Realizing success

1. Realizing success

Instacart saw an initial opportunity. There’s a great story about one of the founders, Apoorva Mehta, doing deliveries, but he didn't have a car. He went to the grocery store, picked up the customer’s groceries, and took an Uber to complete the delivery. Mehta grew Instacart from literally nothing, a feature, but he saw the massive opportunity, so it was worth the work and difficult journey.

2. Align on strategy

The same is true with data and AI monetization. We see multiple opportunities, but we must align with strategy first. A startup’s core strategy is highly fluid. Mehta went through 20 ideas before he finally founded Instacart.

3. Opportunity statements

In a more established business, we must look at what business leaders currently believe are its most significant opportunities for growth. The monetization process starts by taking a strategic objective and turning it into an opportunity statement explaining what we will do. What part of the opportunity can the business seize? Why use data and AI for this opportunity? Opportunities are broken down into use cases that represent the value stream and value creation side of the equation. Use cases take advantage of the opportunity and help define the problem space. Defining the problem creates the connection to expected outcomes. Those can be cost savings or more granular internal KPIs. It can also be customer-facing, defining the customer experience that will be delivered and the value customers will get. Use cases connect to workflows, which helps us move from the value stream to what must be done to create or deliver value. The workflow defines two critical requirement types. Functional, which we're used to, and a new type of requirement, reliability. Models don't function consistently like other software does.

4. Reliability matters

We see this a lot with generative AI. Ask a chatbot a question, and it'll give you an answer. Sometimes it's wrong. That's the reliability component. Just because a model will function doesn't mean it meets the full scope of the user's requirements. We have functional and reliability requirements. Breaking down the workflow helps us understand how reliably a machine learning model must work to meet the business or customer need. Sometimes, this is new functionality, and in other cases, it’s an incremental improvement. Think back to Instacart shoppers, the people who fill the orders. They may be going into the store for the first time and don't know where all the items are. If the app can provide them with a map that's 80% accurate, it's an improvement over their guessing where items are located. Instacart shoppers are part customer and part employee or contractor. They use Instacart’s app and indirectly pay to use it. They also serve Instacart’s consumer segment. They are part of a marketplace, which is another power of a platform. The marketplace monetizes multiple components, not just the customer buying groceries.

5. Understand ROI

Finally, we have to estimate the ROI. To do that, we must understand the workflow as it is or at the current baseline. If an Instacart shopper enters a store for the first time, they don't know where anything is. We can measure how long it takes them to complete the order. The workflow changes if the app gives them a route that’s right about 80% of the time. Instacart shoppers go from guessing to following the route. What's the improvement? What's the value created by changing the workflow? How much time did it take when the person was guessing vs. having the app suggest a route? What is the expected value of that time saving? Because of the platform, we’re not just talking about the savings to the Instacart shopper but also to the customer who has a shorter wait for their order. On one side of the marketplace, we monetize the improvement in Instacart shopper productivity. In another part of the marketplace, we serve more customers faster. What is the ROI of those improvements?

6. Bringing it all together

Let’s recap because there’s a lot of content in a small package. We begin by meeting the business where it is and focusing on what business leaders believe is most important. We take strategic objectives and define opportunities with them. Opportunities break down into use cases that connect the initiative with workflows. This process includes defining the problem space, which covers the business outcomes the initiative will deliver and two requirement types (functional and reliability). Finally, we define the change to the workflow and estimate the value created by that change.

7. Let's practice!

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.