Founder-Led Sales to $1M ARR With Just 10 Customers

概览

Seven Learnings helps online retailers and brands improve pricing, marketing, and ordering decisions through predictive decision automation. Felix Hoffman describes the product as a “Google Maps” for retailers: customers define growth goals, and the system recommends product-level decisions to reach them.

The company began unusually, not with a SaaS contract, but with a consulting project that gave the team access to the large datasets needed to build forecasting and optimization technology. The first live pricing upload failed, but later A/B tests showed meaningful profit uplift, including a 13% improvement.

A major theme is founder-led enterprise sales. Felix says the first 10 customers, and even many after that, required direct founder involvement, personal networks, events, referrals, and customer champions.

The episode also contrasts traditional machine learning with LLMs. Felix argues that enterprise decision automation needs deterministic, cheap, accurate, and explainable systems, so LLMs alone are a poor fit for pricing and marketing optimization.

分段落总结

[00:48] Introduction to Felix Hoffman and Seven Learnings

[事实] The host introduces Felix Hoffman as co-founder and CEO of Seven Learnings. [事实] Seven Learnings reached its first $1M ARR with 10 customers, sold through founder-led sales. [事实] The episode covers how the company used a consulting project to get data, learned from a failed first pricing upload, and why Felix does not see LLMs as suitable for his product. [推测] The episode is positioned for SaaS founders who want practical lessons on enterprise sales, AI product development, and pricing.

[02:22] What Seven Learnings Does

[事实] Seven Learnings helps brands and large retailers improve decisions around prices, marketing, and ordering. [事实] Felix compares the product to Google Maps for retailers: the customer states the destination or growth ambition, and the system helps determine the product-level decisions needed to get there. [事实] The company had recently crossed $5M ARR, with around 40 customers and 60 employees. [推测] The product’s core value is less about giving dashboards and more about automating complex commercial decisions at scale.

[03:48] Origin of the Idea

[事实] Felix worked for six years at consulting firm Kearney on pricing projects in retail and industry. [事实] He then spent two years at Zalando, a major European fashion marketplace, where he saw predictive decision-making in practice. [事实] He believed many companies were not using that kind of approach and saw an opportunity to build a better technology stack. [推测] His founder-market fit came from combining consulting knowledge of customer pain with operational exposure to advanced retail decision systems.

[05:18] Finding Co-Founders

[事实] Felix says finding co-founders was one of the hardest early challenges. [事实] He initially wanted four co-founders because a Berlin program would fund four founders for a year. [事实] Finding technical co-founders in Berlin was difficult because strong technical people could earn well without taking startup risk. [事实] He eventually found a technical co-founder who was a friend and was interested in AI-driven decision automation. [推测] The funding program shaped the early founding team plan, even though the final team did not match the original ideal.

[06:59] Starting With Consulting to Get Data

[事实] Felix felt he knew roughly what product he wanted to build, but the challenge was getting data and a customer willing to engage. [事实] He says the product required large datasets, so working with a small retailer would not have been enough. [事实] The first contract was a consulting project, not a SaaS contract. [事实] The customer allowed Seven Learnings to use its data to develop the product, while also paying for consulting work. [推测] The consulting-first approach reduced data-access risk and helped the team build technology before having a mature SaaS product.

[09:11] Forecasting and Optimization as Core Technology

[事实] Felix says decision optimization depends on accurate forecasts, similar to weather forecasts that improve over time but are never perfect. [事实] The system predicts outcomes from decisions such as coupons, discounts, or increased Google ad spend. [事实] Seven Learnings built two core technologies: a forecasting mechanism and an optimization mechanism. [事实] Building the frontend and broader product took around a year. [推测] The product’s defensibility depends on the quality of prediction and optimization, not just the user interface.

[11:26] First SaaS Customer and Founder-Led Sales

[事实] The first customer came through Felix’s background and network connected to Kearney. [事实] Felix describes this as founder-led sales. [事实] He says founders usually need to sell the first 10 customers themselves, and remain involved even after that. [事实] He believes sales cannot easily be delegated early on. [推测] For complex enterprise SaaS, founder credibility and domain knowledge are central parts of the sales motion.

