Handing a project between teams: what must travel with the files
Organise responsibilities, access trials, decisions and open issues so the receiving team can operate the project.
Article language: English

In this article
Everything is in the folder: the design, source files, presentation and even a recording of the farewell meeting. The missing answer is who will correct an error on the website tomorrow and how they will check whether the change worked. A handover ends only when the receiving team can perform the necessary tasks, understands responsibility boundaries and knowingly accepts the remaining issues.
Access to files confirms they can be opened. It does not prove transfer of knowledge, permissions or operational support. For a useful handover, connect materials to specific actions: what the new team must maintain, how it will decide on changes and whom it will contact when an event exceeds the agreed scope.
Define what actually moves to the other team
Take a fictional example: a design team finishes a service company's new website and marketing, together with a technical person, will take over daily care. “We are handing over the website” is too broad. Separately describe content, publishing, the contact form, language versions, technical updates and incident response.
For each area identify tasks transferred now and those still handled by the previous team or provider. Agree the boundary between a correction and new work, and the reporting route. These organisational arrangements must reflect the actual scope of collaboration. They do not replace the terms of contracts.
The article on responsibility for maintaining a business website (in Polish) provides a useful starting point. A team may accept content publishing without taking over the entire infrastructure. Say so plainly before the first incident sends everyone to a shared inbox hoping somebody else will answer.
Also distinguish the date materials become available, the start of acceptance trials and the transfer of responsibility. They may differ. One “handed over” entry must not obscure the stage when the recipient is still checking its own readiness.
Build the package around the recipient's tasks
Do not begin by copying the whole project drive. First list actions the new team will perform often and those that are rare but important during a problem. For each, identify instructions, the required account, input material and a way to check the result. From that you can derive the files to transfer.
In our example, marketing must publish an offer correction, replace an image and assess a phone preview. The technical person must recognise failed form delivery, check the site and find the right response path. A set of design screens is not enough for any of these tasks.
You can organise the package into:
- Scope of work and responsibilities.
- Instructions for key actions.
- Locations of source material and current versions.
- An account and role register and how to obtain required permissions.
- Known limitations and open issues.
- Contacts, reporting paths and rules for further changes.
Connect publishing instructions to your CMS content process. The guide to CMS publishing (in Polish) discusses the steps and roles. Explain which stage is preview, which is approval and which is actual publication. This matters especially when environments look alike. “I can see it” may mean a correct preview and no change at all for the customer.
Transfer the reason for a choice with its result
The new team should know why certain elements work as they do. Otherwise it may remove a limit protecting an important dependency or maintain a workaround for years after its reason disappears. A description of the current appearance alone does not provide that knowledge.
Include your project decision register with the major decisions recorded in it. The guide to a project decision register (in Polish) explains how to preserve their context. Identify deliberate choices, compromises and temporary arrangements. Do not expect the recipient to reconstruct history from hundreds of messages. Chat archaeology can be fascinating but is difficult to put into a maintenance plan.
If decisions changed scope during the project, retain their current outcome and links to earlier reasoning. A short decision note helps with the next change. The new owner need not ask the previous team about everything that looks unusual at first glance.
For a related approach to summarising a decision, see the management decision memo guide (in Polish).
For a multilingual offer, explain which version is the basis for updates and how differences are agreed. Version control for a multilingual offer helps. Files named “final” and “final_corrected” describe neither scope of change nor who approved it. The multilingual offer versioning article (in Polish) follows changes across the website, PDF and sales materials.
Test access in action, not in an account list
The recipient should perform a trial with their own account and a role matching future tasks. Watching a fully privileged person demonstrate does not establish that the receiving team can do the same. Nor does it establish that the granted access is needed for daily work.
Start with ordinary actions, then test one problem-response path under agreed test conditions. Transfer passwords and other secrets through the company's proper access-management method; do not put them in general instructions. The document should say how an authorised person obtains access, not become a shared store of every key.
If the project involved agents or automation, check the scope of code handover, access and confidentiality. The guide to code handover and confidentiality with AI agents (in Polish) discusses what to check. Check task-executing accounts, dependencies on individuals and the actual ability to take over support. Changing an owner's name in a presentation does not switch the operation of an integration.
Record each trial: task, account or role, expected result and actual result. For a failure, identify the cause and the person responsible for a fix. Acceptance testing should leave evidence, not just a memory that everything somehow worked during a call.

Establish readiness to assume responsibility
In Google SRE's service engagement model, operational responsibility is preceded by a readiness review that includes dependencies, monitoring and incident response. This is an example from operating Google systems. For a smaller project, the useful inspiration is checking the recipient's readiness before formally treating responsibility as transferred.
