Agility, Maturity

and Innovation
The Globant Development Experience

A Globant White Paper

By Andres Angelani,
Chief Solutions Officer, Globant

www.globant.com
Agility, Maturity and Innovation – The Globant Development Experience

A Globant White Paper

By

Guibert Englebienne, Co-Founder and Chief Technology Officer

Andres Angelani, Chief Solutions Officer

http://www.globant.com

I. Abstract

Over the last several years, a number of new technologies and related market
trends, including mobility, cloud computing and software as a service,
“gamification”, social media and “big data” have emerged, and are
revolutionizing the way end-users interface with information technology,
reshaping the business and competitive landscape for enterprises. As these
companies adjust their business models to adapt and benefit from these
changes, they are increasingly seeking solutions that not only meet the rigorous
engineering requirements of emerging technologies, but that also engage the
end-user in new and powerful ways. We believe this dynamic is creating an
attractive opportunity for technology service providers that have the
engineering rigor, creative talent, and culture of innovation to deliver these
solutions.

Globant has mastered a unique software product’s design and development

model (also knows as Agile Pod model) that combines agility and maturity to
drive innovation, while focusing on cost efficiency due to progressive and
strictly monitored gains on productivity and quality, namely Pod Maturity.

These teams (or pods) leverage the value of our eight Studios, dedicated to
maturing emerging technologies and market trends, and providing a constant
influx of mature talent and solutions that create intellectual property for our
clients.

We have also recorded net savings to our clients of over 30% in average, due to
sustained productivity boosts when pods operate at a higher maturity level.

1
These financial and qualitative gains were conceived mastering:

a) Productivity, quality and innovation incentives, which are tied to client-

specific “Pod Maturity” criteria.
b) Utilization of User Experience and related creative practices throughout the
design and development lifecycle
c) Pod autonomy and purpose, successfully combining talent and process
which resulted in record levels of people motivation and retention

This paper explains Globant’s organizational structure, team set-up and

processes to drive sustained improvements and an environment of innovation
for our clients, while we develop software products for them.

II. The Studios: where our talent thrives

We seek to deliver an optimal blend of design and engineering innovation to

harness the potential of emerging technologies to meet our clients’ business
needs. Since our inception in 2003, we have believed that while engineering is
central to information technology, only by combining strong engineering
capabilities with creativity and agility can we deliver innovative solutions that
enhance end-user experiences while meeting our clients’ business needs. Our
commitment to this differentiated approach is reflected in three core tenets of
our company: organization by technology-specialized Studios; emphasis on a
collaborative and open Culture; and innovation and creativity in technology and
design. To contribute to these core concepts, we have made and continue to
make significant ongoing investments in developing an operating environment
that fosters innovation, creativity and teamwork, while ensuring a commitment
to quality and project discipline.

Our Studios identify, create and mature emerging technology and trends, which
are formalized into Practices. The following are our current eight Studio and
their current practices:

Studio Practices
Big Data  BD Architecture
 BD Science
 BD Custom Visualizations

2
 Studio Practices
Cloud & Infrastructure  Enterprise Cloud
 Managed Services & Security
Consumer Experience  Portal Development
 Content Management Systems
 Ecommerce
 Web Scalability & Performance
Creative  User Experience
 User Interface Technology
 Visual Design
 Digital Marketing
Enterprise Consumerization  Collaboration Solutions
 Process Engineering Tools
 Talent Development Solutions
Gaming  Game Design
 Game Engineering
 Graphics Engineering

Mobile  Native
 Hybrid
Quality Engineering  Test Automation
 Load & Performance Testing
 Gaming QA

III. The Globant Innovation loop

Globant pursues a cross-company strategy to ensure innovation is constantly

flowing into our client projects.

As shown in the illustration below, innovation begins at Globant Labs, an

organization within Globant that specializes in research and development of
new technologies – those our research indicates will be relevant in the next six
to 48 months. We focus on what technologies and specific talent will be
required by the mainstream industries we work with in the mid-long term.
Logically, labs are influenced by market research and specific customer
innovation needs.

3
 At th
he Lab, we
w have an ongoing pipeline
p of 40-plus projects thhat range ffrom
custo
om robots,, gestural in
nterfaces, machine
m leearning, annd many otther innovaative
applications off emerging
g technolog gies.

