DoorDash: An EveryDay Tale of giving and taking

Analyzing DoorDash's product strategy and proposing solutions for their AI trust gap and search limitations.

I am going to start with what I love about the platform. Segregation upfront. No time wasting, you want Indian here is the button you click for that. In the mood for Chinese, say no more, here are 26 options within your area. Who does not love an old-fashioned Pizza or Pasta. Obviously Alcohol, because that is where the margins are. Popular choices, your preferred choices based on your past choices and what might you like. I know if I look at my order history 80-90% of my orders will be from the above-mentioned choices DoorDash has done commendably well in presenting to its users upfront. Catch the essence and intent early and you have yourself a thriving business. No time for mood swings or I don't know what I want (hell I even saw a restaurant name a dish after that catchphrase!). To top that off they have integrated AI into the platform, unknown dishes from an unknown ethnic cuisine, say no more we will have our fine-tuned AI model working behind the scenes to cook something up for you. DoorDash has formulated an unparalleled recipe of food outlets partnerships, supply chain and user support request handling (the traditional outsourcing to India + AI supervision on user comments). You could talk about GrubHub, Uber Eats etc. all day, you could still not contest the 66% market share DoorDash has secured in the food delivery industry.

Let's dive into the supply chain first. I have a 40% discount voucher lying around from Uber Eats for ages and they refresh it for me every now and then so it's not outdated. As much as I would love to use that coupon, the stumbling block always has been the final order value for the same dish on Uber Eats vs DoorDash. Uber Eats has CA fees, Driver fees and all these random charges, but what I care about most is the final dollar value that's coming out of my wallet. When you tell me that amount after a 40% haircut is more than the amount I am paying on DoorDash, it's a big fat no. I do have a DoorDash Subscription to be clear. I often wonder if the Uber Eats analytics team look at their metrics and see a rise in hovering and a lot of carts being started but never being converted. If yes, you could point them to me because I think I know the exact reason why. This is not the topic of this article though mind you.

You could pick any other food ordering platform and you could argue they could or they are offering less additional charges compared to DoorDash now. Maybe better promotions. Is it enough for me to leave DoorDash, hell, no. You want to know why? here it is: Let's look at a parallel example to understand this, consider Uber vs Lyft. Now these are my views only which is implied, but for the sake of being explicit, I have to say it. If I am running late and need to order a cab, my muscle memory goes to Uber. I found myself considering why. The answer's simple and is in how I started this paragraph, supply chain. Uber has a better supply of Drivers than Lyft. You could argue this to their incentives or what else, the customer does not care about what inside the black box but what is being offered to them. Returning back to the point, the same goes for DoorDash. It has the best network of food providers who earn the best commissions. Would I want my order 45 mins late than paying $2 extra, I have found out the hard way that I wouldn't.

Let's talk about the app itself now. My order quantity was really low the other day on my Chicken 65, I wasn't thrilled. I let them know. Before I could bounce out of the app, it asked me to provide more description. I was intrigued. It had never asked me that before, Even if I would enter "Quantity Insufficient", it would be accepted as a genuine reason. I gave it a detailed write-up of what was up and there you have it, a 60% refund on that specific item. It has me guessing: has DoorDash really integrated LLMs so well that they could take a user complaint parse it, and gauge the appropriate compensation. Mind-boggling and I am curious to explore more.

Sticking to the AI narrative, I mentioned them generating description of unknown dishes for a user, the idea is not bad in itself. What it's missing is user trust and more importantly what happens when it's broken. Let's say a user has passive knowledge of a dish or is intrigued by something their colleague, classmate or friend mentioned and are excited to explore it further. They are already in the platform ready to order it but not all the way there. They need more information. Makes sense, they are spending their money for this. Now they could do a google search you would argue, but getting that information while being inside the DoorDash ecosystem is much less arduous. They like what they read, it synchronizes with everything that they have heard about the dish. They order it. Time for the hard truths, it's nowhere near good or what they have tried before. That resonating feel is gone both with the dish and the platform. It's not all airtight I guess.

Pics go hard. I certainly think so and they could say a thousand words needed at the same time. With these cool AI-generated videos floating around, what's stopping DoorDash from bringing those onto the platform. You could argue about all the cool tech in the world but first and foremost it should be relevant. AI-generated text descriptions, photos and videos are relevant to the DoorDash platform and does accrue a lot of monetizable value but it comes at an operating cost of the modern world. Trust. There is a reason why big tech companies have a profound trust and safety department these days in which they have poured millions and if not billions yet. How do you earn that user trust that your content won't be misleading?

