Datadog's price list is an org chart

Open Datadog's pricing page and count the rows. Not the tiers — the products.

Infrastructure. APM. Log Management. Real User Monitoring. Synthetic Monitoring. Database Monitoring. Network Monitoring. Serverless. CI Visibility. Incident Management. Cloud Security. Sensitive Data Scanner. LLM Observability. The list keeps going, and it keeps going in a very specific way: each row is priced separately, on its own meter, with its own free allowance and its own overage math.

Most people read that page looking for a number. The more interesting read is structural. A pricing page this fragmented is not a pricing page — it's a public disclosure of how the company divides its own work.

The unit of billing is the tell

Infrastructure bills per host. APM bills per host too, but a different definition of host. Logs bill on two axes at once — volume ingested and events indexed — because ingest and retention are genuinely different costs. RUM bills per thousands of sessions. Synthetics bills per test run, split between API tests and browser tests. Serverless bills per function.

Nobody designs that on purpose from a blank page. A single team building a single product picks one unit and makes everything fit it. A dozen different units means a dozen different origin stories.

And the unit tells you who the buyer is. Per-host pricing addresses someone who owns servers and knows how many they have. Per-session pricing addresses someone who owns a web frontend and thinks in traffic. Per-test-run pricing addresses someone who owns a CI pipeline. Those are three different people with three different budgets, and Datadog has quietly built a price list that lets a sales rep walk into any of their offices without changing the pitch.

That's the strategy the page is admitting to: land in one org, expand into the next one over. The billing unit is the door.

Count the SKUs, find the seams

New Datadog products tend to show up already finished. A SKU appears on the pricing page with its own documentation tree, its own vocabulary, its own onboarding flow, and — critically — its own billing unit that doesn't match anything already there.

That combination is the signature of absorption rather than internal construction. Software built inside an existing platform tends to inherit the platform's nouns. Software that arrives from outside keeps its own. When you see a product whose docs use "monitors" where the rest of the catalog says "alerts," or whose settings live in a separate section of the console, you're looking at a seam.

You can date those seams without any inside information. Pull the pricing page from the Wayback Machine at six-month intervals and watch rows appear. Cross-reference the appearance date with the docs sitemap and the changelog. The gap between "the docs went live" and "the SKU appeared on pricing" is roughly how long it took to make the acquisition billable — and that gap has been getting shorter, which is its own signal about how practiced the integration machinery has become. We've written before about what a changelog leaks about the roadmap; a pricing page leaks the same thing on a slower clock, with money attached.

Where the page stops being confident

The rows aren't uniform, and the inconsistencies are the good part.

Some products carry a generous free tier and aggressive per-unit pricing above it. That's a volume play against a specific competitor — the free tier exists to make switching cheap and the overage exists because they expect you to grow into it. Other rows have no published price at all, just a contact-sales link. That's not always an enterprise signal. Sometimes it means the product's unit economics aren't settled yet and nobody wants a public number they'll have to walk back.

Then there are the bundles. When two SKUs start appearing together at a combined price, someone decided those products share a buyer. When a bundle quietly disappears, someone decided they don't. Bundling changes are among the highest-signal pricing edits a company makes, and they're almost never announced — which is exactly why they're worth catching. The same logic applies to quiet A/B tests on a pricing page, and to the broader patterns you find when you read enough pricing pages side by side.

The honest limit here: none of this tells you revenue, headcount, or what shipped last quarter. It tells you what the company believes is a product versus a feature, and how it thinks its buyers are organized. That's a narrower claim than a teardown usually makes, and it's the one the evidence actually supports.

The read, and the reason to keep reading

A pricing page like Datadog's rewards a slow first pass and then rewards diffing forever. The first pass gives you the map. Everything after that is watching the map change: a unit redefined, a free tier trimmed, a bundle formed, a row moved above the fold.

That's the part manual review loses. You can read a competitor's pricing page carefully once. Reading it carefully every week, across a dozen competitors, and noticing that one line item changed its unit from "per host" to "per container" — that's not a human task. Seeto watches public surfaces like this continuously and surfaces the diffs as discrete change events, so the unit change lands in front of you as a thing that happened on a date. It won't tell you what the change means. That interpretation is still yours, and on a page this structural it's the whole value.

Read the price list like an org chart. Then watch for the reorg.