SaaS distribution strategy

Two companies sell the same product at the same $14,000 a year price. One takes card payments on a pricing page. The other gets bought through AWS Marketplace, where the invoice draws down cloud spend the customer committed to eighteen months ago, so nobody has to open a new vendor request. Identical software, wildly different cost to reach the buyer.

That gap is distribution. Most SaaS teams never decide it on purpose, which is how a company ends up with a Salesforce AppExchange listing nobody maintains and a reseller agreement signed by a founder in 2024 that has produced four deals since.

Distribution is the route to the buyer, promotion is the noise you make on it

Distribution answers a different question from marketing. Marketing asks how a buyer learns you exist. Distribution asks how the software and the money actually change hands, who sits in the middle, and what they charge for standing there.

Six routes cover almost every B2B software company: direct self serve, direct sales, app marketplaces, cloud marketplaces, resellers and managed service providers, and embedded or OEM. You will recognise all six. What almost nobody does is compare them on the same three numbers before committing: the take rate, the months to first revenue, and who ends up owning the customer relationship.

That third one gets ignored and it is the expensive one. A reseller deal that costs 30 points of margin is survivable. A reseller deal that costs 30 points and hides your product usage data from you means you cannot run expansion, cannot see churn coming, and cannot build SaaS marketing programs against your own install base.

The decision also has a natural cadence. Revisit it once a year, or whenever your average contract value roughly doubles, because ACV is what makes a route affordable or absurd. At $600 a year, a field sales rep is a fantasy. At $90,000, a self serve checkout is leaving procurement confused.

Direct self serve: the best margin and the hardest demand problem

Self serve keeps almost everything. Stripe takes roughly 2.9 percent plus 30 cents on card volume, so you retain about 97 cents on the dollar. No rev share, no partner manager, no quota carrier.

The catch is that self serve does not create demand, it only collects it. Companies that call self serve their distribution strategy usually mean they have no distribution strategy and are hoping SaaS content marketing fills the top of the funnel. Sometimes that works beautifully. Linear grew a design-led developer tool largely through peer recommendation and a very tight product surface. Calendly spread because every meeting invite carried the brand in front of a new person.

Where self serve genuinely fits: the buyer is the user, the ACV sits under roughly $5,000, the product delivers a visible result inside a single session, and the purchase does not trigger a security review. Break any one of those and the route strains. Break two and a sales team gets staffed, planned or not.

The honest failure mode is trial to paid conversion. A self serve funnel with a 2 percent visitor to signup rate and a 12 percent trial to paid rate needs enormous traffic volume to matter, and building that volume is a two year project with real cost. Self serve moves your acquisition cost from commission into content and product engineering. It does not remove it.

Direct sales: expensive, slow, and the only route some buyers will accept

A fully loaded US account executive costs $180,000 to $260,000 a year once you add benefits, tooling, travel and the sales engineer time they consume. Standard practice is to expect three to five times OTE in closed ARR from a productive rep, which sets a hard floor on the ACV that justifies the seat.

Below roughly $15,000 ACV, a full cycle rep rarely pays back inside a sensible window. Between $15,000 and $50,000 you want inbound-fed reps with short cycles. Above that you are into buying committees, security questionnaires, procurement portals and a nine month cycle, which is the territory the enterprise SaaS marketing playbook is built for.

Direct sales earns its cost in one specific situation: the purchase requires somebody to absorb organisational friction on the customer’s behalf. Legal redlines, a SOC 2 request from a bank’s third party risk team, an integration question only a solutions engineer can answer. No marketplace listing removes that work. It just changes who does it.

App marketplaces: sitting next to the system of record

App marketplaces put your product inside the platform your buyer already lives in. The economics vary more than people expect, and the difference between the Shopify App Store and Salesforce AppExchange is not a detail.

