The core answer
If your company uses software whose copyright and code stay with the developer, and your business cannot stop when the system stops, ask for a source code escrow agreement before you sign. It is a three-party contract. The developer deposits the code and documents with a neutral escrow agent, and the customer receives a copy only when an agreed event occurs, such as the developer closing down, ending support, or seriously breaching the support terms.
A working agreement answers five questions: what is deposited and how often it is updated; whether anyone verifies that a system can be built from the deposit; which events trigger release and who decides if the parties disagree; what the customer may do with the code; and who pays which cost. An agreement that says only "the developer will deposit the source code with a third party" often fails when it is needed.
Legal basis (Copyright Act B.E. 2537 and the Civil and Commercial Code) Software is a copyright work, and copying or adapting it is a right of the copyright owner, who may permit others to use it within agreed limits. Receiving code files therefore does not by itself give the customer any right to do anything with them. The right must come from the escrow agreement or a licence made in advance. On termination, section 386 of the Civil and Commercial Code provides that where a party has the right to rescind, it is done by declaring intention to the other party, and the declaration cannot be withdrawn. Treat a release notice the same way: state the trigger and attach the evidence before you send it.
Our practical view Escrow is not an insurance policy that ends the matter once bought. It is only a route to getting the code back. What keeps a business running is code that builds, a licence that is clear, and a team that understands the code.
This guide covers the escrow mechanics only. For who owns the code and what must be delivered under a development contract, see software development contracts and source code. If a SaaS provider has already announced it is closing, see SaaS provider shutdown: getting your data and work back.
Who should read which part
| Your situation | Sections to read |
|---|---|
| Not signed yet | Sections 1 to 8 for negotiating and drafting, and section 9 as a clause checklist |
| Signed and in use | Sections 3 and 4 to check that updates and test builds really happen, and section 7 to check the rights you would get |
| Already in dispute | Section 5 on triggers and notices, section 6 on the escrow agent's role, and section 7 on the licence |
1. What is a source code escrow agreement, who are the parties, and when is it worth it?
There are three parties.
- The depositor is the developer or software provider that holds the copyright and the code.
- The beneficiary is the customer company that uses the software.
- The escrow agent is a neutral third party that holds the deposit and releases it on the agreed conditions.
Agreements usually take one of two forms: a three-party agreement made at the start for a single customer, or a master agreement between the developer and the agent that each customer later joins as a beneficiary. The second suits software with many customers.
When it is worth it. When the software is specialist and hard to replace, when your data and processes are tied closely to it, when the developer is small or depends on a few people, or when the contract is a licence rather than a purchase of the code. If you could move to another product within a few weeks, the cost and effort of escrow may not be worth it.
SaaS is different. If the developer runs the service on its own systems, the customer does not install the code. Source code alone does not bring the service back, because the service also consists of servers, databases, configuration and third-party services. Escrow for SaaS therefore often has to cover parts of the production environment, and should be written together with data export rights and a transition period, as described in the SaaS shutdown guide.
2. What should be deposited?
The aim is to deposit enough for a skilled team that has never seen the system to build and run it. Source code alone is usually not enough.
| Item | Why it matters |
|---|---|
| Latest production source code, with change history if agreed | It is the system itself, and it must match the version the customer actually runs, not an old one |
| Build and deployment scripts | Without them the team must guess how the system is built |
| List of libraries, frameworks and their versions | Much code will not build if component versions change |
| Database structure and migration scripts | The system needs a correctly shaped database to start |
| Architecture documents and an administrator manual | So a new team can understand the structure and routine operation |
| Environment set-up steps and the list of external services the system calls | Systems often depend on payment, email, mapping or cloud services of third parties |
| List of open-source and third-party components with their licence terms | Shows what may be reused and what must be licensed from its owner |
| Technical contact names and how to reach the code repository | For the first questions to the people who maintained it |
Passwords and keys. Do not deposit live passwords in a package that will be held for a long time: old credentials expire or leak, and several people can reach a deposit. Deposit a list of the credentials and keys that exist and how to recreate them, and agree a step under which the developer or administrator issues new ones on release. If the system holds personal data about end users, do not deposit real data with the code. Use sample data for testing, and see the PDPC for personal data rules.