For a website, readiness can be described through concrete trials: publishing a change, confirming the form's route to handling, finding the correct source material and recognising a situation requiring technical support. Adapt the detail to project scope and risk. Not every small site needs the elaborate operating model of a large infrastructure team.
Use whole-journey website acceptance (in Polish). Acceptance of delivery and transfer of maintenance may share some evidence but answer different questions. The first checks the agreed result; the second checks whether named people can continue to operate it.
If the new team is not ready, name the missing conditions. It may need access, instructions, practice or a decision about responsibility. Do not sign an unclear “accepted with comments” if those comments concern necessities for daily work. The record should show what can be transferred now and what awaits a condition.
Show open issues without glossing over them or catastrophising
A known-issues list is part of the handover. For each issue describe the effect on users or operations, workaround, owner and agreed path forward. If a repair date is unknown, record when it will be determined. Do not replace specifics with “minor” when the new team does not yet know the real impact.
When the person handling active cases also changes, an up-to-date shift handover is needed: actions completed, expectations and the next response time. The project package describes broader operating capability; that record concerns a specific ongoing case. For a fuller guide, see the article on shift handovers (in Polish).
Separate an existing problem from a development idea. A fault preventing enquiries and a proposal for a new animation should not compete on one unlabelled list. The recipient needs to know what disrupts agreed operation and what is an optional future change.
Prepare your own website incident-response plan too. The guide to preparing a website outage response plan (in Polish) discusses roles and escalation. Explain who receives a report, confirms the event and calls in extra support. A person under pressure should immediately find the right instructions without guessing abbreviations or company sayings.
Ensure old lists do not look like current assignments. Mark completed and replaced materials while preserving necessary history. Handing over five competing plans is not rich documentation; it is five ways to start in the wrong place.
Run a trial with roles reversed
In ordinary training the existing team demonstrates and the recipient watches. In a handover trial, reverse the roles. The person taking over receives a task and uses the prepared learning materials, while the provider observes where more explanation is needed. Do not immediately suggest the next click. A temporary difficulty helps identify a gap before it appears in a real customer case.
The article on worked examples and training materials (in Polish) offers guidance on choosing learning support.
In our example, marketing receives a request to change one offer paragraph without publishing before approval. It should find the current source, prepare the change, check the preview and refer it to the right person. Separately, the technical person demonstrates how they would recognise a report of an undelivered form and where they would find the relevant operational trace. Agree test conditions so they do not disrupt real service or send misleading messages to customers.
Record where the recipient stops. The cause may be missing permission, but also an unclear material name, an outdated screenshot or a missing success criterion. Each needs a different correction. More training will not repair an account with the wrong role, and broad permissions will not explain who approves a content change. Identifying the cause matters more than merely recording a failed trial.
After corrections, the recipient repeats only paths whose results remained uncertain, unless a change affected a wider scope. There is no need to replay all training merely for a sense of formal order. Evidence that the gap was fixed and identification of the instruction version tested are needed. When team membership changes, that record helps choose tasks for a newcomer without reconstructing every conversation.
Ask the recipient to explain in their own words what they own, what they do not change independently and how they recognise a need to escalate. The conversation often exposes a difference between knowing the interface and understanding the process. Someone may publish text fluently yet not know that changing service scope requires agreement with the offer owner. The handover package must contain both kinds of knowledge.
Plan the first weeks after handover
Agree a period of jointly checking work: types of questions, reporting channel, people's availability and when independence will be assessed. Do not promise unlimited support “if needed” without agreed resources. The record should let both teams plan, even if the first days expose a missing instruction.
The Service Manual on running a service sustainably describes planning ongoing improvements with service signals and priorities. It concerns UK public services. A company can similarly distinguish the one-time handover from responsibility for later corrections and changes.
Collect questions from the first period and improve documentation accordingly. If one person repeatedly answers the same question, it is not enough to regard it as explained in conversation. Update the place the next person will use. Handover also supports onboarding employees into company knowledge. The employee onboarding article (in Polish) considers tasks, sources and checking independent work.
At the end, ask the recipient to demonstrate one change, one result check and one escalation route independently. This is not a memory test. They may use the instructions - that is why they exist. The result shows whether the team acquired the ability to work or only a collection of well-named folders.
If you are preparing a website or organising its handover to a team, see the Aeneia Design offer. Bring a list of tasks specific people should take over and an example of the current instructions. It will be easier to determine what belongs in the package and what first needs a practical check.










