PRE-$1M ARR — INSTRUMENT THE BILLING METRIC BEFORE YOU BILL ON IT

Build the meter before the invoice. You cannot price on a number you cannot audit, and retro-fitting measurement into a live product is a repricing event.
Disqualify any prospect whose current billing runs clean. Only buy urgency you did not have to manufacture.
Write the "who gets fired if this breaks" question into your discovery script and record the answer verbatim; that answer is your future switching cost.
WATCH: the number of prospects who describe your problem unprompted, in their own words. Below five, you are too early.
REFUSE: pilots that never touch production data. A sandbox pilot proves interest and nothing else.

$1–5M ARR — SET A CONTRACTED FLOOR ON DAY ONE

Never sign pure-variable pricing. Land every deal as committed platform fee plus metered overage, so a customer's bad quarter does not become your bad quarter with no renewal conversation.
Ship the customer-facing usage dashboard BEFORE your first true-up invoice. Billing disputes at renewal are the cheapest churn to prevent and the most expensive to argue.
Hire a solutions engineer before a second salesperson. In infrastructure, the deal dies in technical validation, not in the pitch.
WATCH: percentage of new logos landing on two products, not one. Below 30%, your packaging is wrong.

$5–10M ARR — BUILD THE EXIT YOU HOPE NOBODY TAKES

Ship a complete, documented data-export path now, while it costs nothing. Buyers with an escape hatch sign faster, and the migration tooling you build for exits is the same tooling you use to steal competitors' customers.
Start audit-evidence work (SOC 2, revenue-recognition controls) at least two years before the segment that demands it. Certification lead time is what actually gates upmarket movement.
Cap bespoke configuration with a written rule: no code written for a single customer. Enforce at deal desk; the roadmap will never enforce it.
WATCH: implementation days to first invoice. Drive it down every quarter or it will drive your margin down for you.

$10–50M ARR — GIVE SERVICES ITS OWN P&L

Report professional services as a separate P&L with its own gross margin from the first dollar. Blended margin conceals the trap until the trap is structural.
Price implementation as fixed fee, never time-and-materials. Fixed fee forces you to productise the deployment; T&M pays you to keep it slow.
Hire a partner-enablement lead before your fifth implementation consultant. Every delivery hire you make is a partner you will later have to displace.
DECIDE: the hard ceiling for services as a share of revenue — set it at or below 15% and publish it internally.

$50–100M ARR — AUDIT DOWNGRADES SEPARATELY FROM CHURN

Split net retention into three reported lines: upsell, downgrade, and logo churn. A single blended number will show 110% while the downgrade line quietly doubles.
Ring-fence a product team from the enterprise roadmap and give it one job: the simple, cheap version. Fund it out of the enterprise P&L, not out of a new budget request.
Refuse the twentieth enterprise edge case in writing, with the deal you walked away from named. Complexity debt is only visible when you price the refusal.
WATCH: expansion revenue as a share of total growth. When new-logo revenue exceeds expansion, your installed base has stopped growing.

$100M+ ARR — RUN THE PROCESS WHILE GROWTH IS STILL DOUBLE-DIGIT

Start the sale process a year before you think you need to. A broad auction run from a position of decelerating growth finds bidders, not buyers.
Assume a broad process yields one financed bid, and negotiate as though it will. (Zuora's special committee contacted 30+ parties, 10+ ran diligence, one fully-financed bid emerged; $10.00/share, $1.7B, closed 14 Feb 2025.)
If the simple product does not exist by now, buy one rather than build it. At this growth rate you no longer have the years an internal build requires.
DECIDE: which owner matches the next five years of strategy. The highest number and the right owner are rarely the same party.