You will not get a private company's financials. What you can get is the shape of how they make money, which is often more decision-relevant than the numbers — because the shape predicts behaviour, and behaviour is what will affect you.

Business models leave fingerprints in public. Here is where they are.

The pricing page is the primary document

Read it as a strategic statement rather than a price list. Four things are legible:

Who they want. Where the tiers break tells you which customer they are optimising for. A jump from $50 to “contact us” says the middle of the market is not their business.

What they consider valuable. Whatever sits behind the highest paywall is what they believe buyers will pay most for. That is a considered judgement about their market, made by people with more data than you.

How revenue scales. Per seat, per usage, flat, or hybrid. This determines everything about their incentives. A per-seat vendor wants deployment breadth; a usage-based vendor wants depth. They will behave accordingly at renewal.

Where the surprise bills come from. Overages, add-ons, and metered items are where customer frustration accumulates — and where discovery questions find purchase.

Archived versions are better still. A pricing page that has moved upmarket over three years is a strategy visible in slow motion.

Free tiers and trials tell you about acquisition cost

A generous free tier is expensive to run. Companies do it when the alternative — paid acquisition or a sales team — is more expensive still, which implies a low-touch, high-volume model.

A short trial gated behind a demo call implies the opposite: a sales-led motion where the cost of a human conversation is justified by deal size.

A company running both is usually mid-transition, and transitions are where competitors are vulnerable.

Hiring reveals the cost structure

Count open roles by function. The ratio is the model.

  • Heavy sales relative to engineering — high-touch, expensive acquisition, likely enterprise.
  • Heavy engineering, thin sales — product-led, betting on the product to sell itself.
  • Heavy support or services — either the product needs help to work, or services are a real revenue line. Both matter.
  • Heavy implementation or onboarding — a long time-to-value, which is a competitive weakness and a switching-cost moat simultaneously.

Roles also date the strategy. Hiring precedes shipping by two to four quarters, so the current listings describe next year's company. See the public record.

Customer concentration, inferred

Rarely disclosed, often inferable. If every case study is a large enterprise, the revenue is concentrated regardless of how many logos appear on the homepage. If the careers page lists named-account roles, they have named accounts that matter.

Concentration cuts both ways: it makes revenue predictable and it makes a single churn event material. For a vendor decision, it tells you how much attention a customer of your size will receive.

Retention signals

You cannot see churn. You can see things correlated with it:

  • Annual-only pricing often indicates monthly churn they would rather not expose.
  • Aggressive multi-year discounts suggest a need to lock in revenue.
  • Heavy customer-success hiring against flat sales hiring suggests retention has become the priority, which usually means it became a problem.
  • Public community activity — an active, long-lived user community is one of the better retention signals available, and it is very hard to fake.

Putting it together

Individually these are weak. The method is triangulation: form a hypothesis from the pricing page, then test it against hiring, case studies and community evidence.

A worked example. Pricing shows per-seat with a low entry tier and an enterprise tier behind a call. Hiring is 60% sales, mostly enterprise. Case studies are all large companies. Community is quiet.

The hypothesis that fits: a company that started self-serve and is moving upmarket, where the low tier now exists mainly as a lead source. The prediction that follows: small customers will get progressively less attention, the entry tier will get worse or more expensive, and the sales motion will lengthen.

That prediction is useful whether you are buying from them, competing with them, or considering joining.

State confidence honestly

All of this is inference from indirect evidence, and it should be labelled as such. “Their hiring pattern suggests an upmarket move” is defensible. “They are moving upmarket” is an overclaim that will eventually be quoted back at you.

The distinction between observation and inference does most of the work here — keep it visible, and a reader can weigh your reasoning instead of taking it on trust.

Our Company Deep-Dive works through business model, financial signals, product, customer voice, brand and leadership on a single company, with sources cited and every estimate labelled as an estimate. The broader structure is in how to research a company.