Mark Roberge on Sales Hiring, Unit Economics and Moving Upmarket

Mark Roberge, co-founder of Stage 2 Capital and former HubSpot sales leader, tells Harry Stebbings that selling a large enterprise deal requires distinct skills: building a champion, navigating organizational politics, and working through procurement and legal review.

Roberge entered sales almost accidentally. He wanted to be an entrepreneur and was deciding between marketing and sales, attracted by the art of a deal and the potential income. While consulting for the early HubSpot, co-founder Brian Halligan asked him to sell, giving him a chance to learn both sales and marketing software.

He had met Dharmesh Shah at MIT, where both pitched business ideas in class. Their friendship led Shah to hire him as a consultant while recruiting Halligan as co-founder and CEO. Roberge spent about six or seven months working part-time before joining full-time.

By then, Roberge recalls, he and Halligan had acquired about 40 customers and HubSpot had a few hundred thousand dollars in annual recurring revenue. With a roughly $5 million Series A in prospect, Halligan invited him to become the first full-time salesperson and sales leader.

The shift from founder-led to team-led sales is difficult, especially for product-oriented founders without sales experience. Roberge says they often overvalue a candidate’s years selling a similar product to similar customers and underassess actual selling skills. He recommends finding an experienced sales-hiring adviser.

Founders should develop the foundations of the sales playbook: understand how buyers describe the problem, how they decide, which customers fit best and what value the product offers. A professional sales leader can add methodology and refine discovery, outreach, objections and negotiation.

For a first sales hire, Roberge gives greater weight to experience with a similar deal size than a similar customer category. A person who has sold million-dollar deals to hospitals may already understand champions, economic buyers, procurement and legal processes; learning banking can be easier than learning all those enterprise skills from scratch.

He uses a role-play to assess candidates. For example, a candidate might conduct a first call with a small-business owner who downloaded a marketing guide. Roberge looks for open-ended questions, follow-up, discovery and qualification rather than a long generic sales pitch.

He then tests coachability: ask the candidate to assess their own performance, give positive and constructive feedback, and repeat the role-play. In a startup, even an experienced salesperson needs to learn the product, customer and method; the speed at which they use feedback matters.

A weak first few months do not always mean a bad hire. Roberge has seen representatives take three or four months to find their footing and become top performers later. He asks managers to diagnose a specific skill, coach it and observe whether the improvement holds.

An excessive fixation on title can be a warning sign about ego and culture, he says. Still, he does not think offering a strong candidate a more senior title necessarily makes it impossible to add management layers later; the real issue is agreeing on the job and assessing performance over time.

Compensation should reflect what the business needs to accomplish over the next six to twelve months, not simply copy a sales leader’s previous plan. Roberge asks which salesperson behaviors will support the company’s strategic objectives and designs incentives around them.

Before product-market fit, he prefers a salesperson who will work alongside engineers on discovery and iteration, with salary and equity rather than relying heavily on commission. If the team needs weeks to revise the product, the salesperson should not be forced to chase a short-term commission while everyone learns.

After product-market fit, the objective changes to finding a repeatable go-to-market motion without losing customer success. Roberge wants representatives to sell to customers who fit the ideal profile and are likely to obtain value, rather than just maximizing initial bookings.

One example is paying half a commission when the customer pays and the other half when the customer reaches an early indicator of retention, such as completing setup and a first transaction. The indicator should be meaningful, reachable early and associated with longer-term success; it is an example, not a universal formula.

In product-led or land-and-expand businesses, paying almost entirely on the initial contract can create the wrong incentive. Customers and the company may both prefer a small initial deployment followed by expansion, while a representative paid only for the first sale has reason to push a large purchase immediately.

As an illustrative plan, he suggests 3% on the initial annual contract value and 5% on later expansion. The salesperson can keep hunting while a customer success team handles onboarding and expansion, but their compensation still rewards bringing in customers who are likely to grow.

Stebbings worries about an enterprise-style sales cost base supporting low-priced products that never expand. Roberge says HubSpot started at around $3,000 a year, at the lower end for a sales-led model, but benefited from substantial inbound demand. Annual contract value alone cannot determine whether a sales team is economical.

The answer depends on unit economics: acquisition cost, contract value, gross margin and the sales motion. A simplified payback calculation divides customer acquisition cost by annual contract revenue, ideally adjusted for gross margin. At $10,000 acquisition cost and $10,000 annual revenue, unadjusted payback is about a year.

Outbound acquisition cost varies with the number of calls required to get a meeting, the proportion of meetings converted to demonstrations and the proportion that close. A hospital executive and a small-business owner may be very different to reach. There is no universal contract-value threshold for hiring salespeople.

Roberge offers stage-dependent payback heuristics rather than hard laws. Near an IPO, he considers under 12 months excellent, 12–15 good, 15–20 concerning and over 20 problematic. At Series A he allows more room for improvement, though above 20 months still concerns him.

