A Series B should be a transition point. The company has moved beyond the earliest questiondoes anyone want this?and is now facing the more consequential one: can this become a repeatable, durable, and valuable enterprise?

That distinction matters more than ever. Capital has returned unevenly: US venture investment reached record levels in the first half of 2026, while investment, fundraising, and exits remained concentrated among a relatively small set of companies and funds. The exit market is improving, but it is not indiscriminate. In other words, a good company may be fundable; a great company must still be investable.

I do not treat Series B diligence as a box checking exercise, and I do not believe a single metric can carry an investment. A company can be growing quickly and still be buying revenue. It can have impressive gross margins and still lack a real market. It can have an exciting AI narrative and still be exposed to model commoditization, fragile unit economics, or shallow customer adoption.

My job is not to reward momentum. It is to determine whether momentum is becoming an advantage that compounds.

The Eight Questions I Need to Answer

1. I Start With Growth Quality, Not Growth Rate

The top-line number gets my attention, but it rarely makes my decision. At Series B, I want to understand the composition of growth before I celebrate its speed.

I look for a clear answer to a few practical questions. How much growth comes from new customers versus expansion? Which customer segment is actually pulling the product through the organization? What does the sales cycle look like by segment? How much of the pipeline converts without extraordinary founder effort? And, perhaps most importantly, what happened to the cohorts that arrived twelve, eighteen, and twenty-four months ago?

The best growth has a recognizable shape. The customer profile is becoming narrower before it becomes broader. Sales objections are predictable. Implementation is getting faster. Reference calls become more enthusiastic than the pitch deck. The company is learning which customers to decline, not merely which ones to chase.

I become cautious when growth depends on a small number of unusually large contracts, a rapidly widening definition of the ideal customer, or discounting that is framed as “strategic.” Those dynamics can create a convincing revenue chart while concealing an unrepeatable business. Revenue is an output; the machine that produces it is what I am underwriting.

2. Retention Tells Me Whether the Product Has Earned Its Place

Retention is where the story becomes real. Customers renew products they rely on, and they expand products that create value they can see.

I want both a revenue view and a behavioral view. Gross revenue retention shows whether the existing base is holding. Net revenue retention shows whether customers are deepening their relationship with the company. Neither metric is sufficient on its own. An expansion heavy result can obscure meaningful logo churn; a stable headline retention number can hide a customer-success team working far too hard to preserve renewals.

So I go beneath the aggregate. I ask what usage looks like after implementation. I want to see whether the product is embedded in a recurring workflow, tied to a budget owner, and connected to a measurable business outcome. I listen carefully for the difference between “customers like us” and “customers would have a problem if we disappeared.”

For usage based and AI enabled products, this diligence is even more important. Usage can be a sign of genuine adoption, but it can also be experimentation disguised as engagement. I care about repeat usage by the right users, whether consumption persists after the initial excitement, and whether the customer is willing to commit contractually as value becomes clearer.

3. I Underwrite Margins as a System, Not a Slide

A healthy gross margin profile is not a cosmetic virtue. It determines how much capital a company needs, how much it can invest in product and distribution, and what kind of business it can ultimately become.

I begin by making the cost of revenue honest. For a software company, that means more than cloud infrastructure. It includes third party data, implementation, human review, support, and any services required to make the product work in the real world. For AI-native companies, I also want to see inference and model provider costs separated clearly from broad infrastructure spend. If usage rises, do unit costs decline, remain manageable, or increase faster than pricing power?

The right answer is not a universal gross margin target. An enterprise product with meaningful onboarding can justify a different profile from a self serve product; a product capturing a high value workflow may support a different cost structure from a generic assistant. What matters is the direction of travel and the operating logic behind it. I want to see pricing, delivery, and product design reinforcing one another not a business hoping that scale will repair an economic model it has not yet solved.

That scrutiny reflects a broader market shift. Enterprise software companies are moving from pure seat based subscription models toward hybrid, usage based, and outcome oriented pricing, making the connection between value delivered, revenue recognized, and cost to serve more important to diligence.

4. Market Size Is a Question of Earnable Demand

Every company can produce a large top down market number. That is not the same as proving the company can win a large market.

I prefer a bottom up view. I start with the customers the company can serve today, the budget it is displacing or creating, the number of realistic buyers, and the expected value per account. Then I ask how the business expands. Is there a natural path from one workflow to another? Does credibility in the initial wedge create permission to sell adjacent products? Does the distribution model become more efficient as the company moves upmarket or across functions?

A narrow starting point does not concern me. In fact, it can be an advantage. Many enduring companies begin with a painfully specific problem for a clearly defined customer. What concerns me is a wedge with no credible adjacency a product that is useful, but structurally confined to a market that cannot support the ambition implied by the round.

I also separate category growth from company opportunity. A large, fashionable category can be a difficult place to build a durable business if switching costs are low, procurement power is concentrated, and every competitor can offer similar functionality. Conversely, a market that looks modest in a spreadsheet may be very attractive when a company has a unique route to expand its share of wallet.