Oncee these tecchnologies and relateed market trends attrract client demand, such
techn
nology is adopted
a by
y one of ou
ur Studios; the result of capacityy planningg and
Solu
ution building creates a formall practice w
with matuure professsionals thatt are
assig
gned to cusstomer proj
ojects.

Figure 1:: representin

ng the loop of innovation
n from the labbs, into the sstudios, cliennt agile podss and
back to th
he labs throuugh customeer feedback

Natuurally, the members of Studios participaate in cliennt projectss, bringingg the

expeerience annd solutionns that thheir practiices have created, making tthem
availlable to all
a our clieents. At the
t projectt level, thhis talent, expertise and
softw
ware fram meworks to o acceleraate producct developpment are leveragedd in
acco
ordance to our
o client’s specific delivery
d gooals.

The blend of practice

p ex
xpertise att the servicce of speccific client delivery ggoals
happ
pens at our client projects through
t aagile podss, small teeams withh the

4
adequate talent diversity to carry out at least full feature development, from
conceptualization to production (more on agile pods later).

IV. The Pod Maturity Model

One of the important differentiators in Globant’s development value proposition

is our pod maturity model.

As mentioned above, the Globant Studios host a number of related practices,

which are combined into agile pods to deliver software products.

The pod maturity model leads into what is the most distinctive and uniquely
value-adding aspect of the Globant development experience: an incentive-
driven methodology that aligns three goals critical to building partnerships:

a) the client’s goals, related to productivity, quality and innovation from the
agile pods engaged to build software products;
b) the Globant’s people (a.k.a. Globers) goals, related to their career
development;
c) Globant’s goals, related to maximizing efficiency, utilization, and growing
the engagement into a long term partnership.

Next, we will begin looking at agile pods, the pod maturity model and exactly
how it provides client-focused value that no other developer can match.

V. Agile Pods – an Introduction

Globant leverages its Studio talent to break down work on client accounts into
custom agile teams of no more than eight members, known as agile pods. Each
agile pod is responsible for managing a specific part of the feature backlog
related to the development of a software product or services platform. So, Agile
pods are cross-functional and combine talent from diverse Globant practices.

Any agile pod has varying levels of technical leadership, product

management/user experience, development, QA and creative talent. These
teams are configured ad-hoc, and are subject to client talent deficiency, level of
innovation, user experience, and technical complexity, envisioned for the
resulting product or system.
5
While the individual leadership ratio of each agile pod can vary, in general
Globant prefers to have one technical lead in charge of every pod. In addition, a
project manager will manage two to four pods while product managers and QA
managers optimally manage three pods each. In partnerships using more than
five pods, Globant would include technical directors who typically manage five
pods, while at a higher-level program managers oversee 10 pods.

The principle is that agile pods are fairly self-sufficient to develop software
product features or “themes” at a minimum level of supervision, thus
minimizing dependency and increasing velocity. In large software development
programs, agile pods can also have developers, creative and leadership from the
client team, as long as the Pod Maturity criteria are the same for all members of
the team. Next, we will see more of implementing these criteria.

Creating a Pod

As mentioned above, Globant combines talent from its various practices to

create agile pods. Let’s set an example to analyze the graph below. A client
envisions an innovative, multi-platform product leveraging a gamified digital
experience. Let’s picture a private investment banking platform that runs on PC,
tablet, smartphones, and provides market data, account, transaction
management. For the next phase of product evolution, the bank envisions to
strengthen the investment community around their product/brand, allowing
users to exchange market insight, participate in investment contests and
challenges. The awards may be reduced transactions fees and better positioning
on investor popularity leaderboards.

The bank could hire or utilize its own team of product, marketing, technology,
build a specification and then find a way to build, using their own resources, or
hiring a third party to help with the development. This is the traditional
approach: the partner that is supposed to build the software product is not taking
part of the initial activities that form the product vision, does not take a role in
design. In most cases, the resulting product differs from the vision significantly.

However, by bringing in agile pods early that blend talent in from specialized,
relevant practices, such as gaming experience, mobile experience and content

6
management, Globant
G crreates an entire
e ecosyystem thatt drives a hhigh degreee of
ovation to the resultin
inno ng product..

