Referral Programmes for B2B: Build or Buy?

Referral and affiliate programmes work in B2B for a boring reason: a recommendation from someone the buyer already trusts beats cold outreach. But the moment you promise to pay a commission, you have taken on an attribution problem, a payments problem and a tax problem at once. Founders and ops leads usually meet this at the same moment — ten partners, a shared spreadsheet, and a monthly argument about who introduced which lead. The real question is whether to build the tracking and payout layer yourself or buy something off the shelf.

What does a B2B referral programme actually need to track?

A B2B referral programme needs three things tracked reliably: who introduced the lead, what eventually happened to that lead, and what money is owed as a result.

That is not the same as click tracking. A B2B deal can take 60 to 180 days from first call to first payment and involve three people at the buyer, so attribution has to survive that gap. In practice that means deal registration — the partner registers a named lead with a timestamp, and the rule for what happens when two partners register the same lead is written down and enforced in code. The payout ledger is the second half: commission tied to collected revenue rather than booked revenue, tax fields on every payout, and an audit trail your CA will accept in March.

When is building in-house the cheaper option?

Building is cheaper when your commission rules are unusual and your active partner count is small — roughly under 50.

Take a concrete case. A team selling ₹40,000-a-year licences at 20% commission, running 25 active partners, typically needs two developers for three to four weeks: somewhere between ₹1.5 lakh and ₹3 lakh at current Indian rates, plus a few hours a month of upkeep. That is a real number, not a rounding error. But the same build often pays for itself within a year, because no off-the-shelf tool models 'commission released only after the client invoice is paid'. If your payouts depend on collections, part-payments or per-deal overrides, custom is usually the honest answer; /services covers how that kind of scoping works.

What does buying actually cost?

Off-the-shelf affiliate software typically costs $50-500 a month, roughly ₹4,000-40,000, and the jump to a genuinely partner-oriented platform is steeper than that.

The cheap end is cheap for a reason and installs in an afternoon, but most of those products are built for B2C checkout: a cookie, a link, a sale. PartnerStack is one of the few that is B2B-native, with deal registration and lead routing, and it is priced for teams that already have a partnerships function. FirstPromoter, Tapfiliate and Rewardful assume Stripe-style subscription billing at the point of conversion — workable if you sell self-serve, awkward if you sell through a two-month pipeline. All of them bill in USD on a card, which means no GST invoice in INR and an unhappy conversation at year end.

What breaks specifically for Indian B2B companies?

Two things break most often: payouts and partner behaviour.

Commission attracts TDS under Section 194H (2% at the time of writing), and GST applies to the commission invoice when the affiliate is registered. Every monthly payout therefore needs a computation, a tax entry and a remittance record — not just a UPI transfer and a WhatsApp thanks. Meanwhile your best partners are often consultants who live in WhatsApp and will not log into a portal to register a lead.

A minimum viable build has to include:

• deal registration with a unique ID and a timestamp
• a written attribution rule (first registered, with an expiry) enforced in code
• a payout ledger driven by collected revenue, not booked revenue
• TDS and GST fields per payout, with a downloadable statement for the partner
• a partner view that works in a phone browser, because that is where they will open it

Should you buy first and build later?

Buying first is usually right if you are still finding out whether the programme works at all.

Run a pilot for 6-10 weeks with 5-10 partners and a cheap tool, or even a form and a spreadsheet, and watch where it hurts. The signals that you have outgrown it are blunt: payouts cross roughly ₹1-2 lakh a month, more than half need a manual correction, or finance rebuilds the same reconciliation every cycle. At that point the rules are known and a build becomes a specification exercise rather than a guess. If you go custom, a fixed-price scope is only possible once those rules exist — /pricing explains how we structure that. Either way, budget the internal hours; a per-seat quote hides them.

Where to start

Write two documents before you evaluate a single vendor: the attribution rule and the payout policy. One page each — what counts as a valid referral, when it expires, what triggers commission, how TDS and GST are handled, and who approves an exception. That page becomes the brief for either route, and it is the first thing we ask for at Maysan Labs when a team wants help building the tracking layer.