Buzzwords like Big Data, Cloud, and AI have been out there now for a couple of years. But today, businesses have a clear focus on the application of data use cases and the challenges around that such as metadata management, governance, security, and maintainability in general. Everybody seems to have some version of a data lake and wants to consolidate it into something (more) useful, or move from an on-premise version to the cloud. There is a general need to streamline current practices while also attempting to give multiple segments of users (data scientists, analysts, marketeers, business people, and HR) access in a way that is tailored to their needs and skills. In other words: businesses today are heavily invested in data and AI, but many have a hard time knowing how to mature it to the next level.
This is exactly where a "maturity model" comes into play. The goal of a maturity model is to help businesses in understanding their current and target competencies. This helps organisations in defining a roadmap for improving their competency. A maturity model is therefore one way of structuring progression, whether the company already embraces data science as a core competency, or, if it is just getting started.
In this presentation on maturity models, we answer the following questions:
1. What exactly is a maturity model and why would you need it? We address this by sharing GoDataDriven's maturity model and describing the different phases we have identified based on our experience in the field.
2. How can you use a maturity model to advance your organisation? Having a maturity model alone is not enough, in order for it to be valuable you need to act upon it. This paper provides concrete examples on how to do act based on practical stories and experiences from our clients and ourselves.
4. 4
Making organizations AI-Driven
we organize, we build, we train
AI STRATEGY EXECUTION
TRAINING &
COACHING
BUSINESS
EXPERTISE
TECHNOLGY
EXPERTISE
5. Applied AI
5
Delivers value to end-users
Through products and services
• Pro-active
• Personalized
• Engaging
• Intelligent
AI solutions can drive both
business and consumer
applications!
Recommendation
SmartsearchAutocomplete
Churn drivers
Generalizing
6. 6
Data Technology People
?
Something feasible, usable, and valuable
+ +
=
+
Question: what are the key ingredients
to delivering valuable AI solutions?
8. You are not alone:
8
"Our international Data Survey 2019 (1350 participants)
showed that data is an essential part of the strategy
for 79% of the organizations.
At the same time, it lists taking models into production
as the biggest new data science trend, above hot
topics such as deep learning (#2) and IoT (#3).
Although companies see data as an essential part, they
have trouble turning their data and insights into
valuableAI solutions."
For more info: http://www.gdd.li/datasurvey2019
That sounds
familiar...
9. 9
Data Technology People
Something feasible, usable, and valuable
+ +
=
+
Question: what are the key ingredients
to delivering valuable AI solutions?
