We routinely field calls from investors asking how AI is going to change the retail landscape.
For starters, retailers can’t spell AI.
It’s just who they are demographically.
We’ve never met someone from Harvard who couldn’t wait to jump into retail.
You have to be crazy – or, most likely, devoid of any other options – to work in retail.
Long hours.
Unreasonably angry customers.
Peers who can’t teach you anything useful.
Not nearly has easy as rolling into a bulge bracket bank, then B school, then PE where you coast on nothing more than a spreadsheet and a golf course membership.
The reasons really boil down to two key truths.
First, retailers do NOT self-discover.
They require very expensive sales efforts.
Toast might have the best product in the history of POS, but they still spend 36% of their operating expenses on sales and marketing.
And this is actually pretty damn good as SaaS companies go, but product led growth companies are closer to the 15-20% range.
Why?
Well, because customers sell themselves.
Atlassian spends < 20% of their operating expenses on sales and marketing.
That’s because Atlassian sells tools to engineers, who sure as shit self-discover because they care about improving their performance.
Zapier, a private company, is probably the extreme, and for an estimated $300M of revenue likely spends ~5% of those revenues on sales and marketing (we inferred this by counting the number of Zapier employees on Linkedin that had sales or marketing titles).
Just a totally different demographic.
Look, investors try to avoid the retailer demographic in their everyday lives as much as possible: it’s why delivery apps have grown the way they have.
So it’s not surprising investors have their finger WAY off the pulse here.
The second reason is very related to the first.
Retailers refuse to pay for anything.
It’s why companies like Toast and Shopify make way, way more money from payments – which are esoteric and effectively hidden from merchants – than they make from software.
Software is transparent.
Get a clean bill at regular time periods with a payment method on file.
Don’t like it?
Easily cancel.
Payments?
Good luck trying to figure that out.
Even people that work in payments don’t know what payments costs.
That’s because the card networks, issuers, and acquirers have intentionally complicated things so nobody knows how much money they make for providing a commoditized service.
Even if AI solved every problem a retailer had, the retailer 1) would never find it on their own, and 2) would never pay anything for it.
Look at this report Anthropic published charting the potential use for AI and the penetration of AI in said industry.
The graphic was reconstructed into something that’s easier to digest:
If you’re looking for restaurants, always go all the way to the bottom of a graphic where positive attributes increases as you move upward.
There is no virality for retail merchants.
Which sucks because the most valuable companies in the world (up until the AI hype) were built on product led growth, meaning that customers found the product themselves.
Google.
Facebook.
Amazon.
Naval Ravikant, founder of Angellist, famously stated
You’re doing sales because you failed at marketing. You’re doing marketing because you failed at product.
Yea, just doesn’t work when customers are so unsophisticated that they can’t self-discover.
The CEO of a restaurant chain – who only has the position because he happens to be the oldest person in the room – going to somehow vibe code a POS alternative?
Give us a break.
It’s also simultaneously hilarious because the people who work in retail spend tons of time on social media apps but can’t be bothered to search for business solutions.
Now, where things could get dangerous for incumbent vendors is if an AI platform charged for their services in a payments model.
This is what we think Toast will do as they run out of TAM.
Monetization occurs with every transaction through the AI system via payment processing.
Someone ordered from you on DoorDash?
+10% transaction fee.
Dine inside?
+7% transaction fee.
AI is the interface that the consumer interacts with (imagine just talking in plain English and the AI generates the order for customer confirmation, or even answers questions to refine the order) and then takes their pound of flesh in the payments stream.
OpenAI is silently doing this, which we’ll cover in another post.
But we still don’t know how you get this into a retailer’s hands without spending a lot of money.
Given the revenues these AI companies need to make their latest valuations work, maybe they’ll figure it out.
Because there is no demographic on Earth that will spend more on payments processing than retailers: they’re all such “super genius negotiators who are smarter than everyone” that they can’t possibly be fleeced.
And AI should totally use this to their advantage while they dunk all over the retailer.
Add comment