Figurre 2: the Stu

udio practices are approppriately combbined to buiild agile podds that deliveer the
gn and the ultimately, thee desired pro
desig oduct

The agile pod has all thee ingrediennts requiredd to be succcessful, now we neeed to
build
d an execuution modell that is aliigned to ouur client gooals, short and long tterm.
Let’ss see how.

VI. Pod Ma
aturity - How
H It Worrks

The Pod Matu urity criterria set meaasurable g oals, shortt and longg term. Thhis is
whatt we call “tthe maturitty path”.

Eachh agile pod

d is rated according
a to
t three foccus areas: velocity (hhow fast itt can
get work don ne), autonnomy/indep pendence (technicall mastery and creaative
ideass/team innovation) annd quality (user expeerience, design, reliabbility, etc)..

Thesse three criiteria are then

t used to
t assign a maturity level of 1--3 (with hiigher
num
mbers refleccting pods that perforrm higher-llevel workk).

7
 The chart below illustrates a simplified example of the criteria used to
determine how agile pods are scored in each focus area, leading to the overall
maturity level score:
Metric
Focus Area Criteria Description Level 1 Level 2 Level 3
Productivity (50% Velocity Number of Hours More than 60 Hours Between 35 and 60 Less than 35 Hours
weight) needed to Hours
complete a story
point

Autonomy & Technical Autonomy Level One or more Minimum of 40% At least 1 Pod
Innovation (20% Mastery the Pod reaches approved code customer members member is an
weight) (Core team thanks to reviewers per in the pod are Core Extended Architect,
growth), Technical development areas members; At least 40% Core
Creative Knowledge 3 process Members, 8
Ideas to improvement process
improve suggestions per improvements, and
process month 1 innovative
efficiency prototype per
and user month
experience

Quality (30% User Robustness of Green static build Green static build Product owner
weight) Experience Development ratio of 73% or ratio of 95% or review with no
& Design, practice, Maturity above. Percentage above. Percentage rework on
Technical of Design of successful of successful content/UX/Design
Quality practices, smoke tests 80% or smoke tests 90% or
Quality of above. Producer above. Product
Content and reviews 100% owner reviews
Technology success. Product approved up to 1
owner reviews rework UX/ design
approved: up to 2 iterations
UX/design rework
iterations

Table 1: a simplified example of Pod Maturity criteria

Globant and our clients collaboratively measure a pod’s maturity level according to
specific metrics every month, for example:

a) Productivity

The first step is to define and what is a story point. We prefer that rework be
considered within the standardization of what activities are required to complete
a point. The number of points attributed to a story or feature is not arbitrary, it
is agreed up front with our clients.

8
Within the first iterations of a project, while teams are ramping up, we observe
the volatility between estimate and actual effort, aiming to improve our
estimates and refine the story point definition. Globant conducts postmortems
and focuses on detecting delays and categorizing issues into intrinsic (fixable by
the pod itself) or extrinsic (fixable reconfiguring the agile pods structure and
potentially recommending improvements at the client operating environments).

Globant helps the client understand what is the baseline in hours per points for
Level 1, and negotiate the productivity boosts expectations for the other two
levels. Globant generally commits to higher or lower productivity boosts based
on the following:

- mutual agreement (Globant and client) on what the bottlenecks to velocity are
(e.g. underlying technical architecture dependencies, multiple vendor overhead,
client departmental restrictions, etc)
- client flexibility to help remove bottlenecks

b) Quality

There are multiple approaches to measuring quality. The traditional approach

would focus on number of defects and categorize them by type and severity.
Globant believes this metric is valid but insufficient, too far along the progress
of an iteration, or even worse, when the QA team has got hold of a build.

We prefer to add a faster quality metric to predict the quality of a build and
forecast the defects of a milestone. We need a continuous build system with
comprehensive automated tests. The agile pod always contains test automation
engineers, “embedded”, working alongside the developers.

When a group of developers is pushed to increase velocity over its normal

capacity, quality suffers. Test automation engineers add value to the stability of
the build. As this happens constantly when changes are made to front-end layers
or back-end components, the embedded test engineers repair the tests that
break, and create additional ones to maintain a high level of test coverage,
helping the developers identify and reduce technical debt.