Third-party components. The developer can license to the customer only what it has the right to license. Open-source and third-party parts are governed by their owners' terms, so ask the developer to list them in a schedule.
3. How often must the deposit be updated?
A deposit that is never updated becomes an obsolete package that no longer works. Agree the following in advance.
- Update cycle. For example after every major release, or on a fixed schedule. If the developer changes the system weekly, a yearly cycle is not enough.
- Version confirmation. The developer states that the deposit matches the version the customer runs, and gives the version number.
- Who is told. The escrow agent notifies the customer when a new deposit is made, or when one is due and has not been made.
- Consequence of not depositing. It should count as a breach, and may entitle the customer to request verification or to withhold the part of the fee agreed for that purpose.
- Keeping old deposits. Agree how many earlier deposits are kept and for how long, in case the latest one is damaged.
4. How do you know the deposit can actually be used?
Depositing without verification is the most common weakness. The box may hold incomplete files or a system that cannot be built, and you usually find out only when you need it.
| Level of verification | What happens | Suits |
|---|---|---|
| Completeness check | The agent checks that the deposit matches the schedule, files open and contain no malware | Software that is not critical |
| Test build | A verifier builds the system from the deposited code and scripts following the documents, and reports pass or fail | Important systems, at least on first deposit and after major changes |
| Full verification | Build, install in a test environment and try the main functions against the live system | Systems the business cannot do without |
Points to agree on verification:
- who the verifier is and whether it is independent of the developer;
- when the customer may ask for verification and how much notice is needed;
- if verification fails, how long the developer has to correct and redeposit, and what follows if it does not;
- who receives the report and how it is kept confidential;
- that the verifier sees the code only to the extent needed and is bound by confidentiality.
Remember that a pass means the system could be built from the deposit on that day. It does not mean the system has no defects or that a new team will find it easy to maintain.
5. Which events release the code, and who decides?
The release trigger is the heart of the agreement. If it is loose, the developer fears release without cause. If it is narrow, the customer cannot get the code on the day it is needed.
| Common trigger | Say clearly |
|---|---|
| Developer closes down, liquidates or stops operating | When it occurs, for example registered dissolution, or no response for longer than a set period |
| Developer is insolvent, bankrupt or in rehabilitation | Which fact is the marker, for example a court order; see the note below |
| Developer ends support or development of the product (end of support) | Written announcement, or no security fixes for longer than the agreed period |
| Developer does not fix a serious defect in the agreed time | Definition of serious, response and fix times, and a written warning first |
| Developer seriously breaches the support contract and does not cure in the time given | Separate from minor delay, and set a cure period |
| Developer transfers the business to someone who does not accept the escrow agreement | The transferee must take over the escrow agreement, otherwise it is a trigger |
Note on insolvency. Naming insolvency as a trigger is worth doing, but its real effect depends on bankruptcy law and court orders. Once the developer's assets are placed under a receiver or a rehabilitation plan, the person appointed to manage them may have authority over how assets and contracts are dealt with. Do not assume the clause works at once in every case. Take advice before sending a release notice. For the creditor side, see the creditor guide to bankruptcy and rehabilitation.
Who decides that a trigger has occurred? Choose a method and write it down.
- The customer gives written notice stating the trigger relied on and attaching evidence. By analogy with rescission, which is made by declaring intention to the other side (section 386), a notice once sent cannot be withdrawn, so check the trigger before you send it.
- The escrow agent informs the developer at once and allows a short period, agreed in advance, for a written objection (a counter-notice).
- If the developer does not object, the agent releases the deposit to the customer.
- If the developer objects, nothing is released until the dispute is resolved, by an independent expert or arbitration as agreed, or by the court. Make this process fast, because the customer's system may be down while it waits.
A trigger that is a plain fact, such as the developer's registered dissolution, can allow release without waiting. A trigger that needs judgment, such as a serious defect or end of support, usually needs the objection step. Contracts are interpreted in good faith and with regard to custom (section 368), but do not rely on interpretation: write the definitions yourself.
