1. The Hard Truth
We started with the smallest part of the framework, going from business strategy to initiatives and from the top level, big opportunity, to the workflow changes and value creation. Hard truth time.
2. Must deliver value
Business leaders won’t wait for you. A key course insight: You must serve two needs at every business. The first is business leaders want AI. The second is we must deliver value quarterly. If we can't deliver quarterly or more often, interest will fade. Buy-in will dissipate.
We must be consistently delivering. We break big products and platforms down into incremental features. Features are critical, but how do we keep all those incremental features following a cohesive direction? How do we follow a single vision so features turn into a coherent product that lives on a platform like Instacart?
3. Start Simple
We do it with a maturity model that breaks initiatives into the simplest components and starts with simple implementations. Don't go straight to AI, even though that’s what business leaders expect. Why? We can deliver a ton of value with expert systems or basic digital software.
Early initiatives give us an in to gather data and deliver it to customers and users with that software solution. Instacart’s route recommendation uses a very complex model, but it could start with something much simpler. A digital implementation could be lists based on what we know from the grocery stores’ aisle descriptions. Instead of a route, version one displays data to the Instacart shopper.
Starting simple before implementing advanced models for route recommendations enables us to deliver value faster. We can continue to deliver incrementally by progressing from the digital solution to data to basic descriptive models. Each one is inexpensive and easier to deliver than the big AI feature.
4. Scaling simplicity
At some point in the maturity model, costs scale faster than returns. More advanced machine learning methods cost increasingly more, and we must stop delivering when costs scale faster than returns. We should continuously evaluate initiatives as approaches get more complex and expensive.
How much value can we deliver with cheap technology versus going straight to AI? Sometimes, we never get to AI, but the business is happy because the data team has delivered value incrementally. Deciding not to leverage AI becomes a cost-saving decision based on ROI.
Begin with features. Features support a small part of a workflow and should deliver immediate returns. The reason we need a road map is features set up for more complex implementations. Every time we do a little bit of work, we get access to new parts of the workflow or new data.
We can gather new data, deliver a model, and evaluate changes in the workflow. Those changes enable another data set, and so on. Features aggregate and eventually become products.
5. Staying ahead of the competition
Products support a complete workflow and deliver higher value. Unfortunately, products are not sustainable revenue generators. Eventually, competition enters the marketplace because they see the same revenue opportunities. They deliver a substitute and, eventually, it's hard to differentiate between the new and old products.
Businesses expect to compete on AI, but often, customers don't know they're using AI. The technology isn’t a differentiator, and customers won’t pay more for something with the AI label.
Businesses grow by finding new ways to deliver value. Products eventually become commoditized. When features leverage more advanced approaches, costs begin to spike. As the product becomes more mature, incrementally improving it to maintain the best-in-class standing is also more expensive, putting downward pressure on margins.
6. Commoditization
Commoditization and the cost of incrementally improving the product mean we need a new opportunity. That's the power of platforms like Instacart. Instacart has turned the platform into a product.
The company has incremental ways to monetize, like new products. They have larger ways to monetize the marketplace. Marketplaces' power on platforms is to create an ecosystem of different customers and partners. The business has the opportunity to monetize each one of them. They can also monetize anyone who wants access to the platform or customers, brands, and grocery stores on the platform. Any company that can monetize access to the platform is a potential target for monetization.
Platforms hold multiple products and support longer chains or multiple workflows. They enable ecosystems so businesses can monetize partnerships. The platform generates data natively, lowering the cost of data gathering and reducing the cost of delivering those AI features to customers.
7. Build incrementally and strategically
Follow the maturity model, deliver incrementally, and use the cheapest technology possible.
When you're building a road map, go from features to products to the platform. Features support small parts of the workflow. Products support complete workflows. Platforms support multiple workflows or longer chain workflows.
8. Let's practice!