Link to podcast + transcript. 9 May 2023. All words by Fidji Simo unless specified otherwise.

Sajith: Enjoyed this a fair bit. Fidji is a super-operator who ran the Facebook app and now is CEO at Instacart. Very curious hire coming from a digital to a largely physical business, but certainly Instacart having a large and fast-growing ad busines helped. 

The podcast is interesting in that it gives you a sense of how Fidji and Instacart are thinking about driving the presently utiliitarian product usage into higher value add experiences leveraging AI, data and consumer insight, such as “I want a birthday party menu for 15 kids and 5 adults” and for Instacart to suggest that and enable one-click ordering. There are many such ideas and experiences highlighting innovative product thinking scattered throughout the podcast such as the Instacart Health product where they will enable doctors to prescribe a diet like they prescribe medicine, and enable that to be provided via Instacart. It is also a useful lens to understand how an innovative player in the ecommerce space is thinking about leveraging AI.


Instacart culture of celebrating cents / sweating the details

Fidji: Just to give you a funny anecdote, during my first week at Instacart, I remember joining our weekly business review, and there was a team that had managed to save $0.01 per delivery through a set of optimization. And as soon as they announced that, the whole room erupted in applause. And I remember thinking, oh, we celebrate cents here. 

And at Facebook, let’s just say, we weren’t focused specifically on the cents. And I actually think that, that culture of sweating the details, sweating literally every penny is what has made Instacart successful and has put us in a position where our unit economics are really strong in a market where there’s been a lot of attempts at doing what we do and attempts that have been unsuccessful. 

I actually think that constraints sometimes create just a better business and better operations. I find that people, when you set some hard constraints like you need to be profitable, you need to have great unit economics, it raises the bar on the type of talent you attract, the type of execution that you get in a way that differentiates the good companies from the great ones. 

Facebook’s playbook for building great products

Fidji: I think it was a lot about understanding exactly what consumers wanted and also understanding little demand. And what I mean by that is the playbook at Facebook was really understanding what users were already doing with a product that wasn’t necessarily built for that use case as a sign that there was something there that you could productize and do something bigger with. 

If you look at Facebook Marketplace, people were already using the core Facebook product to try to exchange goods. So we built Facebook Marketplace as a result of that insight. People were already posting links to YouTube to share videos, oh, actually, it might be better to create a real video experience. 

Learnings fm Facebook playbook for Instacart

…have taken that to heart at Instacart in really understanding what are the things that people are already trying to do. 

And at Instacart, it translates into they’re trying to figure out what their meals for the weeks are, as they’re trying to actually come up with new recipes to diversify their diet. And every time we see something like that, we’d just build a product experience around it to help users with the things that they’re fundamentally already trying to do. 

How AI could change ecommerce + going AI-native

I think the biggest change that we’re seeing with generative AI is the shift towards natural language. Because if you think about commerce fundamentally, it is a pretty weird experience online because you already need to know exactly which products you want, go into a search box, type that product, select that product, add it to a cart. 

It is not how people think about their lives. They’ve gotten used to it because that’s all we could do with the tools that we had. But fundamentally when people think about feeding their families, they think, okay, I have a budget of $200 for 10 meals. I have a family of five. One of them has food allergies. We all like Mexican food. What can I make this week with all of these constraints? That’s the way you think about managing your family. 

And then ideally, you would have an engine that tells you, okay, with all of these things, this is what we would recommend for breakfast, lunch and dinner. And in one tap, you can add all of these ingredients to your cart. That is a much more natural way of thinking about the problem of what’s for dinner, what are we going to put on the table than the current search experience. 

In order to really embrace the shift, you need to think about what it would look like to go AI-native.  And what I mean by that is having AI not bolted on, on the side to your current experience, but thinking about if you were solving the problem that your company is trying to solve, now from scratch with the tools that are currently available, how would you rebuild your products? 

Would you rebuild the same? Or would you rebuild it in a completely different way? Would you have conversational experience instead of having a search box? And that is really important because that echoes in earlier stage of my career where I was asked to figure out how to monetize mobile at Facebook during the shift to mobile. 

And there was a clear contrast between the companies that were embracing the shift to mobile by taking their web experience and desperately trying to shrink them on a small screen versus the companies that were really using all of the tools that a mobile device had from location awareness to ability to take pictures and all of that to create a really unique mobile-first experience. And I think we’re going to see the exact same thing with AI. I would encourage companies to think from first principles, with this new tool set, would you rebuild your products in a completely different way. 