[12:40] Using Paid Pilots and A/B Tests

[事实] Seven Learnings typically demonstrates value through A/B tests, optimizing part of a customer’s prices while the customer continues its existing method elsewhere. [事实] The company charged a monthly fee even for the initial pilot. [事实] Felix says expensive SaaS products need a way to show value clearly. [推测] A/B testing served both as proof of value and as a sales tool for overcoming skepticism.

[13:48] The First Pricing Upload Failed

[事实] Felix says the first live upload of prices was a disaster. [事实] The team could see within a day that the pricing changes were not working. [事实] They were too expensive on high-priced products and had to rework the models. [事实] The customer remained patient because the problem was important and they had no better solution. [推测] The failure shows why early customers for high-impact AI products need both urgency and tolerance for iteration.

[15:58] Profit Uplift and Pricing Model

[事实] Later uploads or tests produced a 13% profit uplift. [事实] Felix says the impact was large enough to support a higher price point. [事实] Seven Learnings charges a monthly fee based on the amount of revenue being optimized. [事实] The company avoids success-based fees because they add complexity and increase pressure around A/B test interpretation. [推测] The pricing model is value-based but deliberately simpler than pure performance pricing.

[18:53] Getting to the First $1M ARR

[事实] Felix says the company reached about $1M ARR with its first 10 customers. [事实] Personal network helped, but events were especially important for acquiring early customers. [事实] Customer referrals and customer champions were valuable because new prospects wanted to talk to existing customers. [事实] Felix recommends putting customer participation in contracts, such as joining a fair or webinar. [推测] Social proof was essential because the company was selling a high-stakes product to cautious enterprise buyers.

[21:03] Events, Outreach, and Account-Based Marketing

[事实] Seven Learnings tried outreach, and Felix says it worked to some extent early on. [事实] He believes cold outreach gets worse every year because spam filters improve and prospects receive more messages. [事实] For a small pool of potential customers, mass outreach risks burning valuable accounts. [事实] He recommends a more account-based approach with highly customized emails for specific target customers. [推测] The company’s sales strategy fits a narrow, high-value enterprise market better than a high-volume outbound model.

[22:24] Ideal Customer Profile

[事实] The ideal customer profile has remained mostly the same: mid-market to larger retailers. [事实] Felix says the approach requires a certain scale because optimizing too little revenue does not pay off. [事实] Seven Learnings starts at about 25 million in annual turnover. [事实] The company serves fashion, furniture, pharmaceuticals, and other non-food online retail categories. [推测] The product is category-flexible because the model focuses on decision dynamics rather than the specific product type.

[23:59] Event Tactics That Worked

[事实] Felix says e-commerce and online retail events were the best fit. [事实] Speaker opportunities worked well, especially when a founder could present. [事实] Master classes also worked because leads could sign up to learn about part of the technology. [推测] Educational event formats likely helped because the product required changing how customers thought about pricing.

[25:04] Common Sales Objection: Price Matching

[事实] Felix says a major objection was that many retailers already used price matching. [事实] Retailers often crawl competitor prices and match them, sometimes assuming Amazon knows what it is doing. [事实] Felix argues that blindly following competitors ignores effects on margin, sales rate, stock, and season-end inventory. [事实] He says companies can still follow competitors, but should understand the predicted consequences. [推测] Seven Learnings had to sell a different decision philosophy, not just a better pricing tool.

[27:11] Why Charging More Can Make Sense

[事实] Felix gives examples where products sell out, such as bikes during COVID in Germany and climate control products during a hot summer. [事实] He says price matching only makes sense with unlimited supply, which many retailers do not have. [事实] If a retailer lacks enough stock, selling out at a low price may destroy profit potential. [事实] For excess seasonal stock, retailers must decide whether to discount now, advertise more, or risk disposal later. [推测] The value of predictive pricing comes from connecting price, demand, inventory, and future costs in one decision.

