The partner scorecard: how to rate a live technology partner

A live technology partner is easy to grade on vibe. The Slack channel is friendly. The last call felt good. Someone called the partnership "strategic." None of that tells you whether to give them another build slot, keep them on maintenance, or sunset. Vibe is how weak partners stay on the list and how strong partners fail to get more investment. A scorecard is how you make that call in a QBR without a two-hour argument.

This is a simple 1-5 card for a partner you already work with: not a selection score for a new logo (that is the partner ICP), but a health read on a live one. Four dimensions, scored with evidence, reviewed on a quarter. You take it into the partner QBR. You do not build a data science project. You build a page both sides can sit in front of without embarrassment. Pair it with partnership metrics for the numbers underneath, and with partner pipeline in the CRM so pipeline scores are not a memory test.

The 60-second version

If you only read one section, read this one:

• Score live partners on four dimensions: product health, adoption, pipeline, and relationship (working cadence).
• Use 1-5 with written anchors. If 4 and 5 cannot be distinguished in a sentence, you do not have a scale.
• Evidence on the card, not in someone's head. A ticket count, an active-account number, tagged pipeline, a last-QBR date.
• The card has a job: grow, maintain, or stop. A score without a decision is a decoration.
• Share a version with the partner. You can leave internal notes off. You should not have a secret grade and a public smile.
• Re-score every quarter in the QBR, not when you are angry.
• Do not average away a 1. A dead integration with a "great relationship" is still a 1 on health.

Why a scorecard beats a vibe

You will have more live partners than you can remember in detail. Even a handful is enough to confuse: the partner you like, the partner who used to ship, the partner a board member likes, the partner whose integration is quietly erroring. Without a card, the next slot goes to whoever asked last.

A scorecard also changes the QBR. Instead of a recap, you walk four numbers and agree what would move each one next quarter. The QBR guide already asks for a short scoreboard. This card is that scoreboard, plus a relationship dimension people pretend is unmeasurable.

It is not a performance review of a person. It is a rating of the motion between two companies. Use it on live partners. Do not use it to reject inbound (use the ICP). Do not use it on a partner who has not had time to ship. Scoring a two-week-old partner a 2 on adoption is noise.

Partner program ROI is the program roll-up. The scorecard is the per-partner input. A program can look fine in aggregate while two partners are carrying it and three are dead.

The four dimensions, on a 1-5 scale

Keep four. More dimensions and nobody can remember the scale. Fewer and you will hide a failure in an average.

Health. Is the connector up, supported, and within error bounds you would accept for your own feature? Count open incidents, time-to-fix, partner API changes you were not told about, and whether support can troubleshoot without a hero. A 5 is boring: it works. A 1 is a known broken path you are still listing.

Adoption. Active use among shared customers, not installs. Direction matters. A 5 is the integration showing up in CS notes as part of the workflow. A 1 is a launch with nobody on it.

Pipeline. Tagged sourced and influenced opportunities in the period, plus any closed-won you can defend. A 5 is a partner who regularly originates or materially helps deals. A 1 is a co-sell story with an empty CRM. If your motion with this partner is integration-only, say so and score pipeline as not-applicable, or score the expansion CS can see. Do not fake a pipeline 4 because the relationship is warm.

Relationship. Named owner, answers in a useful time, attends the QBR, tells you about API changes, enables their field if that was the deal. A 5 is a boring cadence. A 1 is a ghost, or a partner who only appears when they want a press hit.

Weight them by motion. For an integration-led partner, health and adoption should outweigh pipeline. For a co-sell-led partner with a stable connector, pipeline and relationship can outweigh. Write the weights once per partner type, not per meeting.

2 and 4 are the in-between. Force a sentence: "4 on adoption because active accounts grew, but still concentrated in two logos." If you cannot write the sentence, you are guessing.

One page per partner. Header: name, owner on each side, motion, weights. Four rows: dimension, score, evidence, note. Footer: composite, last quarter's composite, decision (grow / maintain / stop), the one bet for next quarter, date of next review.

Scoring rules: score with a second person (PM or CS for health and adoption, sales for pipeline). A 1 on health cannot be hidden by a 5 on relationship in the decision. Missing evidence is a 2, not a 3. Do not score people. "Their PM is difficult" goes in the note.