The calculation is only useful if acquisition costs are complete. At scale, Roberge says, the sales and marketing expense line is a useful starting point, including sales leadership and operations. At early stages, allocating founder time and shared activities can be less clear, but excluding those costs can make the metric misleading.

Lifetime value relative to acquisition cost is another lens, but early-stage estimates rest on limited data. A common simplified model divides annual contract value by churn; Roberge notes that a reliable lifetime churn history may take years. The model’s assumptions and revenue-versus-margin treatment matter.

Asked about small-business churn, he recalls HubSpot reaching 8% monthly churn at its worst point in February 2008 and generally seeing 3–4% a month during that year. He considers those levels very high and says a company with such churn would need many other things to work exceptionally well.

Serving small businesses can nevertheless be viable, especially with a healthy expansion motion. Roberge offers an illustrative range of 80–85% logo retention for a strong SMB business and argues that expansion must offset losses to get net revenue retention above 100%. Those are his heuristics, not guaranteed benchmarks.

He values both gross and net retention, but gives net revenue retention more weight because it incorporates expansion. A sustained 120% net retention rate implies an existing revenue cohort grows 20% over the period without new customers. Gross and logo retention still matter because many dissatisfied customers can damage reputation.

Stebbings argues that some product-led users who leave were never an ideal fit. Roberge agrees that not every departure is a failure but worries that customers talk to one another without analyzing the seller’s ideal customer profile. Poor experiences can travel even outside the target segment.

Roberge illustrates why sub-100% net retention becomes harder at scale. With $10 million in starting revenue and 90% retention, a company needs $1 million in new revenue merely to remain flat. At $1 billion, the same rate requires $100 million in new revenue before any growth.

For most software startups, he says moving into enterprise before a few million dollars in revenue is unusually risky. Some businesses must sell to large governments or organizations from the beginning, but those have different funding needs and are outside his typical investment focus.

A big enterprise logo can take a long time to win; a champion can leave, and procurement may demand security or compliance the startup has not built. Roberge advises identifying the smallest customer segment for which the product delivers value, even if that segment is not the very smallest business.

Experienced sales leaders often say to move upmarket when larger customers pull the company there. Stebbings warns that a pull can arrive before the business is ready. Roberge agrees: being pulled may be necessary, but it is not sufficient, and leaders still need to judge product and operational readiness.

He recalls HubSpot declining a potential Facebook contract of roughly $3 million when HubSpot itself had only $2–3 million in revenue. He says the deal could have distorted the roadmap and distracted the company. The figures are his recollection of the proposed deal, not reported revenue from a closed contract.

A single unusually large inbound deal can also mislead the sales team. Roberge describes another startup whose representatives saw a $200,000 contract against a typical $15,000 annual deal and chased similar opportunities, only to discover they did not yet know how to sell them systematically.

Large customer logos can help with credibility, but he believes their effect is often overstated. Prospects frequently want proof from a company very much like their own, so even a strong case study may prompt questions about industry, location or customer type.

He suggests a green, yellow and red approach to the ideal customer profile. Green accounts are strong fits for active outreach; red accounts should not be sold to; yellow accounts can be served if they arrive inbound but do not yet justify outbound investment. Early boundaries remain hypotheses that customer outcomes should refine.

Roberge recalls that nearly all HubSpot customers came inbound for its first two or three years. He estimates that by year five the mix was roughly 50% inbound, 30% channel and 20% outbound. As the brand grew, attribution became less clean because many prospects had already encountered a free product or event.

Founders often assume a large partner’s thousands of salespeople will readily sell a startup’s product for a commission. Roberge says this greatly underestimates the work of mobilizing a channel: the startup needs executive alignment, a reason the product supports the partner’s strategy, and incentives that attract representatives’ limited attention.

A spiff, he explains, is a one-off sales bonus. The representative has spent time winning a meeting and must choose what to show the customer. A small commission alone may not compete with the partner’s own priorities; the startup needs to make the value and incentive compelling.

He describes Salesforce as an early HubSpot investor and later a channel partner, before the companies became competitors. Salesforce leaders saw that customers needed help generating leads, while its own representatives were not yet equipped to address that demand. HubSpot’s offering fit that strategic gap.

According to Roberge, Salesforce leadership selected HubSpot among a small group of partners whose sales counted toward representatives’ quotas and announced the program at a sales kickoff. Even with that support, HubSpot initially received only about 1% of revenue through the channel.

HubSpot then assigned a strong salesperson to the partnership. Roberge says the employee worked inside Salesforce offices and systems, supported representatives and visited sales teams. After a year, the channel contributed roughly 10% of HubSpot revenue, illustrating the sustained work required beyond signing a partnership.