Website, automation or expansion: how to sequence connected changes
Map dependencies, test assumptions and set an order for website, automation and market-expansion work.
Article language: English

In this article
A new website, automation and entry into another market can all shout “me first” at once. The company calendar need not settle a contest of volume. Check dependencies: what must you learn and what must work before the next project makes sense? A campaign launched before customer handling is ready may simply create a backlog faster. A good sequence lets the result of one change inform the next decision.
Sequencing means understanding dependencies between projects, not choosing the trendiest technology. Ask what must be true for the next stage to make sense, who can confirm it and what small action can test a missing assumption. This puts the team's time where it produces knowledge needed for the next decision.
Describe the shared problem from the customer's perspective
Begin with the customer or employee journey. Who wants to do what? Where is the bottleneck? How does the company recognise it? Instead of three separate labels - “website rebuild”, “AI” and “Asia” - record the actual situation, such as difficulty explaining the offer and a long enquiry response time. Each project should identify the part it affects.
Gather a small set of evidence: example enquiries, versions of material, task completion time and observation of a user journey. You do not need every metric to start, but must distinguish confirmed difficulties from the team's ideas. If nobody has checked what customers understand from the page, a first-impression test can help before a broad rebuilding plan. The guide to a five-second test (in Polish) explains what that short test can reveal.
A shared problem does not mean one provider must do everything. It does allow goals to be compared. The website team knows which information to explain, the process owner knows required inputs and the market researcher knows which hypothesis is unconfirmed. Without that connection, each project can be executed correctly and still fail to unblock the most important task.
Map dependencies before setting dates
For each change, record required information, resources and decisions. A new language version needs a confirmed offer and a contact-handling plan. Automated enquiry routing needs categories and case owners. A campaign needs a working destination and a team ready to respond. Dependencies are concrete when their status can be checked. “Organisational readiness” without further detail is too broad for planning.
Mark which work can genuinely run in parallel. Research can continue while the Polish offer is clarified, but a finished translation should not rest on a text whose scope is still changing. Page layout can start before all copy is complete if the team knowingly accepts the likely later corrections. Record the assumption and when to test it rather than pretending tasks are fully independent. When dependencies change with external conditions, compare the plan across several different business scenarios. This separates broadly useful preparations from moves dependent on one uncertain future. The strategic scenarios article (in Polish) explains how to compare actions without treating a scenario as a forecast.
If considering entry into Asia, agree criteria for choosing the first market. The article comparing Korea, Japan and China (in Polish) sets out questions for that choice. Understanding a country does not establish the maturity of the service process. Equally, working automation in Poland does not prove that an offer fits a local audience. The plan should show these distinct assumptions so one successful project is not automatically treated as approval for every subsequent one.
Compare the next step through evidence and reversibility
Instead of assigning one score to entire multi-month projects, describe the next step of each. What can be prepared, tested and accepted within a limited scope? Which result would change the decision? How much work would reversing the trial require? A small action matters when it answers an important question, not merely because it fits easily into the calendar.
For a website, that might mean developing one service page and testing comprehension. For a process, trace several enquiries from entry to response. For expansion, interview a defined segment and assess the next purchasing step. Do not compare page counts with meeting counts. Compare the quality of evidence each brings to the shared decision and the cost of maintaining its result.
For each proposal, record:
- The problem and the people whose work should change.
- The missing assumption and how to test it.
- Dependencies, inputs and the person confirming readiness.
- The limited trial scope and result to accept.
- Team work required and capacity to handle the result.
- Conditions for continuing, correcting or ending the initiative.
Account for attention. Three small projects may require the same approver. Their availability is a real constraint even if each supplier has an open slot. Map the customer's time and actions required by a service (in Polish), including consultation and acceptance. If this work does not fit the calendar, reduce the parallel scope. A delayed decision usually moves more than one task and creates further versions of material.
Build the plan around assessable outcomes
A development plan should show what the company wants to achieve and how it will check stages. GOV.UK describes a roadmap as a plan of direction that expresses goals and priorities and can change with new knowledge. The principle is useful outside government too; the procedure here is our own application to a company. GOV.UK: developing a roadmap.