5. Management Is the Multiplier I Cannot Model in a Spreadsheet

At Series B, the company is becoming too complex to be run by force of personality alone. The founders must evolve from being the primary source of insight to building an organization that produces insight, makes decisions, and executes without routing every important issue through them.

I evaluate management through evidence rather than charisma. How do the founders discuss a missed plan? Can they distinguish a temporary execution problem from a flaw in the underlying thesis? Have they hired people who are more capable than they are in critical functions? Do the board materials show operating discipline, or do they primarily reinforce the narrative?

I also pay attention to talent magnetism. Strong leaders attract capable people before every uncertainty has been resolved. They create clarity about what the company is trying to achieve, set a high bar, and make difficult decisions without creating unnecessary drama. The best teams are not flawless. They are unusually fast at confronting reality and converting it into action.

My reference calls are therefore not a formality. Customer references reveal whether the company can earn trust. Former employees reveal how it handles pressure. Executive candidates reveal whether the founders can recruit. Each conversation helps answer the same question: when the company encounters the next inevitable setback, will this team become more focused or more fragmented?

6. In 2026, AI Defensibility Must Be Specific

“AI-powered” is not an investment thesis. It is increasingly a baseline claim. SVB’s analysis of 9,000 VC-backed companies found that 42% of businesses that described themselves using AI terminology showed little evidence that AI was central to their technology. At the same time, 65% of US enterprise-software venture capital went to AI startups in 2025. The combination raises the bar: capital may reward AI exposure, but diligence must identify the substance beneath the label.

When I assess AI defensibility, I ask four questions. First, does the company have privileged access to data, feedback, or workflow context that makes the product improve over time? Second, is the product embedded in a system of record or a mission critical decision, rather than sitting beside the workflow as a disposable interface? Third, does it have a distribution advantage, proprietary integration, regulatory posture, or trust relationship that a model provider cannot simply replicate? Finally, do the unit economics improve as the company scales?

The four questions I return to when assessing whether an AI advantage can endure.

The strongest AI companies do not merely wrap an increasingly capable model. They turn intelligence into an operating advantage: better outcomes, faster implementation, lower cost, deeper workflow integration, or accumulated domain expertise. They build evaluation systems, human in the loop processes, and customer data rights with the same seriousness that earlier software companies brought to product architecture and security.

I am skeptical of a moat described only as “we move fast.” Speed matters, particularly in a fast moving technical market. But speed without customer ownership, workflow depth, or economic leverage can become a race in which every participant is running on the same infrastructure.

7. Capital Efficiency Is About Optionality, Not Austerity

I do not reward companies for spending as little as possible. I want companies to spend with purpose.

Series B capital should purchase a set of durable milestones: a repeatable go to market motion, a stronger leadership team, product depth, category credibility, and a business that can choose its next financing rather than be forced into it. I want to see a clear link between dollars invested and the evidence the company expects to create. If the plan calls for a significant increase in burn, I need to understand what becomes true at the end of that investment that is not true today.

This is where a cash forecast, hiring plan, and sales capacity model should all tell the same story. I look at revenue per employee, sales productivity, implementation capacity, cash conversion, and the time required for new spending to become productive. I also ask the less comfortable question: if financing conditions deteriorated, where could the company slow down without damaging the core business?

A capital efficient company has options. It can invest aggressively when the return is visible, slow down when the signal weakens, and negotiate from strength when it raises again. That flexibility is not defensive posture; it is strategic power.

8. I Need to See a Credible Path to Liquidity Before I Invest

I do not expect a Series B company to know the date or mechanism of its exit. I do expect it to understand what kind of asset it is building and who will value it at scale.

For an independent path, I ask whether the company has the potential to reach the scale, growth durability, margin profile, governance, and category relevance expected of a public business. For a strategic path, I ask which buyers would care, what capability or distribution they would acquire, and whether the company is becoming more or less essential as the market develops.

This is not theoretical. Even with signs of recovery in 2026, the venture exit environment remains uneven, and the quality of the eventual asset matters as much as the availability of capital today. In enterprise software, 46% of M&A deals had a US VC-backed buyer, a reminder that credible strategic outcomes can emerge from the innovation ecosystem itself.

I do not treat acquisition as a fallback plan. A company built for enduring customer value often becomes strategically valuable precisely because it has earned a unique position in a workflow, data layer, or market. The goal is not to predict a buyer. It is to determine whether the business is developing attributes that a public market or a logical acquirer will regard as scarce.

The Decision Comes From the Pattern

No Series B company will look perfect across all eight dimensions. The question is whether the strengths reinforce one another and whether the weaknesses are known, bounded, and fixable with the capital being raised.

The companies I want to back tend to share a pattern. Their growth is increasingly repeatable. Their customers stay and expand for reasons that are observable in the product. Their margin structure improves as they scale. Their market is larger than their first wedge, but their first wedge is real. Their leaders tell the truth early. Their use of AI is embedded in customer value rather than marketing copy. Their spending creates options. And the business is becoming an asset that someone will want to own not merely a company that investors are willing to finance.

That is the standard I bring to a Series B decision in 2026. If the answer is yes, I lean in. If the answer depends on the next fundraising round, a favorable market, or competitors standing still, I keep looking.