Organization
Embedding
11. Data, people and technology are only three gears
11
The extent to which your organization embeds AI is crucial for success
1. Analytical capability
• Data
• People & skills
• Tools & technology
2. Organization embedding
13. You can measure AI maturity along these two axes
13
For valuable AI you need to organize both technically and organizationally
1. Analytical capability
• Data
• People & skills
• Tools & technology
2. Organization embedding
14. How to mature your Analytical Capability
14
Inmature Mature
People &
skills
Mostly relient on external consultants
No vision & strategy for growing in-house talent
Knowledge sharing is minimal or ad hoc
Senior talent in-house
Clear careerpaths and training curriculum in place
Constant knowledge sharing; learning is part of culture
Tools&
technology
Dependenton IT for sandbox environment
Self-maintained clusters with lacking maintenance
Monolith legacy systems that can hardly be changed
Sandbox and productionenvironment in own control
Usage of easily scalable cloud technology
Technologyis build modularly and is looselycoupled
Data Working with one-off dumps,replicationof data sets
No data ownership, nor data governance program
Organization not aware of importance of good data
Integrated data, easily accessiblein centralized platform
Chief Data Office inplace for data governance and quality
Importance of good data demonstratedbyAI practice
Examples of maturity indicators according to our clients
15. You can measure AI maturity along two axes
15
For valuable AI you need to organize both technically and organizationally
1. Analytical capability
• Data
• People & skills
• Tools & technology
2. Organization embedding
• Sponsorship
• Funding
• Implementation
16. How to mature your AI Organization Embedding
16
Examples of maturity indicators according to our clients
Inmature Mature
Sponsorship Bottom-up initiatives; low board level involvement
AI use cases are pushed from the AI capability
Little communication aboutAI developments
C-level leads AI movement & provides a purpose
Business pull on AI capability for new use cases
AI community that activates the organization
Funding AI initiatives funded by IT
No innovation budgetfor experimentation
No value measurement
Business pays forAI solutions
Central budgetavailable for research& innovation
Value based prioritization & benefittracking
Implementation AI products developedin isolation without business
Hard to roll out products in organization
AI thinking “comeson top of” regular work
Designprocessin place that guarantees value impact
Business in charge of roll out and productadoption
Organization thinks AI: use cases originate anywhere
17. How do both come together?
17
Organization Embedding
AnalyticalCapability
18. A typical AI maturity journey
18
Continuous
Experimentation
Enterprise
Empowerment
Initialization
Organization Embedding
AnalyticalCapability
Initialization
• Find and initiate first use cases:identify
opportunities,bootup data, people & tools
ContinuousExperimentation
• Expand team, infrastructure and company
presence.Define standard ways of working
EnterpriseEmpowerment
• Grow AI practice across all business units and
put business in driver’s seat; buy-in required!
AI Driven Company
• AI literacy in genes of company; anyone has
skills required to make AI driven decisions
AI Driven
Company
19. Each maturity phase has a different focus
19
Initialization
AI Driven
Company
• Find and initiate first use
cases:identify
opportunities,bootup
data, people & tools
• Proofing value
• “Can we do it?”, “How
hard is it?”, “Where are
the opportunities?”
• Expand team, infra-
structure and company
presence.Define
standard ways of working
• Capability building
• “How to organize?”,
“What skills and tech is
needed?”,“How do I
repeat and scale?”
• Grow AI practice across
all business units and put
business in driver’s seat;
buy-in required!
• Organization embedding
• “How to involve the
business?”,“Who is
paying for new and
matured use cases?”
• AI literacy in genes of
company; anyone has
skills required to make AI
driven decisions
• AI democratization
• “How to supportAI driven
decisionmaking across
the organization?”, “How
to measure everything?”
Focus
• 1 - 5 • 5 - 30 • 30+ • Entire company
Questions
State
# People
Enterprise
Empowerment
Continuous
Experimentation
20. Be successful by focusing on the right area
20
AnalyticalCapability
• People & Skills 4 5 3 4
• Tools & Technology 2 5 3 4
• Data 1 3 4 5
Organization Embedding
• Sponsorship 3 4 5 5
• Funding 1 2 5 4
• Implementation 1 3 5 5
This step is hard
organizationally
Initialization
AI Driven
Company
Enterprise
Empowerment
Continuous
Experimentation
22. The Analytics Translator bridges the gap
between business and technology
22
Analytics
Sponsor
Analytics
Translator
Data
Scientist
Data
Engineer
23. The analytics translator is your
compass to value
Tasksto be done:
• Collect,refine and prioritize ideas
• Develop the use case and business case
• Drive the solution developmentprocess
• Involve key stakeholders from the start
• Proof the value of the developed solution
• Drive adoptionand facilitate the change
• Fulfill AI ambassadorrole within organization
Actions
Insights
Data
Measure
Optimize
Predict
Value
26. 26
• Align with the business
roadmap
• Solve the right business
problem
• Make a solid business
case
• Share domain knowledge
• Involve end user during
experiment
• Validate generated
business value
• Integrate solution in
workflows
• Drive adoption by end-
users
• Evangelize the success
stories
Business involvement is crucial
in every phase of the development process
31. 31
The Analytics Translator understands both
business and AI and takes ownership by nature
“The analytics translator combines:
1) deep domain expertise,
2) an understanding of the different machine learning models and how they can
realistically be applied and
3) digital product management skills (entrepreneurship, product design,
experimentation, backlog prioritization, and stakeholder management)."
Doron Reuter
AI Products & Partnerships Wholesale Banking Advanced Analytics, ING
35. 35
The Analytics Translator: The Must-Have
Role for AI-Driven Organizations
Companies struggle to create impact and value with Data & AI.
The Analytics Translator helps you:
• Focus on building the right things
• Get business involvement from the start
• Embed AI in your organization
It’s a crucial role for success!
https://godatadriven.com/whitepaper-analytics-translator
36. Data Driven Maturity Scan
36
Methodology
The Goal of the Data Driven Quick Scan is to obtain insights in
the maturity of an organization's analytical capability.
The results are intended to help the organization improve its
current way of working, and increase its analytical maturity.
The scan consists of a systematic review of the existing data
operation, focusing on four attention areas. Each area is divided
into data & AI relevant themes, wherein each theme is scored on
a scale from 1 (common practice) to 5 (desired practice). Scores
are based on interviews, architecture and code reviews. The
average of scores per area eventually constitutes to an objective
final verdict per attention area.
Data
Tools &
Technology
Process&
Organization
People
& Skills