A consultant tells you about a company that needs what you sell. Two weeks later your sales team closes that company. Was it a referral?
Everyone has an answer and they do not match. The consultant remembers the introduction. Your team remembers that the company was already in the pipeline from a webinar in March. Both are telling the truth, and there is no document that decides it, so the relationship ends over a fee neither side can prove.
That is what lead registration exists to prevent, and it is the only part of a B2B referral agreement that does any real work.
TL;DR: Make the partner register a named company through a defined channel. Give yourself a few business days to reject it on stated grounds, and say what those grounds are. Anything not rejected is accepted and protected for a set period. Then pick a fee trigger deliberately: on an accepted intro, on a meeting held, or on a closed deal are three different risk positions. The referral agreement generator writes a draft around whichever you pick, and it is a starting point for a lawyer rather than legal advice.
Why is this not an affiliate agreement?
Because the thing being paid for is different, and the mechanism follows from that.
An affiliate sends traffic. The system knows which link a visitor came through, so the document turns on the attribution window, the refund clawback and whose tracking data wins. An affiliate agreement is mostly about measurement.
A referral partner introduces a named company. There is no click to track. The document has to create the record that an affiliate link creates automatically, which is registration, and then protect it. Using an affiliate template for a referral deal leaves out the entire mechanism and keeps the clauses that do not apply.
How does lead registration work?
Four parts, and leaving any of them out reopens the argument.
A defined channel. A form, an email address, a shared spreadsheet, a partner portal. Anything, as long as it is named in the agreement and produces a timestamp. "Tell your account manager" is not a channel, because the record of it is a memory.
What a registration contains. The prospect's company name, a contact person, and a short note on the opportunity. The note matters more than it looks: it is what distinguishes a genuine introduction from a partner submitting a list of companies in your market and waiting to see which ones buy.
A rejection window with stated grounds. You need a few business days to check, and you need to be able to say no. The grounds that cover almost every real case are: the prospect is already a customer, is already in an active sales process, has been registered by another partner whose protection has not expired, or you are already in direct contact. A rejection has to say which applies. A rejection right with no stated grounds and no deadline is the clause partners refuse to sign, and they are right to.
A protection window. Anything not rejected is accepted, and the prospect belongs to that partner for a defined period. This is the clause that turns an introduction into an asset the partner can rely on.
Then one more line that costs nothing and prevents a lot: tell the partner when a registration is rejected, when a prospect goes quiet, and when one signs. A partner working a prospect that is no longer available is wasting effort on your behalf.
How long should the protection window be?
Match it to your sales cycle and add room, because the cost of too short is much higher than the cost of too long.
| Your sales cycle | Protection window |
|---|---|
| Weeks, self-serve with a human touch | 60 days |
| One to two months, mid-market | 90 days |
| Three to six months, enterprise | 180 days |
| Longer, or committee-driven | 180 days with extension on an active deal |
Too short is the common mistake and it is self-defeating. A partner who makes an introduction, watches the deal take five months, and is told the window closed at 90 days does not register the next one. They have no way to make your sales cycle faster, so a window shorter than it transfers your own delay onto them.
The other half of this clause is what happens to a prospect whose window expires without closing. Silence means an argument later, so say it: the registration lapses and the prospect returns to the general pool, and a partner who is still actively working it can ask for an extension in writing.
Which fee trigger should you choose?
This decides the risk split far more than the rate does, and the three options are genuinely different deals.
On an accepted introduction. The partner is paid for the introduction itself. Generous, simple, and it pays out on prospects who never buy. It suits a partner whose value is access rather than selling, and it works where your conversion rate is high and predictable enough to price in.
On a meeting held. The middle position, and the one that needs a definition. "A meeting" has to mean a scheduled conversation the prospect attends, with someone who has a role in the decision, and a no-show has to be excluded explicitly or you will be arguing about calendar invites.
On a closed deal. The common default and the cheapest for you. The cost is that the partner waits through a sales cycle they do not control and cannot see. If you choose this one, the protection window and the reporting matter more, because they are all the partner has.
On the rate: in business software, 10% to 20% of first-year revenue is the common band, with flat fees of a few hundred to a few thousand dollars where deal sizes are predictable. The rate matters less than the trigger and the period combined. Ten percent of year one paid on close is a very different deal from 10% for twenty-four months paid on an accepted intro, and both get described as "ten percent".
What does "revenue actually collected" protect?
You, and it should be in the fee clause rather than assumed.
A fee calculated on invoiced revenue means you owe a partner money on a contract the customer never paid. A fee on revenue actually collected, net of refunds, credits and chargebacks, matches what you owe to what you received. It also needs the reverse mechanism: if revenue is refunded later, the related fee comes back, deducted from the next payment or repaid on request.
Expansion is the other half. If a referred customer upgrades inside the fee period, the fair reading is that the fee applies to the expanded revenue on the same terms, and saying so in writing avoids a conversation where both positions are reasonable. Renewals after the period carry nothing, which is also worth stating.
What is non-circumvention actually for?
Preventing a deal being structured around the fee, and nothing more.
The clause says that for a period after registration, neither party will route the transaction in a way that avoids the fee, for example by contracting with a subsidiary of the registered prospect instead. It is standard and it is reasonable.
What makes it fair is the exception. A prospect you were already engaged with before the registration has to be excluded, or the clause becomes a way for a partner to claim a pipeline that already existed. That exception is also why the rejection grounds matter: if you reject a registration on the basis that you were already talking to them, you have created the record that makes the exception work.
Twelve months is typical. Longer than the fee period is normal, since the point is to stop avoidance rather than to extend the fee.
What happens when the agreement ends?
Registrations that were accepted before termination keep their protection window, and fees on them keep accruing for their normal duration.
Without that sentence, either party can end the agreement to avoid a fee on a deal about to close, and a partner who thinks that is possible will behave accordingly. The exception people usually include, and which is reasonable, is termination for a breach of the conduct clause.
The conduct clause itself is short and worth having: the partner describes your offering accurately and only in the terms you publish, makes no commitment about price or scope, does not send bulk outreach in your name, and discloses the arrangement if a prospect asks. The last one is a minimum rather than a best practice. A partner whose recommendation turns out to have been paid for loses credibility with exactly the audience the referral depended on, and in some industries the disclosure is required outright.
When do you need more than a template?
One case specifically, and it is not a general caution.
Regulated introductions. In securities, insurance, real estate, lending and some healthcare contexts, paying someone for an introduction is licensed activity, and the fact that both parties agreed to it does not make it lawful. If your sector is on that list, that is the specific question to put to a lawyer before using any template, including this one.
Beyond that, the usual limits apply: the governing law clause assumes a US state, cross-border deals bring tax and data protection questions a template does not know about, and anything unusual in your arrangement needs a human reading it against what you actually agreed.
Generate the draft so both sides are reading the same document, read the registration clause properly because that is where the disputes live, and have it reviewed before signing.



