mitigate. Speaker 1: We had weights, bottlenecks, and dependencies. Speaker 1: Developers had a safety net, an insurance plan, because they knew that a tester would review the implemented product afterwards.
Speaker 1: And because of this, they wouldn't focus that much in testing and in possible edge cases. Speaker 1: Since testing was a phase that came after development, there was a bottleneck effect, as developers had to wait for QA to review the feature in order to merge to master. Speaker 1: And also feedback came late, was slow because of the weights, and therefore fixing errors was more expensive.
Speaker 1: Another issue to mitigate was the workload on manual testing after development. Speaker 1: QA engineer was mainly focused on validating regression testing, so they wouldn't focus on other important aspects that happen in the defining stage. Speaker 1: Since testing came later and after development, automation was not so much in the picture, so automated testing coverage was low.
Speaker 1: Another aspect that made us go for the shift left model is that we wanted to release faster and experiment more. Speaker 1: Releasing faster means that we would be able to do more experiments. Speaker 1: Since we wanted to do more innovation, we needed to get more user insights and select or discard product variants based on that.
Speaker 1: With the old approach, experimentation would take more time and therefore user insights would arrive later. Speaker 1: And last but not least, we wanted to improve quality. Speaker 1: Since testing was a phase, the team had the belief that the testers or the QA team were the only ones responsible for quality. Speaker 1: This caused that we were finding much more errors in production or during the regression done before releasing, which are the two stages where bugs are more expensive to fix.
Speaker 1: Since risk analysis was not so important in the software development life cycle would realize of the present of some risk by the end of the cycle, which made it impossible or very difficult to react over them. Speaker 1: So now we've reviewed all the problems that we had before we wanted to improve. Speaker 1: Now here is where the shift left model comes in.
Speaker 1: I want to make a disclaimer here by saying that, as I mentioned, we have several topics in topic teams in the company. Speaker 1: And while this model is being implemented by most of them, teams have different levels and different velocities. Speaker 1: Some teams have adapted really quickly, and while some others are having more trouble.
Speaker 1: This being said, now we will take a look to the solutions from each stack's point of view. Speaker 1: We will start with product and then we will see QA and then development. Speaker 1: I was willing to start. Speaker 1: So from the product point of view, the shift the application includes early collaboration between team members, especially between product and product.
Speaker 1: Tech in the first stages of the cycle, such as strategy and design, as mentioned before. Speaker 1: This means having ideation workshops involving the engineering teams from scratch to share their deeper knowledge from the tech side. Speaker 1: Having experiments and tracking sessions when they apply and defining the requirements through three amigos approach and their proper review by the whole team.
Speaker 1: This helps translate it into robust communication between all the team members through the whole process. Speaker 1: Giving a feeling of joint ownership of the feature to be developed, being able to follow up metrics related and understanding the impact of their work. Speaker 1: That being said, I handed over to Anna to talk about shift left allocation in the QA team.
Speaker 2: Thank you, August. Speaker 2: So the QA point of view, we have seen three important points where our ways of working have improved. Speaker 2: First in the testing area, then in the collaboration and in the communication between team members and then in the scope in terms of where the QA engineers put their efforts. Speaker 2: In the testing side, we have seen how manual testing workload has been reduced.
Speaker 2: We have stopped validating each feature after development and have put more effort in defining acceptance criteria, test plans and all of these before anybody gets to coding. Speaker 2: This has led us to explore more as most of our manual testing is now basically exploratory. Speaker 2: From a cultural perspective, we've seen how the team has evolved their thinking.
Speaker 2: We have stopped seeing testing as a stage and we are now seeing it as a series of interventions that can and should be applied during the whole software development lifecycle. Speaker 2: As for the collaboration collaboration and communication we've seen an uplift in conversations between QA and development regarding testing practices. Speaker 2: We've also seen a deeper knowledge on business impact of our deliverables and also we've seen how the shift left approach has improved leadership and communication skills for all team members.
Speaker 2: As one of the basis of this approach is that communication between all of the team members needs to be higher and sooner. Speaker 2: And in the last point, we've seen a change on our scope. Speaker 2: QA engineer fewer is less about checking and verifying and is more about a strategic thinking requirements definition and risk assessment.
Speaker 2: As the approach name suggests shift left is about bringing quality practices to the left. Speaker 2: So it just makes sense that the QA engineer is more focused on these activities, which happen to be on the left of the software development lifecycle. Speaker 2: Since the figure figure is less about manual testing, we can also spend more time on analyzing quality metrics and providing further insights to keep the lifecycle active and keep iterating it and improving it.
