Prompt pack

Partner ops prompt pack

Twelve copy-paste prompts for the unglamorous jobs of partner ops. Paste into any AI assistant, fill in the angle brackets, get back a draft instead of a blank page.

skip the prompts entirelyintrozy MCP

Every prompt below makes you the integration: export the CSV, paste the thread, carry the answer back. Introzy ships an MCP server that removes the copy-paste step. Connect it and your assistant reads your pipeline, contacts, and referral history directly.

1.Connect your assistant to Introzy: Claude, ChatGPT, or any MCP client
2.Ask for what you need in plain English: "triage my stalled deals" or "draft a status update for Dana"
3.The assistant reads live data, applies the same logic as these prompts, and writes the result back

The recurring jobs (status updates, stale-deal triage, statements) are what Kai runs nightly without being prompted at all. Does your PRM do that?

Start free

// Keep partners in the loop

The most common way a referral program dies is silence. A partner lends you their credibility, the referral disappears into your sales process, and the partner is left wondering whether anyone followed up. These three prompts cover the three moments that matter: the stage change, the loss, and the payout. The reasoning behind the cadence is in the blog post, and the non-AI version is in the templates.

1Partner status update
When a referred deal changes stage
You write short, warm partner updates for a business that takes referrals seriously. Facts: partner [name], prospect [name], introduced on [date], current stage [stage], since the last update we [what happened], next step [step and expected date]. Write a four-sentence status email to the partner: what happened, what happens next, no ask, and a thank-you that does not gush. No exclamation marks.
2The lost-deal email partners remember
When a referred deal dies
A referred deal did not close. Partner: [name]. Prospect: [name]. Reason: [honest reason]. Our best-fit referrals look like: [one-line ideal customer description]. Write a short email that tells the partner directly, explains the reason in plain language, names what converts well for us so their next referral is easier, and makes clear this changes nothing about how much we value their introductions.
3Commission statement explainer
When you send a payment
Turn these numbers into a plain-language referral statement. Terms: [fee rate and basis, e.g. 10% of collected revenue]. Period: [period]. Data: [paste referrals, invoices collected, amounts]. Output: one paragraph a partner can read in thirty seconds stating what was collected, what was earned, and when it pays out, followed by a line-item list showing the math for each referral. If any number cannot be derived from the data, say which one instead of guessing.
Why transparency compoundsStatus update templates

// Turn mess into records

Referrals arrive as email threads, voice notes, hallway conversations, and LinkedIn DMs. None of those are structured data. These three prompts turn messy inputs into records you can actually track: a structured referral from a thread, a log entry from raw notes, and a forwardable blurb for double opt-in intros.

4Thread to structured referral
When a referral arrives as an email thread
Extract the referral from this email thread. Return: who made the introduction, who needs something (the seeker), who can supply it (the provider), the prospect contact with email, what they need in one sentence, urgency, and any mentioned budget or timeline. Then list the required fields that are missing and write the one short reply that would collect all of them in a single question. Thread: [paste thread].
5Notes to referral log
After a networking event or call
These are raw notes from a call or event. Pull out every introduction that was made or promised. For each: who introduced whom, to whom, what for, and the follow-up owed with a suggested deadline. Format as rows I can paste into a tracker: introducer, party A, party B, context, next action, due date. Notes: [paste notes].
6The forwardable blurb
When someone asks you for an intro
Someone asked me to introduce them to [person or company]. Here is their ask: [paste request]. Write the double opt-in forwardable: three sentences in my voice explaining who they are, the specific reason the intro makes sense, and a no-pressure out. The recipient should be able to say no in one word without awkwardness.
How to track referrals in your CRMCRM referral field kitReferral tracking spreadsheet

// Deal health and pipeline hygiene

A stale referral is worse than a stale cold lead because someone else's reputation is attached to it. These two prompts apply a published triage formula to your pipeline export: which deals are cold, which are stalled, and which five need you this week. Run the formula on one deal interactively with the health check tool, or read the logic behind the thresholds.

