Updated June 18, 2026 by Sunny Chauhan.

There are five distinct ways to build a HubSpot integration in 2026, and the founders who pick the wrong one usually figure it out around month four when the cost or capability ceiling shows up. The decision is more architectural than tactical: it determines whether you can list on the Marketplace, how much your customers will see HubSpot vs your product, what fails when HubSpot's API has an incident, and what you're paying every month for the rest of the integration's life. Here's the five-pattern map, the decision tree, and the failure modes I've watched each pattern hit at scale.

Pro Tip

TL;DR Five HubSpot integration patterns: 1/ Middleware-driven (Zapier, Make), 2/ iPaaS-orchestrated (Tray, Workato), 3/ Native Marketplace app, 4/ Embedded iPaaS (white-labeled connector inside your product), and 5/ Hybrid (Marketplace app + iPaaS connectors). Pick by three dimensions: customer overlap with HubSpot, integration depth, and whether you need a Marketplace listing. Native Marketplace app wins for deep, multi-tenant, want-to-be-listed; embedded iPaaS wins for broad-but-shallow integration breadth; middleware wins for low-frequency, simple flows.

The customer sets up an automation in Zapier or Make that connects HubSpot to your product. The customer owns the integration. You provide a Zapier integration (the API + webhook surface that Zapier can call).

When it works: → Customers want simple flows (when X happens in HubSpot, do Y in your product) → Low volume (each flow runs a handful of times per day) → You don't need a Marketplace listing

When it breaks: 1/ Volume scales up. Zapier has its own rate limits and per-flow run charges. At high volume the costs and reliability degrade. 2/ The customer wants something Zapier can't model. Conditional logic, multi-step orchestration, data transforms beyond Zapier's primitives. 3/ You want native-feel UX. Customers don't see your product inside HubSpot; they see Zapier between them.

Cost: Zero ongoing cost to you; customer pays Zapier directly. Engineering cost: 2-6 weeks to ship a Zapier integration.

Like Pattern 1 but with a more powerful platform. The customer (or you, on the customer's behalf) configures a Tray or Workato flow that orchestrates the integration. Better at complex logic, larger volumes, and enterprise-friendly features (audit logs, governance, more sophisticated error handling).

When it works: → Mid-to-deep integration with non-trivial business logic → You want speed-to-market over architectural ownership → Customer's HubSpot environment is one of many they care about (Tray/Workato are good at multi-system orchestration)

When it breaks: 1/ You hit the iPaaS platform's per-recipe or per-execution pricing ceiling. Margins erode. 2/ You can't list on the HubSpot Marketplace via Tray/Workato. Marketplace-listed apps must use direct OAuth integration. 3/ You're paying $30K-$150K/year for the platform plus your engineering time. The math gets ugly at scale.

Cost: $30K-$150K/year per platform. Engineering: 3-8 weeks to build the initial recipe set.

You build a direct HubSpot integration in your own backend, list it on the Marketplace, and own the customer relationship for the HubSpot connection. This is the pattern this entire blog series has been documenting.

When it works: → HubSpot is a strategic customer base (30%+ of your install base) → The integration is deep (bidirectional sync, UI extensions, workflow actions) → You want a Marketplace listing for distribution + enterprise procurement signal

When it breaks: 1/ Engineering investment is real (8-12 weeks for v1). 2/ Maintenance is yours forever. HubSpot ships breaking API changes, you fix them. 3/ Certification is a quality gate, not a quick rubber-stamp. May 2026 introduced security questionnaires.

Cost: $40K-$120K for v1 build. Ongoing: 10-20% of one engineer's time per year for maintenance.

See our HubSpot custom integration build guide for the actual build sequence.

You bake an iPaaS-style connector inside your product, white-labeled as your integration. Customers see "Connect to HubSpot" inside your UI; behind it, an embedded platform handles the OAuth, sync, and API calls.

