Consulting and recruiting
Software implementation partners: what partner agreements say about your work
By SourceX Editorial · Reviewed by Noah Loul ·
Short answer
Implementation partner IP sits in three layers: the software vendor's partner and developer terms, each client's MSA and statement of work, and the firm's own pre-existing tools and records. Client-specific configurations usually stay with the client, while generic accelerators and internal project records are often the firm's. Read all three layers before reusing or licensing anything.
Key takeaways
- Three sets of documents decide ownership: vendor partner and developer terms, the client MSA and SOW, and the firm's own employee and contractor agreements.
- Configurations built inside a client's instance are usually tied to the client's data and subscription.
- A pre-existing IP clause protects an accelerator only if the firm can show the asset existed before the project.
- Contractor-built assets generally need a written assignment; a work-for-hire label alone may not transfer copyright.
- Vendor API terms increasingly restrict using data accessed through APIs for AI training.
Who owns the work an implementation partner does?#
Ownership of implementation work is split across three layers, and each layer controls different assets. The software vendor controls its platform and the terms of its partner program and APIs. The client controls its data and, usually, the deliverables it paid for. The firm controls what it brought to the project and the internal records it created while running the work.
Founders often assume partner status or a services agreement settles the question in one place. In practice a Salesforce, NetSuite or HubSpot consultancy needs to read all three layers together, because an asset the client contract lets the firm keep may still be restricted by vendor terms, and the reverse.
The three-layer rights table#
Use this table to sort any asset by the layer that governs it. Where two layers apply, the narrower one usually controls.
| Layer | Documents | What it typically governs | Question to answer |
|---|---|---|---|
| Vendor | Partner program agreement, developer and API terms, marketplace terms | Platform use, API data, demo environments, vendor confidential material | Can data accessed through APIs be stored, exported or used for AI? |
| Client | MSA, statement of work, data processing addendum | Deliverables, client data, configurations in the client's instance | Who owns deliverables, and is pre-existing IP licensed back? |
| Firm | Employee agreements, contractor agreements, internal policies | Accelerators, templates, methodology, internal project records | Did every author assign their rights to the firm? |
What vendor terms usually touch#
Vendor terms rarely claim a partner's own know-how, but they often limit what partners can do with data and environments the vendor provides. Partner program agreements commonly address trademarks, confidential roadmap material, demo and sandbox environments and conduct with joint customers. Developer and API terms govern data accessed through the platform.
API terms are tightening around AI. HubSpot's developer changelog says its updated Developer Terms restrict using data accessed through HubSpot APIs to train, fine-tune or improve AI or machine learning models, with a carve-out for legitimate single-customer use cases, and state that customer data belongs to the customer. A Hunton Andrews Kurth client alert reported that Salesforce changed the Slack API Terms of Service, effective May 29, 2025, to prohibit bulk exporting of data accessible through Slack's APIs and its use in large language models.
Platform terms also define what counts as the customer's material. In Oracle's NetSuite Terms of Service, Customer Data means content the customer provides that is stored in, or run on or through, the Cloud Service. Scripts and saved searches built inside a client's account sit close to that definition, which is one reason to treat them as client-side assets unless the contract says otherwise.
What client SOWs say about configurations and code#
Client contracts decide who owns deliverables, and most implementation SOWs give the client ownership of, or a broad license to, what was built for it. Read these clauses in every MSA and SOW before treating any asset as reusable.
Copyright law adds a wrinkle for contractor work. The Copyright Office explains that a work made for hire arises when an employee creates it as part of regular duties, or when a work in certain statutory categories is specially commissioned under an express written agreement. Software written by an independent contractor often falls outside those categories, so the firm generally needs a written assignment from the contractor, not just a work-for-hire label.
Evidence matters as much as wording. A pre-existing IP clause helps only if the firm can show an accelerator existed before the engagement, which is why dated commits in the firm's own repository are worth more than a folder of undated templates.
- Deliverables ownership: what the client owns on payment, and whether it includes code, documentation and configuration.
- Pre-existing IP carve-out: which firm tools and materials stay with the firm.
- License-back: the client's right to keep using firm tools embedded in its deliverables.
- Residuals: whether general know-how retained in staff memory stays usable.
- Confidentiality: how client confidential information is defined and how long the obligation runs.
- Return or destruction: what must be handed back or deleted when the project ends.
