The 2026 Partnerships Playbook

The short version

A partner program fails quietly, one unanswered email at a time. Most teams sign three hundred partners and hear from twenty. The rest got a kickoff call and a slide deck, and then nobody had the hours to follow up. The five teams in this playbook run 300 to 1,400 partners each with two to five people, because agents now do the follow-up that used to fall on the floor.

This playbook is for the head of partnerships, or the COO who inherited a partner program that looks bigger on a slide than it is in the pipeline. It gives you the five jobs to hand to agents, in the order to hand them over, with the numbers each team saw. It also tells you what to keep for yourself.

1. 01Give every signed partner an onboarding agent that runs the first 30 days to a set finish line: certified, listed, and one referral sent.
2. 02Map shared accounts every week, not every quarter, and turn each overlap into one specific ask to one named person.
3. 03Let an agent match every referral to a closed deal and draft the payout batch. Approve each batch yourself.
4. 04Keep marketplace listings, private offers, and co-sell submissions current with an agent that reads the CRM.
5. 05Score partner health every week, and spend your own hours on the top 20 partners and the ones about to leave.

Why the math changed

Partnerships used to be a headcount problem, because each partner needed a person to remember it. A good partner manager could keep about 25 partners warm. Past that number, the program split into a top tier that got calls and a long tail that got a monthly newsletter. The long tail is where most programs lose money: signed partners who cost onboarding time and never send a deal.

Agents change the cost of attention. An agent can read a partner’s account overlap, their last six emails, their certification status, and their payout history in under a minute, then write the one message that partner needs this week. It does that for 400 partners before Monday’s standup. The person who used to send 25 careful emails now reads the 30 drafts that the agent marked as worth a human touch.

That does not make partnerships a software job. The partners who move revenue still want a person who knows their business, answers the phone, and fights for them in a deal review. What changes is the work around those relationships: the data entry, the reminders, the reconciliation, and the listing forms. By the contributors’ own time logs, that work was about 70% of a partner manager’s week.

Onboard to a finish line

Onboarding is the first job to hand over, because it has a clear finish line and many repeated steps. Most programs define onboarding as “kickoff call done.” The contributors define it as a partner who has finished certification, has a live entry in your partner directory, and has sent one qualified referral. An agent can chase every step of that. A kickoff call cannot.

1. 01Day 0: The signed agreement lands in the CRM. The agent creates the partner record, the portal account, and a shared Slack channel, and books the kickoff with the right partner manager.
2. 02Days 1–7: The agent sends the certification path for the partner’s type (agency, reseller, technology, or advisor) and checks progress every two days.
3. 03Days 7–14: The agent runs the first account map and drafts three named-account asks from the overlap.
4. 04Days 14–21: The agent writes the directory entry from the partner’s own website and sends it to the partner for approval.
5. 05Days 21–30: If no referral has arrived, the agent tells the partner manager which ask got the most interest and proposes a 20-minute call.

At Fieldnote, which sells scheduling software through 1,400 agencies and consultants, this sequence moved the median time to first referral from 64 days to 11. The share of new partners who sent any referral in their first quarter went from 18% to 47%. Priyanka Castellano’s team of three did not grow. They stopped writing onboarding emails by hand and started to join the calls the agent booked.

Map accounts, then ask

Account mapping is the core of co-selling, and most teams do it too slowly to matter. A quarterly spreadsheet swap gives you a list of overlaps that is three months old by the time a rep reads it. Crossbeam and tools like it make the overlap continuous. Agents make it useful, because they turn each overlap into an action that one person can take this week.

Kwame Oduya runs alliances at Tallyhouse, a spend-analytics company that sells to mid-market finance teams. Tallyhouse shares account data with 62 partners through Crossbeam. Each Monday, an agent reads the new overlaps, keeps the open opportunities over $25,000 in Salesforce, and checks where the partner has the stronger relationship: the account is the partner’s customer and Tallyhouse’s prospect. It then drafts one ask per overlap, addressed to the named partner rep, with the deal context the partner needs to say yes.

The numbers at Tallyhouse: about 140 new overlaps a week, 38 asks sent after the filter, and 9 introductions a week from partners. Partner-influenced pipeline grew from 14% to 31% of new pipeline in two quarters. The win rate on partner-introduced deals is 41%, against 22% for cold outbound. The agent never sends an ask about a partner’s top-10 account by itself. Those go to a person first.

• Ask about one account at a time. A list of 20 accounts gets no answer.
• Address the partner rep by name, not the partner company.
• Say what you give back: a referral, a joint case study, or a seat in the deal review.
• Log every ask and every answer in the CRM, so the next ask knows the history.
• Stop at two asks per partner rep per week.

Referrals and payouts

Referral payouts are where partner trust is won or lost, because a partner remembers a late or wrong payment for years. The work is reconciliation: match each referral to a lead, the lead to a closed deal, the deal to collected cash, and the cash to the commission terms in that partner’s contract. Done by hand, it takes a week each month and still gets disputed.