Using AI at Instacart

we’ve already released a plug-in for ChatGPT, which allows you to go and basically start talking about anything you want to do. I’m organizing a birthday party, and I need snacks for 15 kids, what do you recommend? Boom, you get an answer, and you can order it in one tap on Instacart. That’s pretty magical. Same thing for entire meal plans, which I find amazing. My favorite one because we are collecting those from employees was my very judgmental in-laws are showing up in two hours. 

What can I make to impress them? Same thing. Apparently, ChatGPT recommended a Caprese salad, which means that ChatGPT really hasn’t met my own in-laws who will not be impressed by that. All of these use cases are actually very magical when you go from natural language to a set of ingredients to these ingredients showing up at your door. I think that’s incredible. On meal planning, specifically, the ability to say, this is my budget, these are my food restrictions. 

These are the number of meals I need to plan and have a set of suggested recipes and then being able to, again, in natural language, iterate on that and say, the thing you recommended for Tuesday, that’s not really my jam. Can you suggest something else that’s maybe a little less spicy and have something that will show up. That’s really, really magical. I think going forward, there are somethings that are really incredible where you can take a picture of your fridge, and you can get an immediate answer on well what’s left in your fridge. 

This is the recipe that we would suggest. Or hey, you took a picture of your fridge at the beginning of the week and now this is the difference, we can reorder all of these items. That would be pretty cool. You can imagine taking pictures of recipes on a recipe book, the recipe book we all have on our shelves and never open, you could go and take a picture of that. And now the ingredients show up because it is very cumbersome to open a recipe book, try to take a list of all the items, order them online, all of that could be completely automated. 

And then I think there’s a lot of applications in health, wellness and nutrition where you could imagine having nutritionists being able to make a lot of recommendations in the automated way. And right now, if you are someone who wants diet advice, going and seeing a nutritionist is a very luxury experience that’s not affordable for most people. 

If all of that knowledge is captured by an AI assistant, you can make nutritional advice completely democratized, which could be totally game-changing for the type of problems we have in the country where how it fundamentally starts in the kitchen. 

So I’m really excited about that as well and really reach the promise of having a diet that is completely medically tailored through your exact needs and all of these ingredients showing up at your house. I think we’re way further away from that because every time you touch health, it is obviously very scary to make any mistake. And I don’t think the technology is there yet for us to trust that. So we still need human intervention in the process. But over time, I’m hopeful that we will be able to democratize experiences like that. 

As software gets easy, focus on the difficult

AI is going to make software development easy. And that means that there’s going to be fewer barriers to entry. 

So when I advise companies, I always tell them, focus on doing the hard things. It was for 10 years we have had the luxury of creating a lot of value from fairly easy things in consumer software. And now I think it’s going to come down to really creating much deeper technology. And even for us at Instacart, the way I see it is that we are this unique translation layer between the digital world and the physical world. 

The hard thing that we’ve built is connecting to 80,000 stores across the U.S., connecting to 1,100 retailers and this ability to translate your intent of cooking something for dinner into products showing up at your doors within 2 hours because the physical world is still going to remain pretty difficult to navigate even in an AI- enabled world. 

I encourage companies to really think about what is the thing that — the data set that’s going to create most value or the experience in translating the digital into physical experience in the world that’s going to be really hard. And then you can combine these really hard things that you’ve done with all of the magic that’s coming out of AI, and that’s when you can create really amazing experiences. 

Sajith: see Paul Graham on Schlep Blindness.

Grocery ecommerce

Grocery is the largest category of commerce. We’re talking 1.1 trillion in North America alone. It’s a massive, massive retail category. And yet, it’s a category of commerce that is the least penetrated online. It was 3% penetrated online before the pandemic. And then post-pandemic, we are at 12% online penetration. (India 1% now)

If you look at fashion, if you look at beauty, electronics, they’re all in the 25% to 30% online penetration, which is kind of weird because I don’t think you have many people who are saying that groceries in store is so much more enjoyable than going and buying electronics or buying fashion.  (India overall ecom penetration is 7%)

First principles thinking around product

I always go back to what is the fundamental problem that we’re trying to solve (Facebook was remaining in touch with people). And in the case of Instacart, this question of what’s for dinner is a question that every family asks themselves every week pretty much since the dawn of time, and yet it is still a very complicated thing to do. 