Reusing implementation assets: a decision table#
The table sorts common assets by whether a firm can usually reuse them with the next client and whether they could be considered for licensing to an AI developer. Treat it as a starting point for counsel, not a conclusion.
| Asset | Reuse with the next client | Consider for AI licensing |
|---|---|---|
| Generic migration scripts and templates written before the project | Usually yes, if pre-existing IP is carved out | Possibly, after a rights check |
| Client-specific scripts, flows and saved searches | Usually no, or only rewritten in generic form | Usually no |
| Solution design documents | Structure yes; client content no | Only with client details removed and rights confirmed |
| Tickets, estimates and retrospectives in the firm's own tools | Yes | Possibly, after removing client confidential details |
| Exports of client data or metadata | No | No |
| Training decks built from vendor material | Check vendor terms | Usually no |
| Recorded client workshops | Usually no | Usually no without client and participant consent |
How to protect accelerators on future projects#
Accelerators are protected by habits before contracts. The firm that can show where an asset came from, who wrote it and which clients it touched will win most ownership questions without a dispute.
- Keep accelerators in a firm-owned repository with dated commit history, separate from client work.
- List named accelerators as pre-existing IP in each SOW, with a license-back to the client.
- Require written assignment of rights in every employee and contractor agreement.
- Rewrite client-specific solutions into generic form inside the firm's repository, never by copying a client export.
- Log which accelerators were used on which projects, so an audit can trace them.
Illustrative: a CRM implementation partner audits its assets#
Illustrative: a fictional CRM consultancy holds partner status with HubSpot and Salesforce and implements both for B2B software and professional services clients. Its assets include data migration playbooks, workflow and pipeline templates, Apex classes written inside client orgs, recorded discovery workshops and an internal assistant prototype that drafts configuration notes.
The founder asks counsel to sort everything by the three layers. Playbooks and templates kept in the firm's own repository, with history predating its client work, are confirmed as firm-owned. Apex classes written in client orgs stay with those clients. Several templates came from a former contractor with no assignment clause, so the firm obtains a written assignment before relying on them.
The assistant prototype raises the vendor layer: it had been tested on records pulled from client portals through APIs. The firm stops that use until the current developer terms are reviewed and rebuilds its test set from internal tickets and estimates with client details removed.
How SourceX approaches partner-firm records#
For implementation firms, the Rights step of the SourceX five-step transaction (Supply, Rights, Preparation, Approval, Delivery) applies the same three layers, separating firm-owned records such as internal tickets, estimates and methodology from client deliverables and anything accessed through vendor APIs, which is generally carved out.
Each package that proceeds gets a SourceX Evidence Packet, one record of provenance, licensing rights, permitted use, the privacy record and release authorization that the firm and the licensee both hold. The firm keeps ownership and approves every step.
Frequently asked questions
Does partner status give the vendor rights to our methodology?
Usually not, but read the partner agreement for feedback, marketplace and co-development clauses. Some programs take a license to feedback or to content submitted through partner portals or marketplaces. Internal methodology documents are generally outside those clauses unless you submitted them.
Can we keep copies of client configurations after the project?
Only as the contract allows. Many SOWs require return or destruction of client materials at the end, and configurations exported from a client's instance may contain client data. Keep generic patterns rewritten without client details rather than raw exports.
Are our own Jira and Slack histories ours?
The firm generally controls its own workspace records, but they often contain client confidential information such as screenshots, data samples and system names. Treat them as firm records that need client details removed before any outside use, and check the tool vendor's terms on exports.
What about offshore subcontractors?
Check that each subcontractor agreement assigns rights in deliverables and tools to the firm and passes down the client's confidentiality and data terms. Gaps here are common and can weaken both the firm's ownership and its compliance with client contracts.
Do vendor AI-training restrictions apply to our internal tools?
They can, if your tools read data through the vendor's APIs. Restrictions in developer and API terms attach to data accessed through those interfaces, whoever built the tool. Review the current terms before building any AI feature on API-accessed data.
Sources
- HubSpot's developer changelog says its updated Developer Terms restrict using data accessed through HubSpot APIs to train, fine-tune or improve AI or machine learning models, with a carve-out for legitimate single-customer use cases, and that customer data belongs to the customer. Source
- According to a Hunton Andrews Kurth client alert, Salesforce modified the Slack API Terms of Service, effective May 29, 2025, to prohibit bulk exporting of data accessible through Slack's APIs, persistent copies and use of such data in large language models. Source
- Oracle's NetSuite Terms of Service define Customer Data as content the customer provides that is stored in, or run on or through, the Cloud Service. Source
- Copyright Office Circular 30 explains a work made for hire arises when an employee creates the work as part of regular duties, or when a work in certain statutory categories is created under an express written agreement with a commissioning party. Source
Related resources
See if your company qualifies
A short company assessment. No data uploads are needed.