Oren Baptiste runs partner operations at Paysteady, a payroll compliance platform with 610 accountant partners. Each month, an agent pulls referrals from PartnerStack, deals from HubSpot, and payments from Stripe. It applies each partner’s terms: 20% of first-year revenue for most partners, 25% for the 40 partners in the top tier, and a 90-day clawback if the customer cancels. In August it matched 1,184 referrals, flagged 23 for review, and drafted a payout batch of $186,420.

Oren reviews the 23 flags and then approves the batch. The flags are the useful part: duplicate referrals from two partners for the same customer, deals that closed after the attribution window, and refunds that trigger a clawback. Before the agent, Paysteady had a payout error rate of 3.1% and about 14 disputes a month. Now the error rate is 0.4%, and disputes are down to two a month.

Cloud marketplace listings

Cloud marketplaces are now a sales channel, not a listing page, because buyers want to spend their committed cloud budget on your software. That turns AWS Marketplace, Azure Marketplace, and Google Cloud Marketplace into an operations problem: listings to keep current, a private offer to build for each deal, and co-sell referrals to submit to each cloud’s partner team with every field correct.

Lena Brandt leads marketplace sales at Quarrystone, which sells observability for data warehouses. Of Quarrystone’s new bookings, 41% now close through a marketplace. Lena’s team uses Tackle for listings and private offers. An agent watches Salesforce for deals that reach the proposal stage with a buyer who has a cloud commit. It drafts the private offer with the agreed price, term, and payment schedule, and submits the co-sell referral to that cloud’s partner portal. A person reviews each offer before it goes to the buyer.

Before the agent, a private offer took four days from request to the buyer’s inbox, and about one in five came back with a wrong field. Now it takes six hours. Quarrystone sent 212 private offers last quarter with one person on the marketplace team. The agent also compares each listing with the pricing page and the product docs every week, and opens a change request when they differ.

The tools, ranked

Every stack has the same four layers, even though the contributors use different products. You need account mapping, a partner program platform for referrals and payouts, a marketplace tool if you sell through the clouds, and the CRM that holds the deals. Agents work best when each layer has a good API and a clear record of who changed what, so that is how we scored them.

Two notes on the ranking. First, no contributor runs more than 100 partners from a CRM alone, because the referral and payout data gets messy fast. Second, the Salesforce and HubSpot scores say more about the partner setup than about the CRM. Both work well when one person keeps the partner fields clean. The companies below lead their categories.

• CrossbeamAccount mapping and co-selling software for partner teamsEcosystem-led growth and partner account mapping
• PartnerStackPartner program management and payouts, part of AppDirectAI-native PRM for every partner type
• TackleCloud marketplace listing and co-sell software, part of AppDirectThe platform for cloud go-to-market

Health, and what stays human

Partner health is the last job to hand over, and it needs the most judgment. An agent can score every partner each week on signals: referrals in the last 90 days, certification status, response time, overlap growth, and payout disputes. The score tells you where to look. It does not tell you why a partner went quiet, and the reason is usually a person: a champion who left, a reorg, or a competitor’s better offer.

At Corvane, Adaeze Holmgren gives every partner a weekly score from 0 to 100. A partner who drops more than 15 points in a month goes on her Friday list. Of the 340 partners in Corvane’s program, the list holds 6 to 12 names a week, and she calls each one. In the past year, those calls found 4 champions who had left, 2 partners who had signed with a competitor, and 1 partner that wanted to double its commitment for a bigger discount.

Every contributor drew the same line between agent work and human work. The agent prepares, reminds, reconciles, and drafts. A person decides anything that changes money, terms, or trust. These are the items that all five teams keep with a person, with no exceptions:

• The first meeting with a new partner’s leadership.
• Contract terms, tier changes, and commission rates.
• Joint roadmaps, and anything a partner tells you in confidence.
• A deal review where a partner and your rep disagree about who sourced the deal.
• Any message about a partner’s top-10 account.
• The call when a partner is about to leave.

So what do you do?

Start with the job that has the clearest finish line and the most repetition, which for most teams is onboarding or payouts. Do not start with co-selling. Co-sell asks work only when your partner data is clean, and the first two jobs are what make it clean. Give each job one quarter, and measure it against the one number it should move.

1. 01This week: List every signed partner and mark each one active, onboarding, or dormant. Be honest. Most programs find that more than half are dormant.
2. 02Month 1: Write your onboarding finish line in one sentence, and give an agent the 30-day sequence to reach it. Measure the days to first referral.
3. 03Month 2: Connect referrals, deals, and payments, and let an agent draft the payout batch. Approve every batch yourself, and track the error rate.
4. 04Month 3: Start weekly account mapping with your top 20 partners. Cap asks at two per partner rep per week.
5. 05If you sell through a cloud, give private offers and co-sell submissions to an agent next. Keep a person on each offer review.
6. 06Every Friday: Read the health list and call each name on it.

The goal is not a smaller partner team. Every contributor kept their headcount and changed what their people do. The partner managers stopped being the memory of the program and became the people that partners call when a deal matters. That is the job they were hired for, and now they have the hours to do it.