Speaker 2: For example, we've seen a 55% reduction in definition bugs. Speaker 2: What does this mean? Speaker 2: So internally internally we classify bugs by root cause and bugs that were caused by a lack of definition have decreased by so far by 55% this year compared to the data that we have from last year. Speaker 2: And this is it from the QA side.
Speaker 2: Now Laia will explain to us the development point of view. Speaker 2: Okay, so I would like to start this part by commenting that for me, the most important thing about the shift left implementation is that the developers are not only limited to code. Speaker 2: Because we participate throughout the entire process of creating a feature like we can see on the explanation.
Speaker 2: If we take the left drawings, we are part of the product meetings, providing our ideas and investigating and also raising our hand in case that we see any technical limitation in the future development. Speaker 2: And this is very important for preventing a future mistakes. Speaker 2: So once the development has started, we test our own code with unit test and sometimes also we do integration test.
Speaker 2: On other hand, we do our manual regression in the better environment to take that everything works as we want or we plan. Speaker 2: This delivery methodology, we think that it's really agile because we make many small deliveries instead of one really large. Speaker 2: To try to give value to the users as fast as we can.
Speaker 2: So the sooner that we launch a release, the sooner that we can start monitoring user metrics and see if they like the new implementation or if we need to change any part of the product that we just created. Speaker 2: To close this whole cycle and it's essential to create technical documentation if it's necessary to give visibility to other topic teams or maybe we just want to create it for our own topic team.
Speaker 2: And now, as the talk is almost over, let's take a look a brief summary of the benefits of the shift left presented by this work model. Speaker 2: As we have seen all the roles are involved through the entire future creation process, so we are forced to say that way to understand what is being done at all the times working this way.
Speaker 2: We provide high value from a start to finish from all the team members it's sequential. Speaker 2: As you can see all the benefits are directly related because if we pay more attention on design plan and investigation and stages. Speaker 2: We are moving from reaction on bugs to predict them, so we are going to introduce less bugs and we are going to deliver features faster.
Speaker 2: A really important side note is that small deliveries and problems identified in advance are cheaper, but also faster to solve. Speaker 2: And now we are going to see the things that maybe are a bit complex when implementing shift left in the companies. Speaker 2: At least when we are starting sometimes it's a bit difficult for teams that maybe are not really cohesive or that they are really used to do a very specific role.
Speaker 2: And not getting out of it, but we think that if it's a cohesive team that collaborates well, or maybe a new one, and there should not be conflict, even if at the beginning is out of the comfort zone and the team members need time to adapt to that shift left. Speaker 2: On other side that we have on the right.
Speaker 2: Like we mentioned before, if the roles are not well defined, it can confuse the people involved with so it's really important that the scope is close it and everyone have clear out what they should and what they should not do. Speaker 2: And yeah, this is basically it on our side. Speaker 2: Thank you very much.
Speaker 2: We left here or website and now you can also buy from web. Speaker 2: From web, we have available Spain, Italy and Portugal, and so feel free also to contact us if you need something or you have any question we are going to be available for you and that's it, thank you. Bye.
Transcribed automatically; names checked against the speaker record. Select a time to jump there.
Article based on this talk
What changes when quality stops being QA's job alone?
Four Wallapop engineers show how shifting testing to the start of the software life cycle reshaped product, QA and developer work.
In short
Shift left moves quality and testing to the start of the software development life cycle, so the whole team, not only QA, owns quality. Laia López and her Wallapop colleagues Agustina Santos, Anna Ruiz and Rebecca Porras explain why their company made the move, how each role changed, and what it cost. Their headline result, as the speakers cited, is a 55% drop in bugs caused by unclear definitions compared with the previous year.
Key takeaways
Shift left mindset
Experimentation culture
The SDLC (software development life cycle)
Analytics
Software development
Why the traditional cycle runs late
Agustina Santos opens with the classic SDLC: strategy, design, development, testing (often called regression), deployment and maintenance. Its strength is clarity. Each step has defined roles, inputs and outputs, so everyone can see what is needed to reach a launch. The weakness is timing. Because testing arrives as a phase near the end, problems are found late and fixing them is slow and costly.
"But there are also some cons, such as not being a cost effective or time efficient process, where risks are not minimized through forward planning, which usually ends up being reactive to bugs." (3:48)
Testing as interventions, not a phase
Rebecca Porras then defines shift left: quality and testing practices happen earlier, in planning, design and build. The key conceptual change is that testing is no longer a single stage.