So that’s why a lot of people consider grocery shopping a chore because it is a complicated set of questions that you have to answer before you put food on the table. So again, AI technologies can fundamentally change that and make it easier because you can align them with a fundamental human problem. 

Instacart works closely w offline grocers to support their omnichannel strategy

Publix, Sprouts, Wegman’s are all powered by Instacart.

Fidji: …a fairly sizable part of our GMV comes from transactions that happen on our grocers’ sites. We monetize it the same way as we monetize our core app, which is through a mix of fulfillment and advertising, but it is already very meaningful. 

I was talking earlier about this massive trend towards brands wanting to spend on retail media. And that’s a massive opportunity for grocers to turn their e-commerce website into advertising platforms. 

But that requires a massive investment in technology. We can provide that for them. In fact, we just announced that we are doing that with Sprouts where we power advertising on And that means that overnight, Sprouts can start getting some of that advertising revenue directly to them. And for us, it’s also beneficial because that means that we can take the customers that are very loyal to Sprouts, going straight to and also monetize that customer base. 

So all of these technologies are already part of many retailer sites. And we do that in a modular way. It’s sometimes we power the whole website plus a fulfillment plus the ads. Sometimes, we just do the fulfillment, connect it to a retailer’s website. So we have multiple models to really adapt to where the retailer is at in their own digital transformation and what they want to build themselves versus what they want to complement with our technologies. 

And that’s why we made acquisitions like Caper carts, which are smart carts that allow you to skip checkout because we think that grocers are going to need to really embed these technologies inside their stores because customers are going to expect a rich personalized experience, not just online but also inside the stores. That move to omnichannel is really what the next 10 years are all about in grocery. 

Getting to unit positive in grocery

Fidji: Scale is absolutely everything in this business. We used to lose money on every single order until we got to 100 million orders. That’s a lot of orders. It took a lot of investment to get to the point where we had delivered enough orders and we had enough densities that we could turn a profit. And the way it works in our business is basically having density of orders at the store level so that we can batch orders. 

And what I mean by that is that we send a shopper to the store and instead of picking just one order for one customer, they might be picking two orders and then delivering that to the end customer, which saves a lot of cost, as you can imagine. And that’s something that’s really unique to grocery because most of the time, you don’t need the items to arrive at a certain temperature. That’s much more the case in restaurant delivery. 

With grocery, you can actually batch orders and pick multiple orders at once without any issue. And so that’s a very big driver of our unit economics and something that can only be achieved when you have so many orders within a given store that you can combine that. The other thing is that scale gives you a lot of data. Well, back to the data topic, where with a lot of data, you can optimize much better. 

Sajith: Strategies that drive density of orders at store level such as greater focus on micromarkets to drive volume could help achieve unit economics faster.

Sajith: They have 600k shoppers (gigworkers who pick, pack and deliver). Curious why they dont split the two roles. Have specialists who pick, and delivery folks.

Instacart as an ad platform

If you’re a category leader in, let’s say, cream cheese or gain share if you’re a new entrant. The way we work with them is by offering them a set of marketing solutions to be able to attract that customer. One of them is sponsored products, which is if you type in our search engine cream cheese, you can appear at the top in sponsored units. That’s very typical advertising, but also more storytelling capabilities. 

We’ve recently launched what we call shoppable display and shoppable video ads where brands can have full video messages about the attributes of their products and then within one tap, encourage people to buy them. 

And so we know that our advertising solutions work. And in fact, during these interesting market conditions, what we’ve seen is that brands have tended to cut down on unmeasurable top-of-the-funnel advertising and put more dollars behind very high-performance targeted advertising and with exactly that, you can literally advertise right as a person is building their cart and get your product in the hands of that person within two hours. 

And that’s why you’re seeing a massive rise in what’s called retail media networks and the type of dollars going towards that. We’re really helping brands with that transformation. And one thing I’m really excited about is that it also gives a shot to totally emerging brands because if you are a really small emerging brand, you have just developed this one amazing product, good luck figuring out how to spend your money on TV or radio. 

All of these channels that are just incredibly costly, whereas with us, if you carried at a couple of grocery stores, you could put money behind that and make sure that customers that shop online for that grocery store, discover your products and then you can move your products off the shelf at retail by doing that. And so that’s the Holy Grail for these emerging brands who are trying to get these expansions at retail stores and are trying to their spend dollars in very measurable channels. 