Show the partner a version of this card. Redact internal politics if you must. Do not redact health or adoption.

Using it in a QBR, and what to do with a low score

Send the card before the meeting. The QBR is for the decision and the bet, not for discovering the scores. Walk the lows first. A mixed card needs one or two moves, not eight.

Grow. Health and adoption are 4-5, pipeline is working or you have a reason it should, relationship can hold more load. Actions: another slot, a field play, a joint customer program. Growing a 2 on health is how you multiply incidents.

Maintain. Useful, not a bet to enlarge. Keep the cadence and maintenance. Most live partners should live here. If every partner is "grow," you do not have a scorecard. You have enthusiasm.

Stop. Health 1 with no path, adoption 1 after a fair window, or relationship 1 after you asked for an owner. Stopping can mean sunset, downgrade from co-sell to listing-only, or a hard pause. Write the customer impact with CS before you tell the partner. See the end stages in the SaaS partnership lifecycle.

Do not ambush. If you are going to score them a 1, they should have heard the incidents in the weeks before. Internally, roll the cards into the program view on a partnerships dashboard: how many partners in grow / maintain / stop, and whether your slots match the grow list.

A low score is a project, a conversation, or an exit.

Health 1-2. Incident path, named engineer on both sides, a date for "stable or we unlist." Do not sell the integration in the meantime without a warning to CS and sales.

Adoption 1-2. Joint push to existing shared customers, or an honest look at whether the workflow was real. If nobody wants the job the integration does, more enablement will not save it.

Pipeline 1-2 (when co-sell was the deal). Enablement, account mapping, a single AE pair. If their field cannot attach you after a quarter of trying, believe them.

Relationship 1-2. Ask for an owner. If they cannot name one, you do not have a partner.

Put the fix on the card as next quarter's bet. If the same dimension is a 1 twice, the third QBR is a stop conversation.

Common mistakes, and the fix

Scoring on liking them. The fix: evidence fields. No evidence, cap at 2.

Too many dimensions. The fix: four. Fold enablement into relationship or pipeline if you need it.

A composite that hides a 1. The fix: any health 1 blocks "grow."

Secret scores. The fix: share the operational version. Disagreement is useful. Surprise is not.

Scoring new partners as live. The fix: wait until there is something to adopt. Use onboarding until then.

No decision on the footer. The fix: grow, maintain, or stop, every time. "Continue to monitor" is maintain.

Weaponizing the card after one bad week. The fix: quarterly, plus incidents as evidence, not as a re-grade in Slack.

FAQ

Should every partner get the same weights? Same dimensions, weights by motion. Integration-led and co-sell-led are not the same job. Do not invent a new card per logo.

Do we show the partner the composite? Show the dimensions and the decision. The composite is a convenience for your internal queue. The conversation is about the lows.

What is a fair window before scoring adoption? Long enough to launch and put a few accounts on it. If you are still in private beta, you are not in scorecard mode.

Can we do this in a spreadsheet? Yes. You should, until the book is large. A PRM score that nobody believes is worse than a sheet you update in the QBR.

Who scores? The partner manager drafts, a PM or CS confirms health and adoption, sales confirms pipeline tags. The exec sponsor sees the grow/stop list, not every 3.

What if the partner wants to score us too? Good. Swap cards. A two-way score is how you find that your enablement is their 2.

When do we sunset versus maintain at a 2? If customers depend on it, maintain and fix, or maintain and plan a replacement. If nobody depends on it and the card is a 2 after two quarters, stop. CS gets a vote.

Further reading

• How to run a partner QBR for the meeting this card is built to run.
• Partnership metrics for the outcome numbers behind the scores.
• Partner ICP for scoring candidates before they are live.
• RACI matrix on Wikipedia, if you need to name who is Accountable for grow / stop once the card is filled.

The short version

Rate live technology partners on a 1-5 card with four dimensions: health, adoption, pipeline, and relationship. Write anchors. Attach evidence. Re-score every quarter in the QBR. Use the footer to grow, maintain, or stop, and do not grow a 1 on health.

Share the operational version with the partner. Keep the card small enough to update. Four numbers and a decision will change the book. The rest is decoration.

If you want a first pass at scoring the partners you already have, that is what a Partner Audit is for. We look at the live work, the evidence you can actually pull, and which relationships deserve a slot next quarter.