7Stale-deal triage
Weekly, against a pipeline export
You are triaging referral deals. Rules: a deal with an upcoming meeting or an active reminder is fine, leave it alone. A deal created 90 or more days ago with no logged activity in 30 days is cold: recommend closing it as lost, kindly. A deal with email history that has gone silent past its stage threshold (7 days late-stage, 10 mid, 14 early) is stalled: recommend a follow-up and draft the first line. Everything else is healthy. Apply the rules to these rows and return a table: deal, verdict, evidence, recommended action. Rows: [paste pipeline export].
8The five-deal briefing
Monday morning
From this pipeline export, pick the five deals that most need attention this week and say why in one line each: biggest value at risk, longest silence, closest to a decision, or owed a partner update. End with the single referral partner who should hear from us this week and what we would tell them. Rows: [paste pipeline export].
Referral health checkAI referral routing with human approval

// Attribution and reporting

Attribution disputes are the relationship tax of every referral program. These two prompts build the quarterly report (sourced vs influenced, no double-counting) and untangle the two-partner overlap. Both enforce a written policy. If you do not have one yet, the builder generates it from seven questions.

9Sourced vs influenced report
Quarter end, from a CSV
Build a partner attribution report from this data. Definitions: sourced means the deal originated from a documented partner introduction and only one partner can hold it; influenced means a partner has a logged activity on the deal but did not originate it, and several partners can hold it. Return: revenue by partner split into sourced and influenced columns, deals claimed by more than one partner flagged for review, and totals that finance can sum without double-counting. Data: [paste CSV].
10Attribution dispute untangler
When two partners claim one deal
Two partners claim credit for the same deal. Policy: the first documented introduction wins sourced credit; a later partner with logged, linked activity may earn influenced credit; verbal claims without a record earn nothing. Timeline and evidence: [paste both claims with dates]. Apply the policy, state the ruling with the evidence that decides it, and draft a short, respectful note to each partner explaining the outcome and what a winning record looks like next time.
Sourced vs influenced revenueAttribution policy builder

// Program admin

Two jobs that are easy to skip and expensive to redo: putting fee terms in writing before the first dispute, and reaching out to the next partner before the pipeline thins. The agreement prompt drafts a clause, not a contract; have counsel finish it. The recruitment prompt writes the three-sentence note that works better than flattery.

11Agreement clause drafter
Before the handshake becomes a dispute
Draft one plain-English contract clause for a referral agreement covering: [the term, e.g. 10% of collected revenue for 12 months, paid monthly, 90-day attribution window, clawback on refunds]. One paragraph, no legalese where plain words work, define any term a partner could reasonably misread, and end with a bracketed note listing what our lawyer should confirm for our jurisdiction.
12Partner recruitment note
When recruiting a new partner
Write a three-sentence outreach note inviting [person] to refer clients to us. Our ideal customer: [one line]. What partners get: [fee and terms]. Name the fee plainly, state the effort required (an email, not a sales job), and end with a question that is easy to answer yes to. No flattery padding.
Referral fee agreement templateReferral program starter kitWhat is a fair referral fee?
Related tools

The artifacts behind the prompts

Attribution policy builder

"What do prompts 9 and 10 enforce?"

Answer seven questions and leave with the written policy the prompts apply.

Referral health check

"Where do the triage rules come from?"

Run the stale-deal formula from prompt 7 against one real deal.

Status update templates

"Prefer templates to prompts?"

Five copy-ready partner emails, no assistant required.

Questions

Fair questions

Any of them. The prompts are plain text and work in ChatGPT, Claude, Gemini, Copilot, or whatever your company already uses. Paste the prompt, replace the angle-bracket placeholders with your data, and review the output before sending.

Assistant output is a draft, not a decision. Review anything customer-facing or contractual before it ships, and have counsel review agreement language. Prompts 3 and 11 assume referral compensation is permitted in your industry and jurisdiction.