[29:05] Tariffs and Cost Prediction

[事实] Felix says Seven Learnings predicts not only sales but also costs. [事实] Tariffs can increase purchase costs, forcing retailers to decide whether to raise prices. [事实] If prices rise, retailers must also decide how much less to order. [事实] These decisions depend on how the rest of the market reacts and where competitors source products. [推测] The tariff example illustrates why pricing cannot be optimized in isolation from supply chain and competitive context.

[31:05] Difficulty of Demonstrating Value

[事实] Felix says A/B tests are useful for proving value, especially as companies care more about ROI from AI spending. [事实] He describes the product as almost “profit as a service,” because the company is promising profit uplift. [事实] A/B testing creates operational pressure around test design, split setup, evaluation, and communication. [事实] Seven Learnings is still debating when A/B tests are the best sales method. [推测] The company faces a tradeoff between making sales easier and making implementation harder.

[33:48] Why LLMs Do Not Fit Pricing Decisions

[事实] Felix argues that enterprise decision automation needs systems that are deterministic, cheap, and accurate. [事实] For those reasons, he says LLMs do not make sense for pricing or marketing optimization. [事实] He says founders should start with the customer problem and then choose the best technology, rather than deciding in advance to use LLMs. [事实] He believes successful LLM products will become software products that use LLMs as part of the stack. [推测] Felix is not anti-LLM; his objection is to using LLMs as the default solution for problems better suited to machine learning or deterministic software.

[37:43] Explainability and Machine Learning

[事实] Felix says enterprise customers need explainability when decisions are automated. [事实] He argues that LLMs produce outputs that are difficult to explain. [事实] Seven Learnings uses intermediate predictions such as sales and profit margin, which can be checked against real outcomes. [事实] He believes most enterprise decision-making will rely more on machine learning than on LLMs alone. [推测] The intermediate prediction layer is important because it makes the system auditable and trustworthy.

[39:26] Frontend and UX Still Matter

[事实] Felix believes a good, use-case-specific frontend remains an advantage. [事实] He says explainability often requires visualizations of predictions. [事实] He argues that customers want continuity and consistency in normal workflows. [事实] He says LLM-generated dashboards may work for occasional questions, but repeated business workflows need stable interfaces. [推测] The discussion rejects the idea that agentic chat interfaces will fully replace enterprise SaaS UX.

[41:43] The Risk of Building Another LLM Wrapper

[事实] Felix says LLMs make his company more productive, but he warns founders against defaulting to building another wrapper. [事实] He says early LLM wrappers could succeed because they were first, but now that opportunity is much harder. [事实] He believes founders should build real software that is better or cheaper, not merely a thin layer over an LLM. [事实] He expects some low-value SaaS products to struggle, but does not believe decision optimization SaaS will be replaced by LLMs alone. [推测] The stronger long-term opportunity is in domain-specific software with clear value, not generic AI packaging.

[45:24] Lightning Round

[事实] Felix names management by objectives as an important piece of business advice. [事实] He recommends The Master and Margarita by Bulgakov, saying it is important to calm down in the evening and not focus only on business. [事实] He says hiring a senior backend engineer very early was the best money spent on the business. [事实] His productivity habit is prioritizing things he can finish that day. [事实] His passion outside work is beach volleyball, including playing at Beach Mitte in Berlin.

播客点评/总结

This episode is valuable because it gives a concrete, non-glamorous account of building AI SaaS in a complex enterprise category. The strongest parts are the details on starting with consulting to access data, using A/B tests to prove value, and surviving an early failed implementation.

The discussion is especially useful for founders building products that depend on customer data, measurable ROI, and enterprise trust. Felix’s comments on founder-led sales, events, referrals, and account-based outreach are practical because they come from a narrow, high-value market rather than generic SaaS advice.

A limitation is that the episode is mostly from the founder’s perspective, so the customer side of the early failures, buying process, and implementation burden is only described indirectly. Some claims about LLMs and the future of SaaS are opinionated and should be treated as Felix’s viewpoint rather than proven outcomes.

[推测] The best audience is SaaS founders selling complex B2B products, AI founders deciding whether LLMs are the right tool, and operators interested in pricing, retail optimization, and enterprise go-to-market strategy.