DEEPER DIVES: Instacart's Expanding Into B2B - SMART

Good Morning,

First, as always, thank you for joining.

I’m continuing on a few of the themes I’ve been writing about recently, mainly increasing revenue and profitability in the face of incoming increases. It sucks seeing smaller business teetering on the brink because of an impact to their margins they had no control over.

Hopefully this helps.

Here’s what this issue brings:

• Will Call helps restaurants when they run out of what they need for tonight’s service. With gigwork networks looking for ways to improve profitability and relevance, moving into the B2B world is a solid step forward for Instacart
  

• The world is too noisy for out of the can reports and generic analysis. I’m sharing examples of how and why you need to customize your analytics if you are serious about increasing your sales
  

• Assistant first shopping in coming in hard and fast. For features like Amazon’s new “Buy For Me” to be a hit, will be need more regulation and price transparency?

This Is One Of The Few Instacart Service’s That Is Actually Needed

When I worked in the food industry, I was responsible for last mile (DSD) deliveries for some of the largest retailers in the world (Walmart, Costco, etc).

While DSD or consolidate warehouse deliveries varies by product and vendor, there’s one area that’s dominated by them in the food world.

Restaurants.

Every restaurant that you see is getting their product based on delivery schedules and planned routes.

Depending on the network structure, how those routes are constructed and the frequency that a store will get when it comes to its deliveries varies.

High volume locations can get daily delivery, 7x a week, where another location of the same banner might only get deliveries 2x a week.

And if there’s one thing that’s true when it comes to food, NOBODY ever wants to be out of stock when a customer wants to order something.

No matter what that network structure is, they all have one thing in common.

Special or emergency deliveries.

What most people don’t know however, is that emergency deliveries are a huge area of tension between the retailer/store and the distributor/operations team.

Why?

Because the product that needs to be rushed over usually doesn’t remotely cover the cost of getting it to the location.

Even when the product is “worth a lot”, you have to keep in mind that margins in the food industry are FINE.

And the cost of sending a truck stays pretty similar based on sending 10 cases, 1 pallet or even 5 pallets.

Because of this, requests for emergency orders are often pushed to the next day (when the order can be added to another truck already planned out on the road) or simply pushed and added to the location’s next planned delivery.

In both situations, it ends up with the restaurant/store being out of stock for some period of time.

Instacart has been watching and seen this need grow.

And they are offering up their network of gigwork shoppers to help solve the problem.

They announced last week a new service called “Will Call”.

It’s a white-label application that allows distributors to streamline same-day ordering through their sales reps.

Here’s how it works:

1. A business customer (think restaurant owner) discovers they need an item urgently

2. The customer calls or texts their distributor sales representative as they normally would

3. The sales rep uses the Will Call Delivery application to request an Instacart shopper

4. An Instacart shopper picks up the items from the warehouse and delivers them directly to the end customer

What’s even more interesting is that when items are out of stock or customers are far from the warehouse, distributors can also place orders from retailers on the Instacart Marketplace to ensure customers get what they need.

I have to admit, well done.

This does a number of things that I think create a solid win-win in the market.

• Restaurants can now have access to a more affordable service option

• By accessing a gigwork, capacity is dramatically increased (this is usually a primary factor distributors face when having to execute these services)

• Availability of a local contingency option (i.e. getting ingredients from a supermarket)

• A service Instacart can charge an appropriate rate for (which hopefully means better pay for Shoppers)

I’ve shared before in my articles that one of the biggest challenges for gigwork last mile networks is the disproportionate pay to time ratio.

In most cases, drivers are not able to make enough transactions per hour to make enough money based on the value of what is being delivered (there is only so much people will tip on meals from QSRs).

By moving into the B2B space, Instacart opens up a whole new service revenue stream AND gets to do it with customers that will actually be in a better position to pay for what that hot shot delivery actually costs.

(If you’re a restaurant owner who needs those cases of product that have been selling really well, you want to make sure you have the ingredients you need to sell all of those plates).

The other advantage is that emergency orders can be manipulated to increase the delivery value.

This was a regular thing that my team would manage when doing emergency deliveries in the past - especially when it was to places like coffee shops for example. We would increase the products and quantities on the emergency request (with the customer’s approval of course) and then cancel their next planned delivery.

The emergency order was still more expensive than the planned delivery, but we at least removed some of the cost by not having to go back to the location 2 days later with a nurfed order.

Instacart is piloting this program with GFS (Gordon Food Services) at launch, but they plan to expand the service to additional distributors through the year.

I expect this to be a wild success and that Instacart won’t have any issues onboarding (almost) every distributor out there doing restaurant deliveries.

Here’s Why You Need To Stop Leaning On A Generic Tech Stack

I think Shopify has been amazing for commerce.

It has blown the doors open so that anyone can get into the entrepreneurial life.

They’ve built an ecosystem that allows you to have a polished store, decent tools and an engine that can handle the load.

But when you build for everyone, you inherently trade off customization and nuance. And while there are a lot of common needs of a business, going from good to great means tailoring your solutions and workflows to YOUR business…

(because who cares how well it’s doing for someone else).

In a world that just got turned upside down, and everyone feeling the burn of rising costs, let’s talk about ways to use the data YOU ALREADY HAVE to help you make more money as a D2C or small retail brand.

The Meaty Middle

I hope everyone out there is familiar with the Bell curve.