The build system allows for high frequency of builds and test executions.
When a build is stable and has high test coverage, running many times per day,

9
it is easier to derive the level of code quality that the agile pod is producing.
Therefore one important metric is the percentage of green builds over all builds,
over time. Assuming that the level of test coverage is almost 100%, and
building every 5 minutes, if smoke tests succeed 80% of the time, we can
predict that the team is following good coding guidelines.

Another aspect of quality is user experience and design. This is assessed

alongside the client during bi-weekly demos. We rate feature using various UX
focus tests and other techniques, which are out of the scope of this document.
All these methodologies are within the scope of our User Experience practice.

The technical requirements (performance, availability, security) are of course

assessed by iteration and intensified as the build matures.

c) Pod Autonomy & Innovation Metrics

There are various ways to measure the capacity of Pod’s ability to self-regulate,
complete features and bring in innovative ideas. Globant recommends simple
metrics, generally derived from the following categories:

- Structural, e.g. percentage of core members within the pod (the core member
is defined as the developer, test engineer, or creative Glober experienced in
our clients business, technology stack, communication, specialized in a
certain technology or process, and committed to the program); members of
the pod who have been awarded architects for the program (this applies to
large programs where software architecture is the authority that brings
technical cohesion and sound design patterns to software)

- Behavioral, e.g. how many innovative concepts/prototypes has the agile pod
introduced to the program within a quarter, how many velocity and quality
improvements has the pod introduced, how many relevant solutions,
processes or frameworks has the pod leveraged to increase efficiency

10
Agile Pod Maturity Assessments

Every month and quarter, the Globant Program Manager and Technical
Directors overseeing the agile pods performs a self-assessment. Agile pods are
rated according to the Pod Maturity criteria, placed in a leaderboard, and
submitted to our client for review.

At the end of every quarter, our client audits and decides whether to promote or
demote pods’ maturities. Our client’s decision is based on both auditable data
points, and other considerations such as overall quality state of the entire
program.

The Maturity level of an agile pod only lasts one quarter i.e. if the pod does not
fall within the Level 2 or 3 criteria set, the pod will fall back to Level 1 or 2
respectively.

And while the hourly rates for a pod’s work go up as the maturity level rises,
client experience data shows that in fact, higher-level pods can actually produce
savings of over 20% or more compared to lower-level pods. For example, Level
2 pods perform at a minimum 30% higher velocity with twice the velocity
predictability of Level 1 pods. In a continuous build environment, Level 2 pods
generate at least 30% fewer defects and better architecture, design and higher
automated test coverage than Level 1 pods.

It is also worth noting that all agile pods are committed to a process of
continuous improvement, where they take in client and experiential feedback to
get to the next maturity level.

VII. Client Experience

Our client is a public, Fortune 500, muti-national company, and a leading

consumer brand. The program involves creating interactive digital experiences
for end-users. Globant participates in the concept, definition, implementation
and test of new features, as well as the re-engineering of a services backend.

We agreed to leverage the Pod Maturity model and ramped up a number of

agile pods. Globant used an average of 15-20 pods and reached a headcount
peak of 150. The duration of the program was longer than 24 months.

11
The following describes the experiences throughout the partnership:

a) Adoption

Globant walked into a waterfall-like environment , characterized by a sharp

separation of design, development and QA roles, heavy focus on producing
very extensive business requirement documents, and product ownership in a
different division and scarcely available.

Our client team was asked to innovate their current technology platform and
create an ecosystem of services to boost user adoption and brand fidelity in the
digital space.

During the inception phase of our engagement, Globant participated in creating

prototypes for the new product and contributed to selecting the relevant
technology framework. The first Globant agile pod built a core feature and
technology proof-of-concept.

b) Scale

As the prototypes were reviewed and approved by senior management, Globant

gained the partner role to build the front-end and some core back-end services
for the product.

Globant employed a simple method, which we call “mitoses,” to spin up new

pods. This concept has two steps:

- Step 1. Globant committed to growing 30% of pod member to “core” level.

Being a “core” level pod member relates to our Glober’s level of experience
on the client’s technology stack, and the ability required to develop the type
of features that the pod has been tasked with.
- Step 2. Globant would then spin up a new agile pod, using core members
from an existing team and adding new members to the project. Before
committing to delivering work, the new members were inducted in the
technology stack, processes and methodology by the core members. This
introductory phase is what we call a “Sprint Zero.”

