This article tells the story of a co-founder who started the vertical SaaS company "Butter" during the 2020 pandemic to address inefficiencies in the food service industry. Butter aimed to digitize the outdated ordering and inventory management systems between wholesalers and restaurants, integrate financial services, and create network effects through an ordering app. However, unexpected technical complexity, long sales cycles, low willingness to pay, and too many simultaneous initiatives made scaling extremely difficult. Ultimately, Butter was acquired by GrubMarket, and the founder shares the valuable lessons learned through this journey, offering insights for building successful SaaS in old-school industries.

1. Butter's Origins and Ambitious Plans 🚀

As the 2020 pandemic swept across the world, many businesses began to keenly feel the need for digital solutions. With software like Toast, DoorDash, and Shopify experiencing explosive growth, our co-founders believed the opportunity had arrived for food supply chain technology to ride the wave of digital transformation. This is when we founded "Butter."

At the time, two major problems stood out in the food industry:

1. Chefs' outdated ordering methods: Chefs were still carrying clipboards and paper order forms, calling or texting suppliers to order ingredients. The problem was that they wouldn't know until the next morning whether the ordered ingredients would actually arrive in the kitchen. Imagine the tremendous stress of worrying every morning whether you could actually serve the ribeye steak or cioppino on the menu!
2. Wholesalers' severe technological backwardness: The wholesalers supplying the chefs were in an even worse state, using 1990s-level technology. They were manually entering orders received via text or voicemail, using ancient DOS ERP systems, and recording inventory in Excel spreadsheets. Payments were mostly made by paper check! It took as long as six hours a day just to enter customer orders into the system — imagine if that time could be spent growing the business instead.

We spent months observing suppliers' workflows at 3 AM and thought we had found the answer. Our plan consisted of three major components:

1. Build an all-in-one cloud ERP system: Modernize the old systems to house core workflows and serve as a system of record, thereby reducing customer churn and driving high engagement.
2. Integrate financial services: Integrate payments, lending, payroll, and other financial services to dramatically increase average contract value (ACV) per customer. In 2020, a16z published a special article about the new wave of vertical SaaS enhanced with financial services, which gave us confidence.
3. Introduce seamless communication processes: Create seamless communication processes for both chefs and wholesalers, and build a platform network growth effect by incentivizing more suppliers to join the platform through a DoorDash-like ordering app. It would be a lie to say we weren't influenced by watching the ordering app "Choco" rapidly grow into a unicorn.

Orders were often handwritten with unreadable product names.

Looking at the ERP system of one of our early customers — a supplier of fish to Michelin-starred restaurants — they were still using custom DOS software developed in the 1980s that was no longer maintained. Our plan seemed so obvious that it felt like "this is going to be easy!" But we were deeply mistaken.

2. First Pitfall: Enormous Engineering Complexity and Customization Demands 🤯

We thought building a modern ERP system would be easy. The old systems were clearly outdated, so how hard could it be to build something better? We demoed prototypes to several local customers, received interest in test usage, and decided to accelerate development. But the next stage hit an enormous wall.

We solved technical problems like eliminating customers' local server failure risks, but the industry's requirements were far more diverse than we had realized. Every prospective supplier customer had specific requirements to actually use the system:

• Custom keyboard shortcuts for every operation ⌨️
• Exact column layouts on data entry screens
• Uniquely formatted invoices and warehouse printouts
• Even detailed information to be displayed on shellfish tracking tags.

Each customization demand became a black hole consuming engineering resources. What we thought would be a streamlined solution turned into a highly customized product requiring constant modifications and development cycles to meet each customer's needs and win the next contract. Engineering budgets snowballed as we tried to meet requirements. The "wedge" we had hoped for became a massive monster, and we realized it had no scalability.

(Side note: I recently saw YC's new RFS for ERP software, and if you really want to jump into this, please think very, very carefully.)

3. Second Pitfall: Painfully Long Sales Cycles 🐢

Switching ERP systems is nothing like upgrading your phone software. It's more like open-heart surgery. It affects every living part of business operations — warehouse, procurement, accounting, sales departments, and more. Customers hesitated because they knew how disruptive the process could be. Many companies knew their current systems were wasting time and money, yet put off the decision as long as possible. Even when the owner (typically the key decision-maker) wanted to switch systems, the complex workflows required buy-in and cooperation from every department, which severely prolonged the process.

Perhaps because of their old-fashioned way of doing things, most food distributors/wholesalers we targeted were barely keeping operations running. They had no capacity to actively seek out new software. The situation worsened especially during busy seasons for their end customers — restaurants — limiting our golden sales window to early spring and mid-fall.

These long sales and adoption cycles were devastating for us. We offered clear improvements, but getting companies to make bold changes was always an uphill battle, especially when it involved their core operations. Toward the end, our deal close rate was only 1/5 to 1/3 of our target. We had too many deals in progress but too few closed.

