One of my favourite parts of a GAP analysis is seeing how a business works now.
A spreadsheet fills a gap. A familiar product code carries knowledge that nobody has written down. An extra check follows a mistake. Somebody knows which colleague to ask when the record on the screen does not quite explain what has happened.
Years of everyday decisions leave their mark. I enjoy getting into that detail and beginning to see how the pieces might fit together more coherently.
There is a creative satisfaction in it. There is also a commercial responsibility: the structure we propose will affect how people work, what the business can promise and what it costs to keep the operation functioning.
An Odoo GAP analysis compares how a business operates, what it needs to achieve and how well Odoo can support those needs. It should establish where standard functionality and configuration fit, where information or working practices need attention, and which remaining requirements may justify integration or custom development.
For me, the useful result is a connected recommendation: what should change, why, what supports that conclusion and what still needs to be decided before the business commits.
When should you commission a GAP analysis?
A GAP analysis is useful when a consequential systems decision needs a firmer basis.
That might be before a first Odoo implementation, when an existing implementation is struggling, or when growth has made the connections between several systems difficult to manage.
The warning signs are often familiar:
- Important work depends on somebody knowing how everything fits together.
- Departments give different answers to apparently simple questions.
- A project has a long feature list but unclear priorities.
- Nobody is confident about the meaning or condition of the data to be migrated.
- Custom development is being proposed before the requirement is understood.
- The business cannot explain what would make the implementation successful.
The assessment needs an agreed boundary. Reviewing one difficult process and assessing a business across sales, purchasing, manufacturing and finance are different engagements.
Depth should follow the consequences of the decision.
I follow the commitment through the business
I want to understand what the business is trying to accomplish before deciding how the software should support it.
An enquiry becomes a quotation. A quotation becomes a promise. That promise creates obligations for purchasing, stock, operations, delivery and finance.
Looking at each department separately can obscure those relationships. Sales may have a useful view of an order while the person responsible for fulfilling it has to reconstruct the important detail elsewhere.
I therefore follow representative work across the boundaries.
What has been promised? What evidence supports the promise? Who receives the next task? What do they still have to find out? What happens if the original plan becomes impossible?
This changes the nature of the assessment. We can examine whether the proposed arrangement supports the whole commitment, including the handovers and exceptions.
My experience of running businesses matters here. A system’s usefulness includes what happens when somebody is absent, a supplier disappoints or the customer needs a reasonable answer before every uncertainty has been resolved.
The operation still has to function.
I look for what the apparent inefficiency is protecting
A complicated current state is not evidence that people have been doing things badly.
A manual check may prevent an expensive mistake. An unusual sequence may allow work to continue while something else is delayed. A local record may preserve information the main system has never accommodated.
Some arrangements will deserve to change. Others protect a capability the business would miss.
I want to understand the contribution before removing the mechanism.
That includes recognising where specialist knowledge gives the business its edge. A standard process should make that knowledge easier to use and sustain. It should not flatten a useful distinction merely because the distinction is inconvenient to configure.
This is also where I have to challenge my own enthusiasm. Solving the puzzle can be satisfying. The proposed arrangement still needs to justify its effect on somebody else’s working day.
In Entropy in Business, I explored how individually reasonable decisions accumulate into something harder to maintain. A GAP analysis gives us an opportunity to examine those decisions together and choose deliberately what should survive.
I compare what people say with what the records support
Workshops are essential. They are one source of understanding.
I also examine representative documents, reports, spreadsheets and relevant data. Where appropriate, that includes database extracts and documentation explaining the existing system.
These sources answer different questions.
A colleague can explain why a process works as it does. An order can show what was recorded. A data investigation can reveal that two apparently similar records carry different meanings.
I want those accounts to meet.
If somebody says that a price updates automatically, which price do they mean? A new quotation, an existing quotation, the product’s current selling price or a historic transaction?
If two departments disagree about stock accuracy, are they measuring the same thing?
The disagreement may reveal a difference in timing, definitions or responsibilities. Choosing the most confident account would conceal that distinction.
This matters because an elegant recommendation can be built on an incorrect assumption. Cross-checking the sources helps expose those assumptions while they can still change the design.
It also lets me preserve what the evidence does support without pretending every uncertainty has been resolved.
I separate the requirement from the requested solution
A request often contains both a genuine problem and an initial design.
“We need a dashboard” may mean somebody cannot make a dependable promise. “We need an automated email” may mean customers are not being told when that promise changes.
I want to retain the problem while testing the proposed response.
Consider a hypothetical business where salespeople ask for a stock-availability dashboard.
“Available” could mean physically present, uncommitted to another order, expected from a supplier or possible to manufacture from components.
Those meanings lead to different answers.
I would follow a representative order and establish what the salesperson needs to decide. Then I would examine the information supporting that decision: reservations, incoming dates, existing commitments and the treatment of shortages.
Only then can we sensibly assess the display.
A dashboard will not resolve commitments that nobody records. Nor can it decide who deserves the last unit unless the business has established an allocation policy.
The recommendation might combine standard configuration, better recording, an agreed decision route and a small additional view. It might require less development than the original request suggested. It might reveal a justified need for more.
Either conclusion is useful when we can explain it.
The requirement becomes something we can assess: an authorised salesperson can establish what can be promised, understand the basis for the answer and follow a clear route when an exception needs a decision.
That gives implementation and testing a much firmer starting point than “build a dashboard”.
I investigate the uncertainty that could change the recommendation
I prefer to begin with standard Odoo functionality and configuration.
That preference needs scrutiny too. A feature name is not enough to establish that a business requirement is met.
A system might record the result of a check. The business may need that result to prevent a particular action until an authorised person resolves it. Recording and controlling are different obligations.