6. What does the escrow agent do, and not do?
It does
- receive and keep the deposit securely, and not open it beyond what the agreement allows;
- verify to the agreed level, or arrange a verifier;
- receive updates and notify the customer on the agreed cycle;
- receive release notices, pass them to the developer, count the objection period and release when the conditions are met;
- keep a record of who deposited, who requested, and when it was released.
It does not
- decide for the parties whether the trigger really occurred, unless the agreement clearly gives it that role; usually it follows the conditions and steps written down;
- guarantee that the code has no defects, does not infringe anyone's rights, or is lawful;
- maintain the system or provide technical services in the developer's place;
- grant the customer a right to use the code: the right comes from the agreement and licence, not from the agent.
Choosing an agent. Look for neutrality and no link to the developer, secure storage and backup, the verification levels you need, its own rules for disputes, and the ability to keep operating even if the depositor disappears. Do not let the developer choose alone; the customer should accept the agent too.
7. Once I hold the code, what may I do with it?
This is where escrow agreements most often fall short. The customer can receive the code but has no wording allowing it to do anything with it.
The reason is that software is a copyright work, copying and adapting it is the owner's right, and the owner may permit use only within stated limits. Use beyond that may infringe copyright. The escrow agreement or the main software contract should therefore contain a release licence agreed on the day of signing. It should answer the following.
| Issue | What to write |
|---|---|
| Purpose | Use to maintain, fix and develop the system for the customer's own needs, not to sell in competition with the developer |
| Copying and modifying | Copy and modify the code to the extent needed to maintain the existing system and add capability |
| Who may use | The customer and named affiliates, or also contractors engaged to maintain it, who are bound by confidentiality |
| Distribution and resale | No distribution, resale or licensing to others as a product |
| Confidentiality | The code is confidential; the customer must protect it and limit access |
| Duration and end | While the customer continues to use the software under the original contract and pays the agreed licence fees, or for the agreed period |
| Work the customer builds on it | Who owns modifications made by the customer or its contractor, and who may keep using them |
| Third-party components | The customer may need permission from the owner of those parts if the developer cannot sublicense them |
This is not a transfer of copyright. Escrow gives the customer a right to use on release, not ownership. If the customer wants to own the code, it needs a development contract and a written copyright assignment, as explained in the software development contract guide.
Having code is not yet having a system you can maintain. Plan who will understand the code, how long preparing and testing the environment will take, whether the third-party services the system calls can still be used, and how the business works in the meantime. Keep a written fallback, such as temporary manual working and a list of data exported beforehand.
8. Costs and who bears them
There is no fixed rule about who pays. It is negotiable, depending on bargaining power and how important the system is. Costs usually fall into these parts (no figures are given here because they depend on the provider and the level of verification).
- Set-up and agreement fee. The escrow agent's initial charge.
- Annual service charge. For safekeeping and tracking updates.
- Verification fee. According to the level, and sometimes charged separately for each major update.
- Release fee. The cost of the release process and any objection procedure.
- The customer's own adviser and developer costs. Once the code is received and the system must be rebuilt, this may be the largest real cost.
Common patterns. The customer pays because it asked for escrow, or the developer pays the deposit costs as part of the terms of sale, or the parties split them. If several customers join one deposit, verification and annual charges can be shared. State what happens if a party does not pay (for example the agent may suspend service, which can harm the other party) and whether the other party may pay in its place to keep the agreement alive.
Treat the escrow cost as the price of continuity planning, and compare it with the loss if the system stops and with other options, such as switching provider, running a backup system, or obtaining access to the code repository directly.
9. Clause checklist
- Identify the developer, the customer and the escrow agent, and the authorised signatories (check the developer's company affidavit)
- State the product and versions covered, including future releases
- Schedule listing the deposit: code, build scripts, libraries, database, documents and third-party components
- How passwords and keys are handled: no live credentials deposited, and a method to issue new ones on release
- Update cycle, version confirmation and consequence of not depositing
- Verification level and frequency, the verifier, the report, and what happens if verification fails
- A definition of each release trigger, who decides, notice steps, objection period and timings
- A fast dispute route, and release while the dispute runs if agreed
- The escrow agent's duties and limits of liability
- A release licence: scope, users, duration, confidentiality and third-party parts
- Each cost and who pays it, and the consequence of non-payment
- The agreement binds any transferee of the business, and the developer may not cancel or amend the escrow without the customer's consent
- When the escrow agreement ends, and how the deposit is returned or destroyed
- Confidentiality of the code, governing law and dispute resolution
Hypothetical example: a logistics company using a small developer's warehouse system
This is a hypothetical case to explain the points. It is not a real matter of any client.
