Nobody reads the boring clauses. They are where the money is.
Notice windows, auto-renewal, price escalators, liability caps. Not opinions about them — counts, drawn from the agreements themselves.
This page publishes nothing until the numbers can carry it.
Sidecar's benchmark corpus is being built now. Until a given statistic rests on at least 30 agreements, this page shows no figure for it — not a smaller sample with a caveat, and not a rounded guess.
That is the same rule the product applies to itself. A Sidecar review will tell you your notice window is longer than most comparable agreements only once there are enough comparable agreements to make "most" mean something.
What we will publish: median notice periods, how often agreements renew themselves, typical initial terms, how often price increases are uncapped, and how often liability is capped at something a reasonable person would accept — broken out by agreement type.
What we will never publish: pricing. Dollar benchmarks need sample sizes we do not have and probably will not for a long time, and a wrong one is worse than none. Nothing in this corpus can identify a customer, a vendor, or a contract.
A median is not a verdict.
Normal is not the same as acceptable. A ninety-day notice window is common in some categories, and it is still ninety days during which leaving is not your decision. What a benchmark buys you is the ability to tell the two apart — to know whether you are being asked for something ordinary or something the market does not actually require.
That is the difference between "this clause looks aggressive to me" and "this clause is longer than most comparable agreements, and here is the number." The first is an opinion a vendor can wave off. The second is a negotiating position.
Where these come from.
The corpus draws on agreements Sidecar has reviewed for customers together with material contracts filed publicly with the SEC. It is still being assembled, so the mix described here is how it is built rather than a description of a finished dataset. Public filings skew toward larger companies than Sidecar's customers, and we say so rather than quietly presenting them as small-business norms.
Every agreement is read by the same extraction pipeline the product runs, which is what makes the two sources comparable in the first place. Each statistic is gated independently: an agreement type can have enough contracts to report a typical term length and still not enough stating a notice period, in which case you will see the first and not the second.
Only derived statistics are stored. No contract text, no party names, no amounts, and no dates from the agreement itself — a record keeps the shape of a contract, such as how many days of notice it demands, never when it was signed or what it was worth. A record cannot be traced back to a customer or a counterparty.
Sidecar is business decision support, not legal advice, and not a law firm.
One email per edition. Unsubscribe in a click.
Find out where your own agreement sits.
Upload one contract. You get a verdict, every deadline, the clauses worth pushing back on — and, where the corpus supports it, how your terms compare. First review is free.