Marketplace	Revenue share	Time to listed	What actually drives installs
Shopify App Store	0% on first $1M a year, 15% above	2 to 6 weeks	Install volume, review count, category rank
Salesforce AppExchange	Commonly around 15% of net revenue, plus annual partner fees	3 to 6 months with security review	AE referrals and co-sell, not browsing
HubSpot App Marketplace	No revenue share on standard listings	3 to 8 weeks, longer for certified status	Certification badge and integration depth
Slack App Directory	No revenue share	2 to 6 weeks	In-workspace discovery and workflow fit

Klaviyo is the clearest worked example of marketplace distribution done as a real strategy. Its 2023 S-1 disclosed that customers who also use Shopify accounted for roughly three quarters of annual recurring revenue. That is not a listing. That is a company that built its integration deeper than anyone else’s, bought placement in the merchant’s daily workflow, and accepted enormous platform concentration as the price. Shopify later took an equity position. The dependency runs both ways now, which is the best outcome available in a marketplace relationship and also a reminder of what you are signing up for.

Here is the opinion the category avoids stating. A marketplace listing is a distribution decision that companies treat as a listing task, which is exactly why most of them earn nothing. Somebody in engineering fills in the manifest, uploads four screenshots, ships it, and moves on. Six months later the app has eleven installs and two reviews, both complaints.

Ranking inside these stores follows installs, review velocity, recency of updates and support response time. All four are marketing work. Budget a launch the way you would budget a product launch: a customer campaign to drive the first 100 installs, an email sequence asking happy users for reviews, a named owner who answers support inside a day, and a quarterly update so the listing does not go stale.

Vertical platforms deserve the same treatment and get even less of it. Epic App Orchard, Procore, Toast and the big practice management systems each hold a captive buyer base, and a vertical SaaS marketing playbook that ignores the incumbent platform’s marketplace is ignoring the shortest path to the buyer in the market.

Cloud marketplaces and co-sell: the committed spend accelerant

This is the route most underused by companies who would benefit from it. AWS Marketplace, Azure Marketplace and Google Cloud Marketplace sell third party software through the customer’s existing cloud account, and the listing fee is low: commonly around three percent on public offers, and lower still on large private offers.

The fee is not why you do it. You do it because of committed spend drawdown.

Large enterprises sign multi year commitments with their cloud provider: an Enterprise Discount Program agreement at AWS, a Microsoft Azure Consumption Commitment. Eligible marketplace purchases count against that commitment. So when a VP of engineering buys your $200,000 observability tool through AWS Marketplace, they are not asking finance for $200,000 of new budget. They are spending money the company already promised to spend, against a contract that is already signed, through a vendor that is already approved.

Watch what that does to a sales cycle. The security review shrinks because the cloud provider’s standard contract terms carry a lot of the load. Procurement has no new vendor to onboard. The champion stops needing a business case and starts needing an approval. Deals that took four months close in six weeks.

Datadog is the obvious example. It transacts across all three major cloud marketplaces, and for its largest customers the marketplace private offer is simply how the renewal gets done. The Datadog marketing strategy teardown covers the demand side of that business, but the distribution side matters just as much: the company made itself easy to buy inside the exact purchasing system its buyers already use.

Costs to get transactable: four to eight weeks of engineering for metering and entitlement, legal review of the provider’s standard EULA, and finance work to handle the disbursement schedule, which typically pays out 45 to 60 days after collection. That cash flow delay surprises smaller companies.

Resellers, MSPs and agencies: distribution you do not control

A reseller buys at a discount and sells at list, keeping the spread. Twenty to forty points is the normal band, with the top of the range going to partners who also deliver implementation, first line support and local invoicing.

The margin is the visible cost. The invisible one is the relationship. Your reseller owns the renewal conversation, the billing contact and often the admin account, which means your product analytics show usage without names and your lifecycle email programs have nobody to email. Expansion revenue, the thing that actually compounds in SaaS, becomes someone else’s job.