Company A runs a transport business and uses a warehouse management system built by Company B, which licenses it on an annual basis. The system is tied to daily deliveries, and Company B has two main programmers.
Before signing, Company A asks for an escrow agreement with an escrow agent. Company B bargains over cost and worries about leaks. They agree:
- deposit the code, build scripts, library list, database structure and installation manual, but no live passwords;
- update at every major release and at least every six months (this interval is a hypothetical figure);
- one test build after the first deposit and after every major release;
- triggers: Company B registers its dissolution, stops responding and does not fix a serious defect within the set time after a written warning, or announces end of support in writing;
- if Company B objects, an independent expert decides within the agreed time;
- on release, Company A may copy and modify the code for internal use and have contractors bound by confidentiality maintain it, with no resale;
- set-up costs are split, while annual fees and verification are paid by Company A.
During use, the first test build fails because a script refers to an internal server that was not deposited. Company B corrects it and redeposits within the set time.
When the event occurs, two years later Company B stops responding. Company A sends a warning, the cure period passes in silence, and it sends a release notice through the agent. The agent informs Company B, no objection arrives within the period, and the code is released to Company A.
After release, Company A needs several weeks to find a team with the right technology, prepare the environment, and replace a messaging service the system used and which it has no right to continue using. In the meantime it follows the fallback it had prepared, such as printing job sheets from data exported earlier. The example shows that escrow gives the business a way forward, but does not bring the system back at once.
Common misunderstandings
"We have the code on deposit, so if the developer disappears we can run the system immediately." Not necessarily. Someone must understand the code, the system must be built and tested, and the third-party services it relies on must still be available. Have a separate business continuity plan.
"Once we have the code files, we can do whatever we like with them." No. The developer may still own the copyright. Rights to copy and modify must come from a licence written in advance, and going beyond it may infringe copyright.
"It is deposited, so it is safe and needs no checking." A deposit that is never verified may be incomplete or impossible to build. Require at least one test build and updates on a cycle.
"The moment the developer becomes insolvent, we can get the code." A clause releasing code on insolvency is worth having, but its effect depends on bankruptcy law and court orders, and the person managing the developer's assets has a role. Take advice before you send the notice.
"The escrow agent checks that the software is good and lawful." No. The agent keeps and releases on the stated conditions. It does not vouch for code quality or for the absence of infringement.
"An escrow agreement makes us the owner of the code." No. It gives a right to use on release and does not transfer copyright. To own the code you need a written copyright assignment in a development contract.
Limits of this guide. It covers escrow for software whose copyright stays with the developer. It does not cover development contracts and copyright assignment, personal data protection, tax, or how bankruptcy law applies in a particular case. Information as at 21 November 2026. The effect of an agreement depends on its wording and the facts of each case.
Further reading
- Software development contracts: who owns the source code and how to accept the work
- SaaS provider announces it is closing: getting your data and work back
- Subscription SaaS contracts: pricing models and clauses to check
- Creditor guide to Thai bankruptcy and rehabilitation
- Limitation of liability, indirect damages and indemnity in business contracts
A workable source code escrow agreement states what is deposited, the update cycle, the test build, the release triggers and who decides, the rights of use after release, and who pays the costs, but receiving the code does not mean you can maintain the system at once.
We review software and SaaS contracts, draft and negotiate escrow agreements and the licence that takes effect on release, and set out the intellectual property terms and the legal side of business continuity planning. We are not software engineers, we do not assess the quality or security of code, we do not act as escrow agent, and we cannot promise the outcome of negotiations with a developer or of insolvency proceedings. Fees depend on the stage of the matter and the work required. Send us the documents first, and we will give you the scope of work and fees in writing before we start. Call +66 92 254 2045 or send us the details. See also contracts and documents and intellectual property.
This guide is published by Suwanvara Law Firm — a Khon Kaen law firm established in 1986. General information only, not legal advice on a specific matter.

