In-House Affiliate Program

An in-house affiliate program means the company owns the whole operation. It licenses or builds the tracking software, recruits its own partners, sets its own terms, and pays affiliates directly. Nothing sits between the brand and its affiliates except the software that records clicks and conversions.

The alternative is running on an affiliate network, where a third party supplies the tracking, an existing pool of affiliates, and payment processing in exchange for fees and reduced control. For B2B SaaS, where relationships run long, commissions recur monthly, and partner quality beats partner quantity, the in-house model has become the default choice for many programs.

How it works in B2B SaaS

The company selects affiliate tracking software and integrates it with its billing or subscription platform, so recurring commissions calculate automatically and stop when a referred customer churns. Affiliates sign up on the brand's own page, receive unique links and a dashboard, and deal with the brand for everything: approval, onboarding, creatives, questions, and payouts.

Recruitment is entirely the brand's job. There is no built-in pool of partners, so the team sources affiliates through customer outreach, content creator prospecting, communities, and directories. That is the real cost of going in-house, and also the reason in-house programs tend to end up with fewer but more engaged partners.

A worked example

CloudMetric, an analytics SaaS at $50,000 MRR, launches an in-house program. The tracking software costs $150 per month, and the program offers a 25 percent recurring commission for the first year of each referred subscription. In year one the team recruits 120 affiliates, of whom 30 become active and drive 40 paying customers averaging $100 MRR each.

That is $4,000 in new MRR, against roughly $1,000 per month in commissions plus the $150 software fee. On a network, the same result would typically cost the commission plus an override fee to the network on every payout, often with setup costs and monthly minimums on top. The tradeoff ran the other way on effort: CloudMetric spent real hours on outreach that a network's marketplace might have shortcut. At its scale and margins, owning the partner relationships and the economics was worth it.

Typical ranges and benchmarks

A few numbers recur across SaaS programs:

• Recurring SaaS affiliate commissions commonly run 20 to 30 percent of referred revenue
• Commissions are commonly paid for the first 12 months or for the lifetime of the customer
• Cookie windows commonly range from 30 to 90 days
• In-house tracking software is commonly priced as a flat monthly subscription
• Networks commonly charge a percentage override on each commission paid, often in the range of 20 to 30 percent of the commission, plus platform fees

One benchmark holds in both models: only a minority of signed-up affiliates typically ever drive a first referral. Activation, not signup volume, is where programs are won.

In-house affiliate program vs affiliate network

The confusion people actually have is thinking in-house means building your own tracking from scratch. It does not. In-house describes who owns and runs the program, not who wrote the code; most in-house programs run on licensed third-party tracking software.

On a network, the intermediary owns the affiliate relationships, aggregates many programs in one marketplace, and processes payments. That buys discovery and convenience at the cost of fees, less control over terms, and a layer between you and your partners. In-house gives you direct relationships, your own data, and lower per-sale cost, but recruitment and operations are yours. Many programs start on a network for reach and move in-house as they mature.

How it shows up in affiliate and partner programs

For affiliates, in-house programs mean a separate login per brand, which is mildly annoying, but often better rates and direct access to the affiliate manager, since brands frequently pass network savings into commissions. Professional affiliates routinely join both networks and in-house programs, and judge each program on its terms and tracking reliability rather than its model.

For SaaS companies, an in-house program often becomes the umbrella for broader partner motions. Referral partners, agencies, and resellers can run on the same tracking with different commission rules, which keeps attribution and payouts in one system as the partner program grows.

Common mistakes

The classic failure is underestimating recruitment. Without a network pool, an in-house program with no outreach plan is an empty room. The second is choosing tracking software that does not integrate with billing, which breaks recurring commissions and forces manual reconciliation every month.

Three other errors show up repeatedly:

• Treating in-house as set-and-forget when you now own support, disputes, and payout operations
• Skipping a proper affiliate agreement and disclosure requirements (have counsel review yours; this is not legal advice)
• Paying affiliates ad hoc instead of committing to a clear payout schedule and threshold, which quietly erodes partner trust

Frequently asked questions

The questions founders weigh most when choosing the in-house route.

Is an in-house program cheaper than a network?

At meaningful volume, commonly yes: a flat software fee beats a percentage override on every commission as payouts grow. At very low volume the difference is small, and a network's built-in affiliate pool may be worth its fees. Compare total cost per referred customer, not just platform pricing.

Do I need to build my own tracking software?

No. In-house refers to who runs the program, and nearly all in-house programs license affiliate tracking software rather than building it. Building your own only makes sense with unusual attribution needs and engineering capacity to maintain it indefinitely.

Can I run an in-house program and a network at the same time?

Yes, and some brands do it for reach. The critical requirement is deduplication: a clear attribution rule so a single sale never pays out in both systems. If you cannot enforce that, run one model at a time.