When it works: → You support many integration destinations (not just HubSpot) → You want one place to manage all your CRM integrations → Integration depth is shallow-to-medium per destination

When it breaks: 1/ Deep integration patterns (UI extensions, workflow actions) are hard or impossible through embedded iPaaS. 2/ You can't list on HubSpot Marketplace via embedded iPaaS; the OAuth flow has to be your own. 3/ Per-customer per-connection pricing adds up at scale.

Cost: $30K-$200K/year in platform fees depending on volume. Engineering: 2-6 weeks to integrate the embedded SDK.

You build a native Marketplace app for the core HubSpot integration, and use middleware or embedded iPaaS for less strategic destinations. Your HubSpot story is deep and listed; your other CRM integrations are good-enough via platform.

When it works: → HubSpot is a strategic destination + you have 5+ other CRM destinations to cover → You can't afford to build deep native integrations for every CRM → You want a "first-class HubSpot, capable everything-else" positioning

When it breaks: 1/ Two integration architectures means twice the operational complexity. Onboarding, observability, support docs all branch. 2/ You're paying for both engineering time and iPaaS platform fees.

Cost: Hybrid of Pattern 3 and Pattern 4.

Three questions:

→ Yes → Native Marketplace app is in play. Continue to Q2. → No → Native Marketplace app is hard to justify. Consider middleware (Zapier) or embedded iPaaS.

(Drivers: enterprise procurement, distribution, brand signal, customer trust)

→ Yes → Native Marketplace app (or hybrid). Middleware and embedded iPaaS can't get you listed. → No → Embedded iPaaS or middleware are real options.

→ Deep (bidirectional, real-time, UI extensions, workflow actions) → Native Marketplace app. → Medium (bidirectional sync, no UI) → Native Marketplace app or embedded iPaaS. → Shallow (one-way sync, low volume) → Middleware works.

The combinations:

Rough numbers for HubSpot integration at 50, 500, and 5000 customers:

→ Pattern 1 (middleware): Customer pays Zapier. You pay $0/yr. Engineering: 4 weeks. → Pattern 3 (Marketplace app): Engineering: 10 weeks. Ongoing: 2-3 weeks/yr. → Pattern 4 (embedded iPaaS): Platform: $30K-$50K/yr. Engineering: 3 weeks.

At this scale, Pattern 1 is cheapest if you can tolerate the limitations. Pattern 4 buys speed.

→ Pattern 1 (middleware): Customer Zapier costs scale per-customer. Your reputation if Zapier breaks is bad. → Pattern 3 (Marketplace app): Same engineering investment. Ongoing maintenance grows but stays manageable. → Pattern 4 (embedded iPaaS): Platform fees scale per-connection. $80K-$200K/yr range.

At this scale, Pattern 3 starts looking cheaper than Pattern 4 in total cost.

→ Pattern 1: Not viable for most products at this scale. → Pattern 3: Engineering cost is fixed-ish. Per-customer cost approaches zero. Lowest unit cost by far. → Pattern 4: Platform fees can hit $300K+/yr. Margins squeezed.

At this scale, Pattern 3 dominates on unit economics.

At Appnigma we generate Salesforce 2GP Managed Packages, which is the Salesforce equivalent of a native Marketplace app. We're not yet generating HubSpot apps but the architecture maps cleanly: OAuth flow, CRM API integration, webhook subscriptions, UI Extensions for in-app surfaces.

The future state where no-code platforms generate HubSpot Marketplace apps would change Pattern 3's cost structure substantially. Engineering investment drops; the choice between Pattern 3 and Pattern 4 becomes about ongoing platform fees vs zero. Until that future arrives, Pattern 3 requires real engineering investment.

• [ ] Measured HubSpot customer overlap (current and projected) → Yes / No

• [ ] Identified integration depth required → Yes / No

• [ ] Decision on Marketplace listing (yes or no, with reasoning) → Yes / No

• [ ] Cost-modeled each viable pattern at 1x, 10x, 100x current scale → Yes / No

• [ ] Identified operational complexity of multi-pattern (hybrid) approach → Yes / No

• [ ] Confirmed engineering allocation matches the chosen pattern → Yes / No

• [ ] Plan for migration if scale forces a pattern change → Yes / No

A B2B CPQ product had 40% of their install base on HubSpot, wanted an in-HubSpot quote-builder experience, and needed enterprise procurement to see a Marketplace listing.

→ Q1: Yes, 40% on HubSpot. → Q2: Yes, Marketplace listing required. → Q3: Deep (in-HubSpot UI for the quote builder).

Pattern 3 picked. 12 weeks to ship v1. Marketplace certification passed first try. 18 months later they were at 2x their pre-listing HubSpot install rate, mostly from in-Marketplace search discovery + enterprise procurement now seeing them as a credible vendor.

Counter-scenario: a marketing analytics ISV had 8% HubSpot overlap, shallow integration needs (push event data), and no Marketplace listing intent. They picked Pattern 1 (Zapier integration) and shipped in 4 weeks. The integration handles their needs; they revisit when overlap grows past 25%.

A HubSpot integration platform is any system that connects HubSpot to other software. The five common patterns: middleware (Zapier), iPaaS-orchestrated (Tray, Workato), native Marketplace app, embedded iPaaS (Workato Embedded, Paragon, Merge), and hybrid combinations. The right pattern depends on customer overlap, integration depth, and whether you need a Marketplace listing.

An iPaaS (Tray, Workato) is a third-party orchestration platform that runs flows between HubSpot and your product. A native Marketplace app is a direct integration in your own backend, listed on HubSpot's Marketplace. iPaaS is faster to build but can't be listed; native is more engineering but unlocks Marketplace distribution.

A category of products (Paragon, Merge, Workato Embedded, Tray Embedded) that lets you bake an iPaaS-style connector inside your product, white-labeled as your integration. Customers see "Connect to HubSpot" in your UI; the embedded platform handles auth and sync underneath.

No. Marketplace-listed apps must use direct OAuth integration with HubSpot. Zapier and Workato integrations are the customer's integration with HubSpot through a middleware, not your app's integration.

Three questions: 1/ Is HubSpot 30%+ of your customer base? 2/ Do you need a Marketplace listing? 3/ How deep is the integration? Yes-Yes-Deep almost always means native. No-No-Shallow almost always means middleware or embedded iPaaS.

At 50 customers: middleware is essentially free for you (customer pays Zapier). At 500 customers: native ($40K-$120K build + maintenance) competes with embedded iPaaS ($80K-$200K/yr platform fees). At 5000 customers: native dominates on unit cost; embedded iPaaS fees become prohibitive.

Zapier is fine for low-volume, simple flows. The failure modes are: 1/ Volume scaling breaks pricing and reliability. 2/ Complex logic exceeds Zapier's primitives. 3/ You can't ship a Marketplace listing. 4/ Customers see "Zapier" in their integration story, not your product.

Sunny Chauhan is the founder and CEO of Appnigma AI, a no-code platform that generates Salesforce AppExchange-ready Managed Packages from natural-language prompts. He holds Salesforce certifications in Platform Developer II, Platform App Builder, Administrator, Data Cloud Consultant, and AI Associate. Appnigma has helped B2B SaaS companies including Warmly, Hyperbound, Pylon, Seam AI, and Avoma ship native Salesforce Managed Packages. The integration pattern analysis above draws on those engagements and the Salesforce-side equivalent of each pattern.

Originally published June 18, 2026. Last reviewed June 18, 2026. Pattern descriptions, cost ranges, and decision frameworks based on Salesforce ISV experience adapted to HubSpot specifications current as of the published date.

• HubSpot Custom Integration Build Guide

• HubSpot API Integration Guide

• HubSpot Webhooks Implementation Guide

• HubSpot OAuth and Scopes Complete Reference

• HubSpot API Rate Limits Guide

• HubSpot App Cards and UI Extensions

• HubSpot Sandbox Setup Guide

1/ HubSpot Developers, OAuth Quickstart Guide 2/ HubSpot Marketplace Certification Requirements 3/ Truto, How to Build a HubSpot Integration 2026 Architecture Guide 4/ HubSpot Developers, CRM API Reference

Which of the five patterns describes your current HubSpot integration, and is it the right one for where you'll be in 12 months?