"It is important to mention that the traditional approach has a testing phase, whereas in the shift left testing are a series of interventions along the whole cycle." (4:31)
That change shifts accountability too. "So the whole team is accountable for quality." (5:00) The team describes the benefits as better communication, constant feedback, earlier risk mitigation, faster production and bugs caught sooner. Rebecca also positions shift right (monitoring user behavior, analyzing business metrics, running performance and security tests or deliberate failures in production) as a useful complement rather than a rival. She is candid about the downsides: role boundaries can blur, the cultural change is drastic, and test documentation must become more exhaustive because the whole team now uses it.
The problems Wallapop set out to fix
Laia López sets the context. Wallapop is a Spanish platform for buying and selling second-hand goods, organized into multidisciplinary topic teams that own part of the product and include engineers, analysts, marketing and other roles.
Anna Ruiz then lists the problems the company wanted to solve. Developers waited for QA before merging, creating bottlenecks. Feedback arrived late. QA engineers spent their time on manual regression, so automation coverage stayed low. And the belief that only testers owned quality meant more bugs surfaced in production or in pre-release regression, the most expensive places to fix them.
"Developers had a safety net, an insurance plan, because they knew that a tester would review the implemented product afterwards." (10:46)
Speed and learning mattered as much as quality. "Releasing faster means that we would be able to do more experiments." (12:04) Anna adds an honest disclaimer: not every topic team has adopted the model at the same pace. "Some teams have adapted really quickly, and while some others are having more trouble." (13:26)
What changed for product, QA and developers
On the product side, Agustina describes engineers joining ideation workshops from the start, experiment and tracking sessions where relevant, and requirements defined through a three amigos approach and reviewed by the whole team. The result, she says, is a sense of joint ownership and a clearer view of the metrics each feature affects.
For QA, Anna describes the biggest shift. Instead of validating each feature after development, QA engineers now put their effort into acceptance criteria and test plans before anyone codes, and most remaining manual testing is exploratory. The role moves toward strategic thinking, requirements definition and risk assessment.
"We have stopped seeing testing as a stage and we are now seeing it as a series of interventions that can and should be applied during the whole software development lifecycle." (15:46)
The team measures the effect by classifying bugs by root cause. "For example, we've seen a 55% reduction in definition bugs." (17:12)
For developers, Laia stresses that the job now reaches well beyond writing code. Developers sit in product meetings, raise technical limitations early, write unit and sometimes integration tests, and run their own manual regression before release. They ship many small deliveries rather than one large one, so user metrics arrive sooner, and they close the loop with technical documentation for their own and other topic teams.
"Okay, so I would like to start this part by commenting that for me, the most important thing about the shift left implementation is that the developers are not only limited to code." (17:39)
Benefits and the price of change
Laia sums up the benefits as connected: more attention to design, planning and investigation moves a team from reacting to bugs to predicting them, which means fewer bugs and faster delivery.
"A really important side note is that small deliveries and problems identified in advance are cheaper, but also faster to solve." (20:08)
The challenges are mostly human. Teams that are not cohesive, or that are used to a narrow role, can struggle at first, and unclear role definitions can confuse people. The speakers' advice is to make sure everyone knows clearly what they should and should not do, and to give the team time to adapt.
Watch the talk
Shift Left in the SDLC (Software Development Life Cycle), Laia López with Agustina Santos, Anna Ruiz and Rebecca Porras, 22 minutes: https://www.womentech.net/video/laia-lopez-shift-left-in-sdlc-software-development-life-cycle-apply-speak-87178
About the speaker
Laia López is a Frontend Engineer at Wallapop, based in Barcelona, Spain. She presented alongside her Wallapop colleagues Agustina Santos (Data Analyst), Anna Ruiz (QA Engineer) and Rebecca Porras (QA Engineer).
Questions this talk answers
What is the shift left approach in software development?
Why did Wallapop decide to adopt shift left?
How does the QA engineer's role change under shift left?
What results did Wallapop see from shift left?
What are the downsides of shift left?
Drafted with AI from the talk transcript. Editor review pending.
Frameworks and ideas mentioned
Shift left
Shift right
Software development life cycle (SDLC)
Three amigos
A/B testing
Cite this talk
Laia López. "Shift Left in the SDLC (Software Development Life Cycle)." WomenTech Network, 2023. https://www.womentech.net/video/laia-lopez-shift-left-in-sdlc-software-development-life-cycle-apply-speak-87178