Name stages by outcome, not only activity. “Approved offer card and tested contact path” is clearer than “website work”. State evidence, owner and dependent task for each stage. If the outcome is not confirmed, the team should know which later work remains paused and which can continue without relying on a mistaken assumption.
For the website project, prepare your own brief. The guide to a premium website brief (in Polish) explains how to structure it. Assess the operation through an automation audit. The guide to auditing business process automation (in Polish) describes how to trace an automated process from input to outcome. Do not confuse accepting a document with confirming a change in operation. A well-prepared brief lets you commission the right work, but only a checked result shows whether the company can take the next step. Both moments belong in the plan.

Hypothetical scenario: a technical company with three ideas
Imagine a hypothetical technical company considering a new website, AI enquiry qualification and entry into Korea. This is an invented scenario, not a real client result. A review shows that salespeople describe the core product's scope differently and enquiries have no common categories. Launching all three projects at once would reproduce those differences in several places.
The company first agrees the offer and traces the handling of several cases. In parallel, it conducts limited segment research in Korea. Only after those results does it choose website content and enquiry-routing rules. The team sets criteria for its own AI trial. The team does not assume technology will make missing commercial decisions by itself.
For ideas on choosing trial criteria and a reference result, see the AI pilot guide (in Polish).
If segment research challenges the product's use, the company can change localisation scope before producing all materials. If the handling process does not need AI, it can improve it more simply. In both cases, the earlier stage provides useful knowledge. The plan's value is the ability to change direction based on evidence, not an obligation to finish every idea written down at the first meeting.
Set review points and the right to stop
Review should follow an important result, not merely a calendar date. The team checks whether the assumption was confirmed and dependencies remain current. It chooses to continue, adjust scope or stop. If a decision needs more data, record the specific gap and how to fill it. A general “we need more analysis” does not close the conversation.
Keep reasons in the project decision register. The reason for postponing an initiative matters as much as approval to start. A new team member can see which assumptions would bring an idea back for consideration. The whole discussion need not be repeated merely because someone lacks last quarter's context.
The project decision register guide (in Polish) suggests recording the reason for a choice and when to revisit it.
After a stage, check the maintenance load. A new site, database and automation need owners. If the team is already occupied maintaining existing results, another project can make operations worse. Consider simplifying, ending an unnecessary process or changing scope. Planning also includes deliberately reducing what the company must continually maintain.
Check sequence against your team’s actual availability. The guide to project priorities with a small team (in Polish) discusses how to plan around the availability of people needed across projects. Two independent projects may still need the same approver. Putting both on a calendar does not create another copy of that person, even when the spreadsheet stubbornly displays green cells.
How to run a workshop to set the first sequence
Invite people who know the problem, make the decision and will operate the result. You do not need the whole company, but the absence of a process owner makes feasibility hard to assess. Prepare brief proposal cards and available evidence. At the meeting, first confirm the shared goal, then missing assumptions, and only later discuss project and supplier names.
For each idea ask: what must be ready the day before launch? Record answers as testable conditions. Break “sales materials” into approved scope, language version and a person handling contact. “Integration” may require access, input data and an exception rule. This exposes real dependencies often hidden beneath one start date.
Then check which conditions several initiatives share. Clarifying the offer can unblock the website and localisation. Defining enquiry categories helps manual handling and later automation. Preparatory work need not have a spectacular name to be a good first step. Its value lies in the important decisions made possible after the result is accepted.
Separate matters urgent because of an external deadline from those merely voiced most loudly. If an event has a fixed date, check the minimum scope needed. Perhaps one offer and manual handling are enough instead of an entire portal. Reducing scope must preserve the goal and usefulness condition. Simply deleting difficult elements without reassessing the process can deprive it of meaning.
Finally, choose the first action and state the result expected at review. Give other proposals a status and a condition for returning to them rather than leaving them in unclear waiting. Send participants a short version of the plan and ask owners to confirm scope. A plan is useful when responsible people understand the next task, not merely agree with the company's general direction.
If the planned changes involve AI tools, see Aeneia Audit and discuss your needs and the possible scope of a review of their use and information flows.
Aeneia Audit Bring the planned changes and the problem they should address. Discuss your needs and the possible scope of a review of AI tools, current work and information flows. Discuss the scope of an AI review