Three situations make that trade worth it. Geographies you cannot staff, where a local partner handles language, invoicing and the relationship norms you would spend two years learning. Regulated or public sector buyers who can only purchase from an approved vendor list. And high volume SMB markets where managed service providers already sit inside hundreds of small companies and bundle your product into a monthly bill nobody reviews.

Run it properly or not at all. That means deal registration so partners do not compete with your own reps, published margin tiers, a partner portal with current collateral, and a named partner manager. Channel conflict is not an edge case, it is the default state, and it is best resolved with a written rule rather than a quarterly argument. Partner-sourced demand also needs its own measurement, separate from the SaaS demand generation reporting you run for direct.

Embedded and OEM: your engine inside someone else’s product

Embedded distribution means your software runs inside another company’s product and their customers never see you. Payments, messaging, identity, analytics, e-signature and data enrichment all have large embedded businesses. Twilio built much of its early volume this way, sitting under products whose users had no idea a third party handled the SMS.

Deal shapes vary. Revenue share of ten to thirty percent of the end price is common, as is a flat platform fee with usage tiers on top. Contracts run two to five years, which makes the revenue unusually predictable for SaaS.

Two real costs. First, the engineering: an embedded partner needs white label theming, tenant isolation, an admin API and a support escalation path, and that work does not benefit your direct product. Second, concentration. When one partner reaches a third of revenue, their roadmap becomes your roadmap and their renewal becomes your board meeting.

I would only build embedded deliberately after the direct business is working, with one exception: infrastructure products whose natural buyer is a developer at another software company. For those, embedded is the primary route and direct is the side business.

What each route costs and when it starts paying

Route	Take rate or fee	Time to first revenue	Typical share of new ARR	Pays off from
Direct self serve	About 3% payment processing	Immediate once traffic exists	40% to 100% at low ACV	Day one, if demand exists
Direct sales	25% to 40% of ACV fully loaded	6 to 12 months per rep	50% to 90% above $25K ACV	$15K ACV and up
App marketplace	0% to 15% plus partner fees	2 to 6 months	5% to 40%, occasionally more	When the platform holds your buyer
Cloud marketplace	Around 3% listing fee	3 to 6 months to transactable	10% to 50% of enterprise ARR	$50K deals into cloud-heavy buyers
Reseller or MSP	20% to 40% margin	9 to 18 months to productive	5% to 30%	Markets you cannot staff directly
Embedded or OEM	10% to 30% revenue share	9 to 24 months per partner	0% to 40%, lumpy	After direct is repeatable

Two things fall out of that table. Cloud marketplaces are the cheapest intermediated route by a wide margin and the one most companies skip. Resellers are the most expensive and the one founders sign first, usually because a partner asked and it felt like free growth.

How to pick two routes and deliberately ignore the other four

The sequencing question comes up constantly and the answer is boring. Get one route working end to end before adding a second, then choose the second based on where the first one hits a wall. Self serve hits a wall at the security questionnaire, so the second route is usually sales. Sales hits a wall on procurement friction at enterprise size, so the second route is usually a cloud marketplace. That progression is also roughly the arc described in the SaaS go to market strategy playbook, and you can see it play out concretely in a worked SaaS marketing strategy example.

What to do next

Pull last quarter’s closed won list and tag every deal with the route that carried it. Most teams find one route quietly producing 80 percent of revenue and three others producing rounding errors with full-time attention attached to them.

Then answer one question: does your buyer hold a committed cloud spend agreement. If the answer is yes and you are not transactable on AWS or Azure, that is the highest return distribution work available to you this year, and it is mostly engineering and legal rather than marketing spend. If the answer is no, the decision is between deeper marketplace commitment and more sales capacity, and it turns on ACV.

Everything else, including the reseller agreement somebody signed two years ago, can wait until you have done that. For the demand side of the same decision, the companion guide on SaaS distribution strategy covers how to generate pull through a route once you have chosen it, and the broader guide to building a SaaS marketing strategy puts distribution back into the wider plan.