An Amazon PM on New Product Thinking in the Age of Machine Learning

With retail and marketplace experience, Himankini walks us through the importance of collecting customer data, reviews and experiences before beginning to work on a product in the current environment.

She explains how design thinking can be applied to building ML models, and how to make sure that your approach to PM is user-first and not tech-first. (She’ll also tell you why that’s so important!)

Meet Himankini Shah

Himankini Shah

Himankini Shah is an experienced Amazon Senior Technical Product Manager. She has previously been a part of Amazon Retail, Marketplace and Sharepoint.

She has strong analytical, data-driven and result-oriented skills acquired from managing solutions to deliver complex and multi-functional projects. As a Product Manager at Amazon, Himankini defined the product vision, roadmap, and support for machine learning solutions within Worldwide Operations. She also launched automated brand protection solutions to ensure a brand’s authoritative catalog is accurately represented on Amazon.com.

New Product Thinking in the Age of Machine Learning

What is New Product (Design) Thinking?

New Product Thinking is heavily influence by Design Thinking.

Tim Brown Design thinking

New Product Thinking is a tangible tool for working back from your customers, allowing you to develop and innovate the products that meet customer needs faster and with fewer iterations.

It works by breaking the stages of developing a new idea into an iterative process that is customer-focused.

Challenges with technology-first solutions (vs user-first)

Taking a tech-first approach presents product people with a few problems, because you’re not serving customer needs directly.

Solution without problem: Working backward starting with technology and solutions first vs focusing on the user and problem you are facing first.

Tech-first wrongly informs UX: Many algorithms assume which data levers to pull therefore limiting the customer interaction before defining the user goals.

Algorithms need to learn: Machine learning is learning user-based algorithms. In this, your first foot doesn’t guarantee to be your best one.

glitch computer GIF by Rylsee

How Does a Design Thinking Framework Work?

Designers, engineers, data scientists can work together to bridge the gap between MLor AI to solve customer problems.

New Product Thinking essentially divides product development into the 5 stages of the design thinking framework. These stages are:

  1. Empathize – Research your user’s needs
  2. Define – Define the problem you want to solve
  3. Ideate – Create your ideas
  4. Prototype – Build something to test your ideas
  5. Test – Give your prototype to users and start collecting feedback

Phase 1: Customer empathy

Building customer empathy is more than just getting to know them. It’s about gathering information on their pain points in order to know how best to serve them.

This stage of development involves developing a deep understanding of the customer’s needs, as well as recognizing their hidden needs. You need to understand the causes of their behavior, not just track their behavior. This should lead you to definining new metrics, and using customer research methods to obtain more insights.

When it comes to customer research, you need to be finding a balance of quantitive and qualitative data. You cannot use one and not the other.

Qualitative data helps you to discover your customer’s motivation, desires, pain points, and opinions. Common methods include focus groups, surveys, and interviews.

Quantitative data helps you to understand the behaviors and actions your customers take using more statistical methods to product measurable data.

You might also be interested in: How Qualitative Research Can Make You a Better Product Manager

Phase 2: Defining the problem statement

Defining your problem statement involves understanding and modularizing the current state and the outcome you are looking into.

To do this, you need to find meaningful stories/themes from customer research to identify problems and potential solutions.

Himankini also recommends building Insight Statements; a description of the current customer experience, why it’s frustrating for customers, and the desired end state.

Create a customer journey map

A very standard tool in building customer understanding, used across many industries, is the customer journey map. Using these, you can break down the customer journey into Why, What, How, and CX.

Blackflagship black flag ship compass GIF

Why:

  • Gain a clear picture of where the customer has come from and what they’re trying to achieve.
  • Understand what questions customers have, and how they are feeling.
  • Identify gaps, pain points, and opportunities to enhance their experience

What:

  • Visualize the customer experience
  • A customer journey map is a compact visualization that tells the story of today’s customer experience from beginning to end.

How:

  • The art and science of mapping a story.
  • Begin by identifying the key stages a customer passes through in their interaction with your product

CX:

  • Add customer feelings, motivations and questions they face at each stage of the process.
  • Includes customer goals, touch-points, emotional response, customer thoughts, overall customer experience, and ideas to improve.
working sci fi GIF by Leon Nikoo

Phase 3: Ideation 

Once you’ve understood your customers, and mapped out the problem that they have, it’s time to think of solutions.

Albert einstein quote

So in this case, spend the maximum amount of time understanding the customers and their problems than working out the possible solutions.

Once you have your insight statements, you can think about how to transition them into ‘might we’ statements. This turns your challenges into opportunities. Turn your insight statements around and frame them as ‘how might we..?’

This questions is broad enough for creative freedom, but narrow enough to make it manageable.

Thinking styles:

Once you have your How Might We statements, you might rely on one of the following three thinking styles:

  • Divergent: Exploring new ideas through brainstorming
  • Emergent: Leverage ideas from divergent state to explore new ideas
  • Convergent: Sifting through ideas and deciding upon winners and losers

How to brainstorm on solutions: 

Himankini gives us her four top tips for effective brainstorming:

  1. Tap onto a broad body of knowledge and creativity to product ideas in a group
  2. Promote openness and generate a lot of ideas 
  3. Don’t be solution-oriented or worry about immediate feasibility
  4. Look out for problems solved before perhaps in other industries or technologies

Prototype

design technology GIF by Morena Daniela

An initial prototype should be the lowest fidelity that the customer can respond to. It’s a powerful tool to get feedback from customers, by giving them a minimum viable product to play with and try out.

Though iteration, you’ll evolve the details and resolution of your prototype over time to come to a high fidelity prototype.

Final Advice from Himankini

Embrace Failure

If you aren’t failing you may not be thinking big enough or moving fast enough.

Gather Feedback

  • Evaluate the feedback, evolve your prototype and solicit your feedback again
  • Get inputs from customers and key stakeholders
  • Invite honesty and openness
  • Adapt on the fly

Did you like this talk? You can find many more just like it on our YouTube channel!

Future of pm banner

Enjoyed the article? You may like this too: