Custom-coded workflow actions let a developer run real code as a step inside your HubSpot workflows, so automation can do far more than the standard point-and-click actions. We build, test and document these actions as proper software, not no-code glue.
A custom code action is a programmable step inside a HubSpot workflow. It reads data from the enrolled record, runs your logic in JavaScript or Python, calls HubSpot or third-party APIs, and returns output values that later workflow steps can use. These are the jobs it does best.
Call a third-party API for firmographic, address or company data at the exact moment a record enrols. The action writes the results straight back onto the contact, company or deal, so downstream steps and reps work from enriched, current data rather than blanks.
Push or pull records to and from an external system such as an ERP, billing platform, support tool or internal database. The code fires at the precise point in the process where the data is needed, keeping both systems in sync without batch jobs or manual exports.
Normalise phone numbers, names, addresses, VAT numbers and domains, then catch duplicates before they pollute the CRM. This is logic the standard workflow actions simply cannot express, written once and applied consistently on every enrolment.
Score, route and round-robin leads using your own rules and external signals rather than HubSpot's built-in scoring. The action evaluates each record against custom criteria and assigns or branches accordingly, so the right lead reaches the right owner.
Derive values across associated records, for example summing deal values or producing a weighted score, and write the result to a property. The action does the maths HubSpot's formula fields and standard actions cannot, then exposes it for reporting and branching.
Send a webhook or notification to a tool like Slack, create a record in another platform, or kick off a downstream process. The same approach can power bespoke business rules behind a customer-portal action without standing up any separate infrastructure.
We treat each action as a small piece of production software: scoped, engineered against HubSpot's real runtime limits, tested on live data and documented for your team.
We define the enrolment trigger, the input properties (up to 50 per action), the logic and the output fields you need mapped into later workflow steps. We confirm your tier covers it, because custom code actions require Data Hub (formerly Operations Hub) Professional or Enterprise, and design around the hard limits before a line of code is written.
Our engineers write the action in JavaScript on the Node.js runtime, or Python where it fits, using the provided packages such as @hubspot/api-client, axios and lodash, or requests on the Python side. Arbitrary npm or pip installs are not supported, so we build cleanly against the supported runtime rather than fighting it.
Any HubSpot or third-party API access runs through a private app with the correct scopes, and every key or token is stored as a HubSpot Secret and read from environment variables, never hard-coded. We keep within the 1,000-character total secrets limit and handle authentication for each external system the action touches.
Each action must complete within 20 seconds, 128 MB of memory, a 65,000-character output limit and the 50-property input cap. We batch or re-architect anything that risks a timeout, add proper error handling and retry logic (throwing in the catch block so HubSpot retries cleanly), and respect HubSpot API rate limits, which matter most on large initial enrolments.
We test the action on sample records before go-live, confirm the output fields flow correctly into the rest of the workflow, then document how it works so your team can maintain and extend it. The result is a transparent, owned automation rather than a black box.
Everything you need to know about our services
Custom code actions are gated to Data Hub Professional or Enterprise (this was called Operations Hub until the September 2025 rebrand; the feature and pricing are unchanged). They are also available on the equivalent Marketing Hub or Sales Hub Enterprise combinations where workflows allow it. They are not available on Free or Starter tiers. Data Hub Professional starts at roughly £600 per month and Enterprise higher again, so this licence sits alongside our build work. Please confirm the current figure on HubSpot's pricing page. If you are not sure what you hold, we will check during scoping.
Actions run on HubSpot-managed serverless infrastructure (AWS Lambda), so there is no server for you to host or maintain. We write them in JavaScript on the Node.js runtime, or Python on the Python runtime (Python is currently in beta). Each execution must finish within 20 seconds, use no more than 128 MB of memory, return no more than 65,000 characters and accept up to 50 input properties per action. We engineer around these limits from the start, batching or re-architecting any logic that would otherwise time out.
Indicatively from £2,500, which covers discovery, building, testing and documenting a real custom-coded automation of one to a few actions. That figure is a starting point, not a quote. The actual cost depends on scope, the number and complexity of integrations, the volume of data being processed, your error-handling and rate-limit requirements, and whether existing infrastructure needs work. Larger multi-system builds scale up from there. We work to a fixed price wherever the scope allows, and our Default to Build process carries our guarantee: delivered on time, or you get 20% back. The HubSpot tier licence is separate and paid to HubSpot.
We are a software engineering firm that happens to be a HubSpot Diamond Partner, not a configurator. We have an in-house engineering team, hold HubSpot Custom Integration Accreditation and Cyber Essentials Plus, and our own site runs on React UI work and HubSpot serverless functions, so the capability is real and demonstrable. Where it fits, we deliver productised, customisable solutions hosted on Railway that you can run yourself, so you own transparent, maintainable software rather than a workaround held together with no-code glue.
Credentials are never hard-coded. Any HubSpot API access uses a private app scoped to only what the action needs, and every external key or token is stored as a HubSpot Secret and read from environment variables at runtime. HubSpot caps the total length of all secrets at 1,000 characters, which we work within. This keeps sensitive credentials out of the code itself and out of version control.
This is exactly the failure mode we engineer for. Because each execution has to complete in 20 seconds, we keep external calls tight and batch heavier work. For rate-limit and transient errors in Node.js we throw the error in the catch block so HubSpot retries the action rather than silently failing. We also account for HubSpot's own API rate limits, which bite hardest on large or initial workflow enrolments, and design the enrolment strategy accordingly.
Decision-stage reading to help you scope the right approach.
When a native action is enough, when you need custom code, and when to call a developer.
The limits, retries and maintenance debt teams underestimate when building it themselves.
A decision matrix for the three approaches, by use case, reliability and cost.