12
While scaling up teams, this very simple concept created efficiencies, and team
bonding quickly as new members adapted to client’s business domain, and
inherent processes and tools. Learning curves and ramp-up issues were greatly
minimized.

c) Pod Configuration – Defining the key communication channels with Client

Teams

Globant ramped up agile pods taking the following considerations, which

resulted in best practices and roles inserted in the agile pods structure:

- Collaborate with our client’s product team: A Globant associate-level

product manager was inserted at the pod level. The Globant product
managers would generally service two to three pods and focus on “landing”
the feature stories, helping the product managers from the client to relect on
decisions, acknowledge pod capacity and load, write acceptance criteria,
socialize the concept for the product/feature to the rest of the pod, follow up
on progress daily, communicate issues and provide solutions to the customer
product team, and help the product team “groom the backlog” i.e. prioritize,
or trim scope as a function of pod capacity, feature value and desired time-
to-market.
- Take ownership of scrums: The agile project manager would take a couple of
pods and run daily scrums, unblock stoppers and coordinate tasks, flagging
dependencies; these scrum masters would participate in scrum of scrums to
resolve conflicts and fix delays, and also help product owners communicate
pod capacity to stakeholders to prevent over-commitments.
- Add creative roles: These roles would also “interface” with the creative team
from the client, as branding restrictions were very strict and closely
monitored by the client creative teams.
- Ensure high test coverage: Globant integrated test automation engineering
into the agile pods, whose main was to “enrich” the build process with an
extensive set of “smoke tests.”
- Enforce a cohesive architectural blueprint: The technical director, a role
overseeing agile pod’s technical leads, would manage approximately five
pods and focus on ensuring quality deliverables from a technology

13
 standpoint, architecture matching to client expectations and consistent across
pods, and component re-use.
- Add overall operational oversight: A program manager for the operational
management of the pod, logistics, facilities, and communication with higher
level stakeholders.

Since this was a multi-year, and large program, Globant also organized monthly
steering meetings with client executives to review:

- velocity (productivity)
- quality
- features delivered (value)
- risks and issues
- maturity level of each agile pod, velocity, quality and innovation
leaderboards

These meeting helped align high-level program goals at the higher levels of
both companies, and helped Globant make an investment.

d) Implementation of Pod Maturity

All new agile pods where rated at Level 1 (entry level). A Pod Maturity criteria
sheet was built similar to Table 1. Globant agreed to start the Pod Maturity
rating as of the third sprint for new pods, to allow for stabilization of velocity,
and team bonding. To support Pod Maturity, Globant implemented the
following steps:

- Pod identity: Each pod was given a name, and an avatar. Common values
were created for pods and an internal micro-site created for the program.
- Program branding: Leveraging our client’s powerful brand image, Globant
co-branded our facilities to exalt the partnership.
- Development environments: Globant and our client set up a site-to-site VPN
to work off the same code repositories and be part of the same build process,
minimizing delays and unnecessary coordination overhead .
- “Gamification” of the team environment: Globant added an element of play
to instill collaborating behaviors; here are some examples:

14
 o pod maturity leaderboards
o celebrations when agile pod members are promoted to core member
o celebrations when agile pods are promoted to higher maturity
o “shootings” when builds are broken, fun ways to put a spotlight over
members that break the continuous build
o velocity competitions with awards
o quality competitions with awards
o pod innovator of the month, quarter, etc.

e) Training Tracks

Globant implemented specific training tracks. The tracks were instrumented

and managed by the leads of the following practices:

- Product Management: Program scope, which consists of vision, feature

review, roadmap
- UI Engineering: html5, css3, responsive design
- Mobile Engineering: hybrid development (native, html5)
- Test Automation: specific tools and technology set for the program
- Client Services layer: APIs, protocols, etc.
- Core member track: i.e., how to become core pod player: expected
behaviors, our client’s business, leadership, peer review, client technology
stack, environments

f) Productivity Results

The following table shows the average productivity increases measured as

reduction of hours to complete a story point. Level 0 is for comparison, and
depicts velocity metrics of teams working in similar technology and product not
following the agile pod model. Level 0 alludes to staff augmentation contracts
where the customer sets all process rules constraining vendor independence and
creativity. The data was collected after two quarters, once Globant had enough
samples at each maturity level. The amount of effort, in this case, takes into
account total production time: time to define, develop, rework, and test within a
sprint.