Well, RLHF (Reinforcement learning with human feedback) comes to mind. LLMs are bottlenecked by the best model available in the market. Suppose DoorDash has that best model, how do you know it's still giving you what you are asking for. I want you to think about customization and personalization here, these are big terms in today's market. This is how big corporations make the big bucks. It's about satisfaction for every user who comes to the platform, not some cohort or localized domain of people. This is a pivotal problem. In Product Management terms, we call this creating a moat. No single platform can achieve this on its own but what they can do is leverage its vast network of partnerships.

The DoorDash platform cannot allocate resources to verify each description or photo being generated but the delivering restaurant can. It is their bread and butter after all. They could verify in seconds what their services are. The caveat that might come up is how is this possible to verify every text, photo or video generated for their offerings. You don't have to. What happens when you go to buy a car, you talk to a salesman from the particular shop you are interested in buying the car from who walks you through the car features. Same logic would apply here, uncertain about a dish, send a "more information requested" prompt to the restaurant, let them answer your concerns about the dish or anything pertaining to the restaurant asynchronously. Obviously, there would be a limit to the number of requests you could send in a day. Then you use that information exchange to update your descriptions your photos, videos and even fine-tune your models again. Certainly could have a FAQ. The user is also more assured about making an informed decision. It is a win-win.

Videos are not talked about enough. What if DoorDash has chefs showcasing how the dish was made videos or live streaming integrated into the platform. Sounds promising! Not possible, no worries. It could certainly do animations and when you have video generation platforms like Veo 3 and Midjourney coming out, it's never too late to think out of the box.

Even if DoorDash nails this all down which it has huge propensity to in my vision. The key blocker that remains on the platform is keyword search. Before I comment on the quality of search, I want to talk about the presentation of results. Currently, it shows the restaurants that matched your item search, if the restaurant has the exact same thing you want, it shows "based on your search" in the item description. It also does not always depict the item upfront when you tap into that particular restaurant. Plus you get a list of similar dishes at other restaurants horizontally. The list is buried within the presented restaurant options. That list horizontally to me should be the main dish in itself. Why not simplify things and show you found a specific item match at a particular restaurant instead of making the user labor through each restaurant option. If you think it's a spelling error, recommend so in the zero state queries instead of sending the user on a wild goose chase and wasting their time.

This is still plug and play. The real meaty part of the equation here lies in semantic search. Imagine a pop-up assistant in chat where I tell it what I am in the mood for or what ingredients I am looking for and it recommends choices available within the platform at that time based on my provided preferences. This here is a winner to me. You have drastically reduced the effort on the user part, you have considered their need, you already have their food order history, now you just have to make the right recommendation. To have the right recommendation system is a talk in itself for another day.

We have already discussed what DoorDash is doing well or what it can do better in certain regards and handling new technology. Now let's talk about what it is not doing well and does not want to talk about as well. The Service fees + Membership fees + Tipping business.

I too pay DoorDash a monthly/annual membership fee. That's not enough, they want me to pay a service fee associated with each order value, the minimum is set on that. If that is not enough to drain your wallet, here comes the tip for the dasher. Trust me as a DoorDash user for over 3 years, I could vouch to the fact that you are not getting your food untampered or at all, if you don't include a tip. I have heard some contacts complain of the orders being delayed significantly if the orders do not have a tip included.

Imagine Uber or Lyft taking 100% fare from the trip and the drivers had to resort to tips for gains. The whole notion is bizarre. By the time you reach the option to tip the dasher on the DoorDash platform, the financial motivation is gone. It just does not make sense to pay both the service fee and tip, I mean what is the operating cost to maintain the platform? Isn't that why you are making us pay the membership fees in the first place or is that a guarantee to an elite society not accessible to the common men? Take a % commission (like Uber/Lyft) and give the rest to the dashers. Leverage public good and CSR responsibilities if you have to.

Another point of discussion is that publicly available information on the platform poses a threat to user privacy: reviews and tip amount (at least most of it) being accessible to dashers before the order is delivered. This creates bias. My own personal experience where the dasher did not follow delivery instructions and I was hasty to put in their review. They came back and tampered with the order before I could get to it. This only lessens user satisfaction.

The feel I have been getting out of the DoorDash platform is that of an insurance company trying to get out of paying their claims. I get it, food ordering is a risky business, maybe the margins are less (well restaurants puff prices on the DoorDash app) but I suspect DoorDash has internalized the maximum compensation per user per month that they are going to hand off no matter how many issues you encounter with your orders. I can see this as them trying to protect themselves but I wonder if it's sustainable in the long term and will they lose customers? Something like insurance coverage comes to mind, user can opt to insure their food or not just like their flight tickets. The pricing levels would certainly vary.

Well, the needs of the many overshadow the needs of the few and I think that's where DoorDash is exploiting the market. Our lives are certainly better with it but are our wallets?