The quiet SMB revolution,eleven years later

Someone I recently met sent me an article they co-wrote, The Quiet SMB Revolution of the 2010s. Joe Lonsdale and Reid Spitz published it in April 2015, describing why small businesses had become a more viable market for technology startups.

The historical obstacle they described was distribution:

Each business is only a small customer, and it is expensive to engage with each one on an individual level.

Then they described a feedback loop. Smartphones made owners more comfortable with technology. Customers used services like Yelp and OpenTable, which gave businesses a reason to build a digital presence. That activity produced more information about businesses, which helped software companies find potential customers. Better distribution and cheaper technology made more of those customers economical to serve.

Eleven years later, I think that is still a useful way to look at the next opportunity ('History doesn't repeat itself but it rhymes'). What becomes affordable now that used to require too much individual attention?

They also said small businesses require “lower implementation and maintenance costs and a shorter learning curve,” which made this software “almost a consumer/enterprise hybrid.” I read that as software a company can adopt without an IT department that still handles consequential work like payroll and employee benefits.

Several objections to serving small businesses remain are still hanging around though: The subscription is too small to justify the effort. Every shop uses different systems. Every customer needs something custom. The work starts to look like a services business, with more people needed for every new account. And the owner has no time to learn another application. i.e. The deal feels like an enterprise deal but at a small price point that doesn't make sense.

I think agents give us a reason to revisit those objections, starting with what the customer is buying. What an owner will pay for access to a tool tells us only part of the story. Having customer inquiries answered, estimates followed up, or invoice problems resolved has a value of its own. A service that reliably handles that work can justify a different price. The value has to come from the work it completes not just show the owner the work that has to be done (Looking at all those B2B SaaS dashboards).

The software in the 2015 article already automated meaningful work, including payroll. Agents could extend that automation to jobs where the steps depend on information scattered across the business. I've written about how the owner connects records, experience, and judgment. Much of the opportunity is in helping with work the owner currently has to piece together.

Imagine a distributor receiving a supplier invoice that doesn't match what arrived. The purchase order is in one system. The delivery record is somewhere else. An email explains a substitution, and the owner remembers agreeing to a different price.

An agent could gather those records, ask about the missing agreement, and prepare a request for a corrected invoice. Once the owner approves it, the agent sends the request, tracks the response, and checks the correction against what was agreed. The owner decides the disputed terms while the agent handles the follow-through.

That also gives us a way to think about integration and adoption. The agent can work through email, messages, phone calls, and the applications the business already uses. The owner can start by explaining a job and reviewing its results, without first moving the whole business into a new system.

The next distributor will have different suppliers, approval limits, and ways of recording deliveries. Much of the underlying work will still be familiar: compare what was ordered, received, and billed; resolve the difference; follow up. Those common steps give the provider something to reuse, with each business's rules and exceptions kept specific to that business.

This is where I see another feedback loop. Each installation can show us which parts of the work repeat and which need individual treatment. Reusing what repeats should make the next installation easier. Agents can also help build the connections, test the workflows, and carry out the ongoing work. My bet is that we can reduce the human effort required to set up and serve each additional business while still giving it help that fits. That's the answer I want to test against the objection that every customer becomes a custom services engagement. I don't think SMB's have to get stuck on the B2B SaaS bell curve black hole where they always will need to get additional software or customizations to fill in what the SaaS won't do for them.

Existing software companies have a reason to pursue this, too. Adding AI inside an application is a useful starting point. Taking responsibility for a job that crosses email, accounting, scheduling, and someone else's software asks more of the product. It may also require changing how the company charges and supports its customers. That gives a company built around doing the work a different starting point.

Lonsdale and Spitz described changes that made small businesses easier to reach and serve. Eleven years later, I think agents could change both the value of what we deliver and the cost of delivering it and completely change the market dynamics.

I've written before about AI helping small businesses stay independent. That's why this matters to me. I want an owner to be able to buy dependable help with the work the business needs done, keep control of its information and decisions, and gain capabilities that used to require a larger organization. Doing that at a price a small business can afford is an opportunity worth building for.