15
The metrics average
a datta collecteed over thhe first sixx-month peeriod, in m
man-
hourrs:

Ave Producttivity (15
5 pods ssample)
100
90
80
Hours per Point

70
60
50
40
30
20
10
0
Leveel 0 Levvel 1 Leevel 2 Leevel 3
Hours per Point 0
90 72
7 50 33

We can calcu ulate the velocity

v im
mpact associated too rate inceentives inn the
owing tablees, comparred to projjects not foollowing thhe Globannt pod matuurity
follo
(Levvel 0):

Rate
Veloocity gainss Net cost # of Pods
Maaturity leveel Incenttive
(hou
urs perr impact att
ogression * (cost per
pro
poin
nt) * (savingss) * M
Maturity
hour) *
Lev
vel 1 10%% 18 12%
% 7
Lev
vel 2 25%
% 40 31%
% 6
Lev
vel 3 35%
% 57 51%
% 2
Pro
ogram Lev vel Cost-E
Efficiency 25%
% 15

ompared to
* co o Level 0 (not
( an agille pod)

As seen
s in thee table aboove, the raate incentiive within this engaagement m
model
drov
ve to higheer marginall gains in velocity,
v oonce pods reached hiigher matuurity.
At th
he program m level, thee 15 pods measured rendered a savings iimpact of 25%
whille performiing at high
her quality.

The agile podss were rateed at the en

nd of each quarter annd the rate incentive only
lasteed for the following
f quarter,
q wh
hich meanss that:
16
- only agile pods that had attained the maturity level during the previous
quarter deserve the rate incentive during the current quarter; and,
- agile pods that perform at a lower maturity level fall back to the
corresponding lower level and lose the incentive the following quarter

g) Product Quality, Innovation and Team Dynamics Results

We found that progressive pod maturity led to fewer defects at the end of the
sprint, while working at higher velocity. The postmortems provided us with the
following insight:

- Pods became more cohesive over sprints i.e. a strengthened sense of self-
regulation allowed for autonomy whereby they would either improve or
eject non-performing team members.
- The incentive to innovate on better processes increased velocity and
escalated blocking issues earlier, helping the technical directors and program
manager act upon dependencies and minimize bottlenecks that would lower
overall efficiency
- Pods gained a better understanding of their own capacity, how many stories
they could deliver per week, learned to better the estimates and became more
predictable; a big jump in estimate accuracy happened when transitioned
from Level 1 to Level 2
- Pod members embraced the fact that there is a career path within the
program, eliminating the need to jump onto a different project or leave
Globant; overall the program was low in attrition (lower than 10%) over 24
months.
- Pods submitted their ideas, prototypes and accepted the continuous
challenge of staying up in the charts. Those pods lower in the velocity or
quality leaderboards quickly started to rank higher on the innovation
leaderboard proposing new ideas and improvements to processes, tools, and
frameworks, and showing that they could turn their reality around; obviously
helped by their senior technical directors .
- Due to the level of feature autonomy and incentives to improve, we found
that Globers that would not normally step up to the plate and propose ideas
were now more open and communicative, adding to the overall spirit of
collaboration that we envisioned.
17
Overall, the agile pods created the inner team dynamics that fostered drive
for improvement. We used lots of “carrots”, team regulation strategies,
game dynamics, peer pressure, and no sticks.

VIII. Conclusion

Globant has created an entire cross-company ecosystem where expertise from

every part of the company is brought together into small teams called “agile
pods” that combine diverse skillsets to provide clients with efficient
development of the latest technologies. Our pod maturity model allows us to
use qualitative metrics to collaboratively agree with clients on the maturity level
and associated pricing their agile pod warrants.

Maturity levels are regularly reviewed to ensure that agile pods are constantly
focused on improvement and also that pricing levels fairly reflect the output of
the agile pod. In this way, Globant maximizes the value its dedicated practice
areas can deliver clients by eliminating technology “silos” and ensuring that
specific developmental skills, such as mobile platform development, are
augmented by other necessary skills such as visual design and delivered in a
fair, timely and measurable manner.

18