Where the distinction matters, I use focused investigations to establish the relevant behaviour.
The question needs to be specific enough to answer. For example: after a price change, do existing quotation terms remain intact while new quotations use the revised price?
A bounded test can provide useful evidence about that mechanism. Its conclusion should remain tied to what was demonstrated.
I keep documented capability, observed behaviour, proposed design and business decisions distinguishable. That makes the remaining work clearer.
It avoids two expensive assumptions: that the standard system must cover the requirement because a related feature exists, or that an extension is necessary because the fit has not yet been investigated.
A genuine residual need can then be described precisely, with a clearer basis for estimating, testing and maintaining it.
I treat history as something people need to use
Data migration decisions shape the eventual operation.
It is possible to move records successfully and still leave people unable to answer the questions that matter.
A customer calling about an old transaction may need more than an invoice total. The person helping them may need the original description, the relevant product configuration and the history of any changes.
A current product record cannot automatically explain what was supplied years earlier.
I therefore ask what the business needs to do with its history. Find a transaction? Investigate a complaint? Establish an outstanding balance? Support a replacement or credit decision?
Those are practical requirements for the migrated system.
They influence what must be preserved, how records need to connect and what must be checked. Counts and totals can be useful controls, but they do not settle every question about meaning or usability.
There is also a limit to what any migration can recover. If the original operation never recorded a change, we should not manufacture a precise history to make the new system look complete.
The stronger outcome is usable information with understood limits.
I assess the transition as well as the destination
A proposed future state can look convincing while leaving a difficult question unanswered: how does the business get there while continuing to operate?
Orders arrive during implementation. Product information changes. People make decisions. The team may also improve data and configuration as it prepares the new system.
Those changes need a deliberate treatment.
Otherwise, the final move risks losing work completed during implementation, replacing a useful correction with an older value, or leaving uncertainty about which record is authoritative.
I want the analysis to identify the transition obligations early enough to influence scope.
What needs to be rehearsed? Which records and balances must be checked? Who can decide that the result is adequate? What happens if a necessary condition is not met?
Training belongs in the same conversation. Attendance tells us somebody joined a session. A representative exercise tells us more about whether they can carry out the work.
These are part of making the proposed operating model usable.
I make scope choices with their consequences attached
Once the findings begin to make sense individually, I examine them together.
Product definitions influence quotations, procurement and manufacturing. Customer identity affects permissions, correspondence and finance. Several visible requirements may depend on the same underlying capability.
That shared work should be recognised once, with the dependent outcomes made clear.
It also affects phasing. A smaller first release can be sensible, but it needs to complete useful work. Removing a prerequisite can leave a collection of apparently finished features that cannot support an ordinary transaction together.
I want the scope discussion to make those consequences visible.
What becomes possible with this selection? What remains manual? Which decisions or data must be ready? What is deferred alongside the visible feature?
This gives the business a more meaningful way to adjust investment. It can choose a coherent outcome and understand the compromises.
What should you prepare?
You do not need perfect documentation before starting. Establishing what is missing is part of the assessment.
The most useful preparation is:
- Your reasons for considering change: the problems, risks and outcomes that matter.
- Access to people who know the work: including those handling exceptions and covering for absent colleagues.
- Representative examples: ordinary transactions, difficult cases and things that already work well.
- Relevant materials: forms, spreadsheets, reports, process notes and system documentation.
- An account of your systems and data: where information sits and what access can safely be provided.
- Known constraints: budget expectations, deadlines, contracts and activities that must continue.
Access should be proportionate. A selected sample or walkthrough may answer a question without requiring unrestricted access to sensitive information.
What should a useful GAP analysis leave you with?
I want the outputs to support decisions at two levels.
Business owners need a readable recommendation: the material findings, proposed approach, priorities, consequences and choices.
The people responsible for delivery need enough supporting detail to understand the required outcomes, their dependencies and how adequacy will be demonstrated.
For the agreed scope, the assessment should establish:
| Output | Practical value |
|---|---|
| Current-state findings and strengths | Shows what needs attention and what should be preserved |
| Clear required outcomes | Gives the project a basis beyond a feature list |
| Reasoned solution treatments | Explains where configuration, process change, training, integration or development fits |
| Data and transition requirements | Makes migration and operational continuity part of the scope discussion |
| Open questions and decisions | Gives unresolved matters an owner and a consequence |
| Connected priorities | Supports a viable first release and deliberate later phases |
| Validation expectations | Makes clear what the eventual implementation needs to demonstrate |
Recommendations should remain traceable to their basis. The report should also distinguish what is known from what is proposed, conditional or awaiting agreement.
That makes the scope discussion more honest and the resulting commitment more understandable.
How does this strengthen the implementation?
The immediate benefit is a better basis for decisions.
A proposed extension has a clearer justification. A migration has a purpose beyond moving data. A first release has an operational shape. An unresolved policy remains visible before somebody accidentally implements an answer.
This also improves the basis for estimating. The work includes the obligations needed to achieve the outcome, including material exceptions, validation, training and transition.
It cannot remove every unknown before implementation. It can make the important uncertainties explicit enough to investigate, allow for or decide.
Later results still need measuring. Claims about saved time, improved service or reduced rework should be tested against an agreed starting position.
The analysis earns its place by giving that later work a more dependable foundation.
That is what interests me about gathering the fundamentals. I enjoy seeing an operation in its accumulated detail, understanding what people have made it do and beginning to recognise how the pieces could support one another more coherently.
The result should leave the business better able to judge the proposal, choose its investment and recognise whether the promised outcome has been delivered.
If you are considering Odoo, or trying to understand why an existing implementation is difficult to operate, you can explore Polyphony’s approach to Odoo implementation. You do not need a finished feature list to begin the conversation.