Data-Driven

Experiential and Incremental Implementation

Veröffentliche eine Notiz:

Motivation

We have started this blog-series with the question why it is so difficult to become data driven and explored the approaches to accomplish this in Part 3. In this article we go in more detail and focus on experiential and incremental delivery.

The main goal of experiential and incremental approaches is to gain a faster time to market in data driven engagements. The secret lies in creating something that meets the minimum requirements to showcase its business value, allowing for further development and growth into a more comprehensive solution. We refer to the solution as comprehensive rather than final, because delivering in increments means that products or data platforms evolve over time and are constantly improved.

An initial, yet essential milestone is called Minimum Viable Product (MVP), it will play a central role as you will see below. Additionally, the concept of an experiment is key, as incremental approaches allow trying things out that seem to have value based on initial discussions. If it fails in the end – probably due to an unfeasible concept – the point is to fail fast and learn, not to spend weeks or months running into a dead end. However failing is not the goal, so we want to be sure we start in the right direction from the beginning. Enterprise Design Thinking is key here, this helps to ensure strong user focus. The needs and hopes of all relevant stakeholders especially the end user are at the core.

The first sections explains the methodology that has proven to be successful at IBM. After that, we introduce our dedicated Client Engineering approach with some client references to prove the value of the overall approach.

Methodology

To build successful data driven applications or products, the requirements of all users and relevant stakeholders need to be the top priority when it comes to planning a new data driven engagement. This may sound trivial but was often underestimated in the past where technological aspects were prioritized over the actual consumers.

Enterprise Design Thinking addresses this with outstanding results in many engagements. From the beginning (requirements gathering), the solution is developed in a multi-functional team. This way users stay engaged, business stakeholders ensure their requirements are met, while technical staff can focus on feasibility. Enterprise Design Thinking offers value not only in the design of new products but also in providing a framework and goals for conducting workshops, particularly when involving individuals with diverse roles and skillsets.

Design thinking workshops are usually structured along a path from generic to specific concepts

  • Business Framing: Help define valuable Use Cases
  • Use Case Discovery: Sharpen identified use cases
  • Scoping: Prioritize business value against feasibility

MVP = Minimum Viable Product

A MVP is the version of a product that has just enough features to be viable, which means it is valuable for the targeted users. Hence the MVP should focus on basic functionality and does not necessarily need to be deployable to a production environment. Nevertheless it needs to be reusable or extendable so that subsequent iterations can build on it, thus delivering improvements incrementally. This distinguishes an MVP from a PoC and implies to follow architectural guardrails right from the beginning.

The desired outcome and value is defined by the corresponding business stakeholders. In addition, MVPs can also encompass purely technical solutions, involving stakeholders such as developers or operations staff. In general any product that provides value to humans in their day-to-day business is a good candidate for incremental delivery and MVP approaches e.g., infrastructure automation.

Enterprise Design Thinking

IBM’s Enterprise Design Thinking (EDT) is a framework based on proven methodologies from the Lean Startup context that have been adapted to the specific requirements of the enterprise landscape.

Innovations cannot be created in a vacuum. At EDT’s core is the observation that developing or integrating technology for the sake of it, seldom matches expectations. Solutions need to consider the context, the needs and pain points of the people eventually using them. Ignoring the end user needs often causes significant additional costs, for example adapting the solution to requirements that should have been taken into consideration from the very beginning.

We ensure the necessary focus by

1.) leveraging a toolbox supporting every possible step in a customer’s decision-making journey (from general interest to specific requirements and use cases). This toolbox also provides technological patterns (Automation, Data & AI, AIOps etc.), where we identify the sweet spot between postitive impact on the user, value for the business, and technical feasibility. 2.) co-creating and closely aligning with our clients during the (MVP-) Build Phases. Business stakeholders and end users are thereby empowered to give feedback in time to adjust between iterations. At the same time the required knowledge transfer between to the client’s technical experts happens as the solution evolves, allowing the client to quickly extend the innovative MVP after our engagement is completed.

Example: Data Discovery Workshop

Before the MVP kickoff it is important that we define a statement that formulates goals and success criteria – the MVP Statement. A good MVP statement ensures all stakeholders align to the value that shall be created, sketches the motivation behind and defines measurable goals to proof success.

