Before you shortlist anyone

Most failed Odoo projects did not fail because the partner was bad. They failed because the client was not ready to be a client, and nobody said so at the start.

This stage costs you nothing and removes most of that risk. Do it before you speak to anyone.

Today — in most companies 7 steps

1. Sales quote from a price list last updated in March.
2. The warehouse keeps its real stock figure in a notebook, because the system is wrong.
3. Purchasing buys from memory of what ran out last time.
4. Finance types every invoice a second time into the accounting program.
5. Three departments each keep their own Excel version of the truth.
6. The monthly figures arrive on the twentieth of the following month.
7. Decisions get made from numbers everybody privately distrusts.

The cost is not the software you do not have. It is the re-typing, the disputes and the decisions taken late.

After — on one system 4 steps

1. A price, a stock figure and a balance are entered once, by the person closest to them.
2. Departments read the same record instead of their own copy of it.
3. The monthly figures arrive in the first week, and they reconcile.
4. Arguments are about what to do next, not about whose number is right.

What you are buying is one version of the truth. Everything else in a proposal is detail.

The four answers every partner will want

If you cannot answer these, you are not ready to buy — and any price you are quoted is a guess.

Question	Why it decides the project
Who is the sponsor?	The person senior enough to change how the company works, not just to approve an invoice. Projects without one stall the first time a department refuses to change.
Who decides?	One named person who can settle a disagreement between departments in days. A committee is not an answer.
What can you spend?	A range, not a number. Partners cannot design without one, and refusing to say it does not get you a lower price — it gets you a defensive one.
What is the real deadline, and why?	“June” invites agreement. “Before the October audit, because otherwise we run two systems through the close” gets an honest answer about whether it is possible.

Who does what

You are not buying a system that arrives finished. Roughly a third of the work is yours, and it needs named people with real time.

Role	Who it is	Time it actually takes
Sponsor	Owner, MD or a C-level	A few hours a month, plus every escalation
Decision owner	Head of operations or finance	A day a week for the whole project
Process owners	The best person in each area	Two days a week during design and testing
Data owner	Someone who knows your master data	Heavy during migration
Testers	The people who do the work daily	Two to three days a week during UAT

The most common failure is naming your busiest people and giving them no relief from their normal work. If nobody can be released, the project will take longer — decide that now, not in month four.

Get your data looked at

Your item list, customer list and opening balances will move into the new system. If they are wrong now, they will be wrong there, and cleaning them during migration is the most expensive time to do it.

Before you shortlist, have someone answer:

• How many products, customers and suppliers do we actually have, and how many are dead records?
• Do the same item or customer exist twice under different names?
• Are opening balances agreed with your accountant?
• How much transaction history do you genuinely need to carry over? Usually far less than people first say.

Cleaning this while partners are still quoting costs you nothing and lowers every price you receive.

Decide what success means

Write three to five outcomes, in numbers, before anyone shows you software. They keep the project honest and give you something to measure at the end.

Good ones look like this:

• Monthly close finished within five working days, instead of three weeks.
• Stock accuracy above 97% at every branch, checked by count.
• Quotation to confirmed order in under 24 hours.
• One place where margin per shipment can be seen without anyone building a spreadsheet.

Vague ones — “better visibility”, “more efficiency” — cannot be delivered or refused, which means they will not be.

What to do about the software itself

Two useful hours: open a free trial, and let your own people use it for a week on their own work. No salesperson, no demo script.

You are not evaluating features. You are finding out whether your team can live in it, and which of your processes are obviously different from how the system expects to work. Those differences are the real content of the project.

Readiness checklist

Copy this and fill it in. When every line has a name or a number, you are ready to approach partners.

SPONSOR ................. name, role
DECISION OWNER .......... name, role, hours per week available
PROCESS OWNERS .......... name per area (sales, purchase, stock, finance, ...)
DATA OWNER .............. name

BUDGET RANGE ............ implementation from ...... to ......
                          licences per year ......
                          internal cost accepted (people's time) ......

DEADLINE ................ date ......
                          the reason it is that date ......
                          what happens if it slips ......

SUCCESS MEASURES ........ 1. ......
                          2. ......
                          3. ......

SCOPE INTENT ............ areas in this project ......
                          areas explicitly for later ......

DATA .................... records to migrate, cleaned by ......
                          history to carry: ...... months
                          opening balances agreed with ......

Send the finished version with your requirements document. Partners who read it will price more accurately, and the ones who ignore it have told you something useful.