Instacart native brands

It’s all about tapping into a category of products that consumers have demand for. I think it’s a little bit of stream of just designing for Instacart. It’s more designing for the online consumer, which we have a proxy for. And there’s this company — full disclaimer I’m an investor, but it’s called Starday Foods. And I love that company because basically, what they’ve done is really thought about how do you build emerging food brands with all of the new tools that online allows you to do? 

And what would it take to build a CPG product like we build software products? And what that means is that they A/B test the name of the product. They A/B test all of the colors of the packaging. Instead of having focused group of five people, they put that through Instacart ads instead. And through Instacart ads, they try to figure out, oh, okay, when the product is this way, actually, people react better, when it’s that way people react worse, and they optimize our products the way you would optimize a feature online. 

Encouraging team members to highlight and celebrate their superpower

I organize these dinners with my team, where I push them to talk about their superpower. No one will volunteer, yes, amazing, blah, blah, blah. 

So I usually just push them a little bit. This is what I see you do really spectacularly well, how do you do it? Try to articulate it for someone for whom that’s not a natural scale and try to give them tools so that they can become good at that. 

Her passion project around chronic care 

I have partnered with two co-founders to create a medical and research center around complex chronic conditions. 

About 80% of health care costs come from complex chronic conditions that affect multiple body systems. And yet, the health care system is still organized by specialties that look at you as if you were just one organ. You go to a cardiologist and they look at your heart. 

You go to the neurologist, and they focus on your brain, but a lot of complex chronic conditions are actually at the intersection of immunology, cardiology, rheumatology, and that’s why patients are bouncing around from specialist to specialist with no one looking at the full picture. And that’s one of the things that we are trying to solve with my medical and research center, but it’s, I would say, a micro example of the bigger problem in health care, which is so many different silos and myopic views that don’t serve the whole patients very well. 

Instacart Health

Another thing that we’ve launched recently is called Instacart Health. 

And what it is, is taking our entire infrastructure of, again, access to 80,000 stores, 1,000-plus retailers and making it available to the health care industry so that they can scale food as medicine. And what I mean by that is making it really easy for doctors and providers to prescribe food as easily as they prescribe medication and have that delivered to your door the same way you would have a prescription delivered to you because it is pretty crazy when you think about it, that about 87% of health care costs come from diet- related diseases. 

And yet to this day, it is much easier for a doctor to prescribe medications and to prescribe a diet. For us, what we are thinking about is, hey, we built that entire technology and platform infrastructure so that anyone in the U.S. — we reached more than 85% of the U.S., can get fresh food in under two hours. 

How can we put that in the hands of a hospital system or doctors so that they can just prescribe food as easily as they prescribe medication? And how can we enlist payers, health insurance companies to reimburse the cost of fresh food because you would save them so much in healthcare costs down the line? 

Leveraging content to drive sales

People spend way more time watching content about food than preparing and eating food. That tells you everything you need to know about the role of content. But the way I think about it is that content is fundamentally inspirational. And I think when you develop a business that is based on inspiration, not just utility, you have a fundamentally more valuable business. And this is something that we’re trying to crack with Instacart where Instacart was a fairly utilitarian business. 

…we’re trying to move to a model that is much more based on inspiration because when you ask people, what do you not like about the things you eat every week. Very often people tell you, well, we eat the same thing every week. We are just in a rut. We don’t want to think about it. We just eat the same meals over and over again. Content is a way to make it easy for people to get out of that, inspire them to change their diet. 

We launched Lists, which is influencers, celebrity, nutritionist being able to create this list of products. So if you’re on a low-sodium diet, these are the products that you should buy. All of these things are a way to inspire people, nudge them in a direction of the experience being a little bit more exciting and nudging them toward products that they would not have thought to buy otherwise. 

And so that’s the role that content plays. And I think the real question with AI is which content gets fully commoditized versus which content ends up being truly unique. And I don’t think we’ve fully seen that play out yet. But if you think about recipe content, you can now, with AI, create entirely new recipes that do not exist in the world. 

You can just create recipes that are like very specific to your taste and your family, et cetera, and AI will generate that for you. I think that there’s going to be a really interesting transition in what we think of in terms of content experiences and how this of blogger content needing to be produced is spending a lot of time doing that. Some of that in some categories of content is going to go away and allow for an explosion of these much more inspiring experiences.