Agile Approaches: Incremental Delivery

The following content relies on practices proven in engagements of the authors. It does not claim to stand as a scientifically representative survey and some aspects can be considered opinionated but though is underpinned with experience from many successful MVPs.

Team structure

  • Diverse, multidisciplinary teams work best: Embrace diversity, establish a positive and open atmosphere – in this way you will be able to mine the best ideas and align against a common goal.

Team Operations

These are lessons learnt from the Build-Phase of an MVP. In a nutshell: Have a structure, stay flexible, adapt if necessary.

  • Start every iteration with a planning session where the backlog is scanned and the corresponding tasks are assigned to the team members.
  • Have a short daily standup, so all members are up to date.
  • At the end of the week perform a playback to get all stakeholders up to date. Playbacks are to demonstrate what has been build. They are aim especially the key stakeholders – i.e. representatives of the personas that are mentioned in the MVP Statement. Those can confirm the project is on track and decide on priorities (reflected in the backlog – this is crucial for the next planning session).
  • Retrospectives target the MVP (-build) team to discuss success achieved so far and areas for improvement. They are an opportunity for the team and each individual to give and get feedback to learn and grow. Plan it at the end of a week, not necessarily every week, but after the first iteration and at the end of the engagement are good checkmarks.

Tooling

  • Principle: Use lean tooling that every participant can access. Artifacts should be governed in established repositories (e.g. Git), all documents should be in an open format so they do not require propriety software or licenses.
  • Kanban boards are an intuitive to use project management utility that gives enough structure and a simple birds eye view of the whole engagement. Trello offers an intuitive Kanban interface which can be used out of the box and further provides options for specific customizations.
  • For diagrams (e.g. architecture overview, component diagrams etc.) draw.io is widely used through the industry. It offers templates and icons for many technical environments and solution components.
  • Establish a common (text-) chat / voice / videocall platform where all stakeholders have access. Slack or Teams are a natural fit for ad-hoc communication.
  • EDT relies heavily on the use of whiteboards. If your team operates face to face, it should have lots of Post-Its and markers at hand. Tools like Mural or Miro do a great job providing an intuitive and easy to use environment for virtual whiteboards. Also some cloud based office solutions are also capable to do e.g. virtual brainstorming sessions.
  • For complex projects consider using a more sophisticated tooling that is aligned with the methodology, e.g., at IBM we use our Cognitive Architect, which has support for architectural deliverables like Architectural-Decisions, Requirements etc., a draw.io integration for diagrams and supports linkage of specific assets to the solution components.

Example: IBM Client Engineering & Value Engineering Method

Short introduction Client Engineering

IBM Client Engineering (CE) is a division within IBM that delivers valuable and scalable business outcomes across all industries by providing technical expertise and support to clients. The deeply skilled multi-disciplinary teams aim to help clients address their unique challenges by leveraging IBM’s technology and resources. By using a human-centered approach, IBM CE assists clients in designing, developing, and deploying solutions using IBM’s software, hardware, and services.

Short introduction Value Engineering Method

The Value Engineering Method is a human and design-led approach to solve complex business problems with transformative technology. This approach uses a proven co-creation methodology that aligns experts, utilizes experience based on thousands of projects, with key client stakeholders to quickly identify business or technology challenges and build Proof of Experiences (PoXs) that solve and ‘de-risk’ those problems using IBM hybrid cloud and AI technology. The following figure illustrates the 4 key phases of the Value Engineering Method:

INNOVATE

The initial INNOVATE stage helps to align on a solution to a compelling business opportunity and define a scope to prove value. This includes a range of workshop formats like Framing, Discovering, Solutioning and/or Scoping.

Key Deliverables of Innovation Phase: MVP Statement including Goals & Success Criteria

PREPARE

The second step is all about getting ready to start co-creation. That means, to ensure the internal and external team is ready to start, to setup the environments and to commit as a team to building the MVP rapidly.

CO-CREATE

Based on agile methods and principles the team builds, validates and improves the solution over a series of sprints and playbacks.

TRANSITION

For this stage, the team needs to ensure the client is on a path to scaled production. During this phase, a roadmap for adoption, sophisticated knowledge transfer and a validated business case should be created together with the customer.

Enough theory? The next chapter introduces two specific use cases, we have been built with customers based on the Value Engineering method.

Use Cases and References

There is no better way to prove the effectiveness of methodologies and approaches than with success stories. The following two stories provide a glimpse into what IBM Client Engineering has co-created with clients, boosting their innovation agenda and significantly improving their process efficiency.