4. Third Pitfall: Low Willingness to Pay and Long Revenue Activation Periods 💲

The biggest shock came with pricing. Most of these businesses had razor-thin margins — often as low as 5%, sometimes none at all. For most, revenue was the only top priority. Therefore, selling something that wasn't directly tied to revenue growth was extremely difficult. They were accustomed to paying $80/month for QuickBooks, and getting them to pay enough for a full-scale ERP upgrade to justify our development costs was an enormous challenge.

We expected the fintech and ordering app components to significantly expand average contract value (ACV) per customer beyond existing SaaS subscriptions, and in theory, they did.

But in reality, activating payment/app usage revenue also took an incredibly long time. This was an additional problem on top of the already long ERP sales cycle. Both payments and the e-commerce app ultimately depended more on the preferences of end restaurant/retail customers, which were also old-fashioned. Suppliers didn't want to force credit card payments on their customers, and many customers stubbornly preferred ordering by phone or text, feeling a (arguably misplaced) sense of security that they'd be better taken care of that way.

As a result, achieving 20-30% of target adoption rates and additional revenue typically required 2-3 months of intense effort. This was too much effort for too little return.

5. Fourth Pitfall: Too Many Simultaneous Initiatives 🎯

Finally, our organization also struggled by trying to do too many things at once.

Since this article is meant to highlight the key problems and lessons we experienced, I'll skip the step-by-step details of every experiment we tried during our journey. However, I will say that we failed to find a true wedge before attempting to unlock additional revenue streams and network effects. We were running large-scale campaigns on three fronts when we should have stayed in guerrilla warfare mode to secure a single big win.

These execution issues prevented us from quickly and iteratively testing hypotheses, and the team was burning out. Even when acquisition offers came in, there were still several "what ifs" worth testing, such as moving upstream to higher-margin parts of the food supply chain or expanding the go-to-market strategy for a new lightweight AI-based product. But after approximately four years of development, we decided to combine our product and technology with an organization that already had established distribution channels.

6. Hard-Won Lessons 💡

Don't get me wrong — our team and I accomplished many things we're proud of. We woke up at 2 AM and literally slept in warehouses for weeks for successful onboarding, built extremely complex yet intuitive software that customers loved, and developed rigorous implementation playbooks, among countless other efforts. But overall, we failed to build a truly scalable venture business.

Our experience taught us one big lesson: Don't jump into an idea based on superficial hypotheses alone. Listen to the people who work on the ground, in the warehouses, and behind the kitchens. Talk to people not to confirm your preconceptions, but to find the truth.

Building a successful product in an old-school industry especially requires these four things:

1. People who deeply empathize with the problem: Unless the pain of maintaining the status quo is greater than the friction of change, nobody will switch systems, no matter how great your product or grand your vision. Inertia is powerful, but change only needs a few catalysts.
2. Deep enough pockets: If the customer can't afford the solution, all the value you deliver won't help your bottom line.
3. A product experience 10x better than the existing system: The more traditional the customer, the bigger the improvement needs to be. There's a reason they've stuck with their ways for so long — you need to solve the hardest problems in their world.
4. Simplicity is key: Your initial wedge product must be extremely simple to adopt and address a single fundamental customer unit. Are you selling a fancy Japanese bidet, or are you trying to replace their entire plumbing system?

To add a more nuanced point: I'm not saying that a startup will fail if all of the above criteria aren't met. But when all these elements are present together, there is usually a clear path to success.

To elaborate further, for a successful SaaS startup, deal size must be proportional to the time needed to close a deal. David Sacks presented a deal size/cycle time quadrant in his article "The Difficulty Ratio" that helps clearly visualize this relationship. I strongly recommend reading the full article, but in summary, high ACV/low deal velocity or the reverse can work, but if both are low, the company dies.

Relating back to the four criteria above: if your target customer doesn't have deep enough pockets but you have a clear wedge product with short sales and onboarding cycles, you can win in the SMB category through fast deal velocity. Conversely, if your product is complex but offers a clear value proposition that customers are willing to pay a premium for, you can secure significant deals in the enterprise quadrant.

But in Butter's case, unfortunately, we fell into the death quadrant of slow deal velocity + low ACV. (To be precise, ACV wasn't that low when factoring in additional revenue streams, but the deal velocity including the full revenue activation timeline wasn't just slow — it was at snail pace.)

Conclusion 🐌

Looking back, we underestimated the complexity of building vertical SaaS in a sector dependent on deeply entrenched practices and outdated technology. We thought simply offering a digital solution would trigger adoption. Instead, we learned that we had to prove the product could deliver major improvements over existing systems and meet customers' needs in a way they understand. Because if the product doesn't feel like an immediate fit, they'll stick with what they know.

In an upcoming article, I'll cover in more detail how AI is opening up new opportunities in these old-school sectors and how we're using AI to work through dead ends. Stay tuned