What’s funny about a normal distribution when it comes to eCommerce is that it usually doesn’t accurately represent your sales patterns (outside of a high level view of totals over a long period of time) but most of the efforts brands take happens here.

When in reality, it needs to be focused here:

And that’s because you will often see patterns link this in your data:

(Both of these charts are over simplified, but it’s to help make the numbers easier to understand)

These two charts give you different things to focus on when it comes to improving your overall profitability and your ability to generate more revenue from your raving fans.

Use the insights you get from your order value distribution to optimize things like:

• Shipping speed / service

• On-hand inventory

• Promo codes and application

• Shipping container (e.g. polybags, boxes, etc)

• Returns policy

The data from the re-order distribution helps you get more value from:

• Loyalty programs

• Identifying your best customers

• Channel performance

• Campaign performance

*Note: Your re-order data is often far more interesting when it’s wrapped in different conditions or criteria

Another valuable type of data is understanding how your sales/customers are distributed across a particular geography and what strategic moves you want to make to then grow your sales.

The above map is of the Los Angeles area.

Here you see three specific (and custom) geographic boundaries that are being used to track and improve D2C sales.

These are listed by priority:

1. Black

2. Navy

3. Teal

You can use these territories to create specific data analysis or leverage them in a weighted priority matrix.

Ranking Matrix

Too often brands treat all sales (and customers) the same.

No one wants to treat customers differently, they are all important, but the reality is, certain customers do more for your brand than others.

There’s nothing wrong with acknowledging and owning this.

That doesn’t mean you give everyone else poor service or offer a lower quality product, it means that enhance the experience of your best customers (the people that can’t wait to tell their friends about you - driving even more sales).

Creating a custom ranking matrix is all about WEIGHTS.

These are the different priorities you will place on the various elements that you include in the grid.

The goal is to include something that’s comprehensive, but still is straight forward, easy to change and DOESN’T try to get overly clever.

Here are some examples of fields that I recommend:

• Customer type (New, Active, Loyal, VIP, Multi-Year Low Value, Multi-Year High Value)

• Number of Orders During Period (this could be 30, 90 or 180 days)

• Average Order Value

• Return Rate ($)

• Target Geography (Yes, No)

• Number of Orders Target Categories

• Number of Referrals

• Engagement (Email, Social, Text, etc)

From here, each element (column) would have a specific weight that you would multiply the actual values in. This creates a new table that calculates a score for your customer based on the weights of the data in your system.

Custom > Canned

I hope you can see why taking more ownership and customization of your data analytics drives better results for your business.

While generic off the shelf apps or plugins help get you started, they aren’t going to be able to get to those deeper levels of value that are highly relevant for your specific business.

And while this might seem like a lot of work, the best part about his is that it’s a process and a system.

That means that you will only define and design it once. You will find over time that you’ll adjust weights or make changes to what’s included, but your core framework is there.

The best part about doing business in the 2020’s is that there are a TON of options to automate and streamline workflows.

When I do this type of work for clients, I build all of the ETL steps into a structured workflow or script.

This allows for my clients to simply download key source files (if your system doesn’t allow for direct access) for the period they want to analyze, hit run and (almost) instantly get all of the formatted and transformed results.

If you aren’t doing these things already, I highly recommend you invest the time in building out these frameworks. Your bottom line will thank you.

“Hey Alex, Buy [THIS] For Me”

Amazon is introducing another new AI shopping feature to its platform.

"Buy for Me" functionality allows users within the Amazon app to purchase products directly from third-party brand websites if they aren't available on Amazon.

Available in beta for select U.S. users on iOS and Android, "Buy for Me" uses agentic AI (built on Amazon Nova and Anthropic’s Claude models) to source and buy products from external sites.

It integrates these listings under a "Shop brand sites directly" banner, handles checkout with encrypted user data, and tracks orders via a dedicated app tab.

Amazon is doing everything they can to keep users within its ecosystem (even when purchasing elsewhere), effectively becoming a universal shopping starting point. They understand how important it is to own the attention of consumers when it comes to commerce.

But the tool isn’t without its drawbacks.

And it raises some interesting ethical concerns.

Users agree to potentially pay up to $10 more than the initial estimated price by Amazon (the final price is set by the third-party brand). In addition, Amazon's return and refund policies do not cover these third-party purchases - you are subject to whatever policy the external brand provides. Lastly, promo codes can’t get applied and you are limited to single-item purchases (at least during the test period, we’ll see how that evolves).

For features like “Buy For Me” to become a mainstream solution, issues of price transparency and consumer protection jump up the priority list.

The UK’s Digital Markets, Competition and Consumer Act 2024, effective as of April 2025, offers a useful lens here.

It bans sneaky "dripped" fees (costing UK consumers £2.2bn annually) and mandates upfront disclosure of all fees in advertised price.

The UK law also cracks down on fake reviews (a £217bn driver of online retail trust, with 1 in 10 reviews on eCommerce platforms suspected to be fake).

Amazon’s own lawsuits against fake review brokers show it’s aware of this issue, but "Buy for Me" adds complexity because purchases occur on external sites where Amazon’s review vetting doesn’t apply.

While direct application of rules like the UK's dripped pricing ban might differ, the underlying principles of fair pricing, clear accountability, and combating deceptive practices are essential for building the consumer confidence required as agentic AI takes over more of our daily transactions.

That’s it for this week. Thanks for being here.