OLGA – Trial assistant for case processing in diesel exhaust proceedings

One such project was delivered by IBM Client Engineering together with the higher regional court in Stuttgart. The IBM team co-created and deployed a mass trial assistant for case processing in diesel exhaust proceedings. The AI-supported system is used for document processing and is intended to relieve the burden on all parties involved.
The court in Stuttgart is burdened with the task of processing hundreds and thousands of appeal cases related to the emissions scandal “Dieselgate”. The individual handling of these cases results in a significant reading workload for judges, who must go through lengthy legal documents in electronic files. This process ties up their capacities for years, leading to dissatisfaction and frustration among court personnel due to the continuous influx of appeals in similar cases.
To address this issue, IBM’s mass litigation assistant, powered by IBM’s Watson AI, was introduced at the suggestion of the Ministry of Justice in Baden-Württemberg to handle Dieselgate lawsuits. Tasks such as parameter extraction and automatic population of predefined categories are well-suited for IBM’s AI-powered search platform, Watson Discovery. It retrieves data from unstructured documents and reduces the time spent on manual searches by over 75 percent.
Within just five weeks, a prototype was developed to ensure a solid foundation of data, minimal susceptibility to errors, and high user acceptance. Additionally, reliability, transparency, and traceability are crucial requirements. Data privacy, availability, and scalability are also taken into account. Based on the successful MVP, the solution is now incrementally extended with the customer and deployed into production.

AI assisted patient discharge – TIMETOACT Group builds Data Platform for Agaplesion AG based on IBM technology

Hospital providers in Germany face medical and financial consequences if a patient is discharged too early. If a patient returns for unplanned follow-up care within 30 days, health insurers will not cover any subsequent cost. Discharge decisions are often made within minutes not considering all health data.

Based on highly confidential patient data and specifically trained machine learning models IBM Client Engineering, TimeToAct and Agaplesion AG have built a solution to predict whether a patient is likely to need to return to hospital or not. A machine learning model, developed in Watson Studio and deployed on Watson Machine Learning, is actively analyzing pertinent health data and undergoes continuous monitoring for both quality and drift using Watson Open Scale. The solution is helping medical staff to make a transparent decision about whether and when a patient can be discharged, reducing the risk of relapses requiring returns within 30 days.

 The Key to Success for the experiential and incremental implementation have been the following:

  • Innovate: identify opportunities to support medical staff
  • Co-Create: build a technical MVP that analyzes relevant medical data and predicts probabilities for returning patient or not
  • Scaling: grow relevant data sources and integrate AI into the discharge process to support the medical staff

MVP Solution Overview

More insights and details about the solution are available here:
https://www.timetoact.de/details/business_intelligence_data_warehouse_system_agaplesion

Next Step: Become Data-Driven

If you want to get deeper insights please check our landing page IBM Client Engineering. If you already have an idea to build something with us, please contact the authors – we will be happy to jointly explore the opportunities.

Senior Client Engineering Solution Architect, IBM Technology, DACH

Florian Scheil

Business Technology Leader, IBM Client Engineering DACH

Anna Helle

Advisory Client Engineering Designer, IBM Client Engineering DACH

More stories
By Sascha Slomka and others on Oktober 24, 2023

AI Governance

AI governance has received a lot more attention as AI regulations are being formulated and passed. But AI Governance is not only about regulation, it is the key discipline to master the complexity induced by the variety of AI frameworks, models and tools. AI Governance relies on proper Data Governance which has been discussed in […]

Weiterlesen

By Andreas Weininger and others on September 12, 2023

IBM’s Data Platform for Data-Driven Enterprises

What technology does IBM have to offer to help you become or strengthen your position as a data driven enterprise? IBM recognizes that most enterprises don’t start on a greenfield, but instead already have a landscape of data stores and analytical systems grown over many years. Therefore, IBM’s approach to a modern data platform focuses […]

Weiterlesen

By Sascha Slomka and others on Juli 18, 2023

Experiential and Incremental Implementation

Motivation We have started this blog-series with the question why it is so difficult to become data driven and explored the approaches to accomplish this in Part 3. In this article we go in more detail and focus on experiential and incremental delivery. The main goal of experiential and incremental approaches is to gain a […]

Weiterlesen