How to read this

Predicting technology five years out is unreliable. Predicting what it means for a business function is a little more stable, because the underlying jobs change slowly.

So this article goes function by function — what the work looks like now, what it plausibly looks like later, and what is worth doing this year regardless.

The final section is the useful one.

Accounting

Now. Documents read into drafts, a person checks and posts. Reconciliation matches proposed, a person handles the exceptions. Month-end close is manual.

Likely. More of the routine close prepared automatically — checks run, missing documents chased, exceptions assembled. The accountant reviews a prepared package rather than building it.

Unlikely to change. Somebody signs the accounts and must be able to explain them. Provisions, accruals and estimates stay judgements. The close stays a review by a person.

The shift: from producing the numbers to interrogating them.

Sales

Now. Summaries before calls, drafted follow-ups, prompts for the next activity, prioritised pipelines where there is enough history.

Likely. More preparation — a pack before each meeting with the history, the open items, the pricing position and what changed since last time. Better prioritisation as data accumulates.

Unlikely to change. The conversation. Negotiation, reading a room, knowing when a deal is not real. Stage changes should stay with the salesperson who knows what happened.

The shift: less admin around selling, roughly the same selling.

Purchasing and inventory

Now. Documents read, three-way matching made practical, forecast-based reordering where stock records are accurate.

Likely. Reorder proposals prepared across the whole range with supplier reliability factored in, presented for weekly review rather than decided item by item.

Unlikely to change. Someone counts the receipt. Everything downstream depends on that one human action, and no forecasting improvement substitutes for it.

The shift: from deciding each order to reviewing a prepared list.

Support

Now. Context summarised for the agent, replies drafted, tickets routed, articles suggested.

Likely. More routine resolution handled end to end, with narrow scope and a fast route to a person. Better prediction of which customers are about to have a problem.

Unlikely to change. The difficult conversation. The customer who is upset and needs a person who can decide something.

The shift: agents handle more, with better preparation, and the hard cases still land with people.

Manufacturing

Now. Scheduling assistance, component forecasting, document reading, consumption variance flagging.

Likely. Better scheduling as operation data accumulates. Predictive maintenance where equipment is instrumented — which is a separate investment, not a software upgrade.

Unlikely to change. Bills of materials still have to be right. A wrong header quantity against the product’s unit of measure will produce wrong numbers quietly, whatever else improves.

The shift: planning gets better, the shop floor changes only where equipment does.

Management

Now. Questions asked in plain language, faster answers, analysis assembled rather than commissioned.

Likely. Analysis arriving before the meeting rather than being built during it. Changes flagged as they happen rather than discovered at month-end.

Unlikely to change. Deciding what matters. A number tells you margin fell. Knowing whether that was a deliberate strategic order or a problem is understanding, and it is not in the data.

The shift: from assembling numbers to discussing them.

What will not change anywhere

Five things, and this is the part worth acting on.

Bad data produces bad answers. More capable systems reach wrong conclusions faster and state them more convincingly. This gets worse, not better.

Unclear processes stay unclear. If nobody agrees who approves a discount, no technology settles it.

Someone is accountable. For the accounts, to the board, to the customer.

Judgement about people stays with people. Credit, relationships, difficult accounts.

Implementation decides the outcome. The gap between businesses will keep coming from how carefully the system was set up, not which features were bought.

The gap that widens

The prediction worth making, because it is about behaviour rather than technology.

Two companies buy the same system with the same features.

One has clean master data, agreed definitions and a clear process. The other has duplicate customers, product costs three years out of date, and three people who each define revenue differently.

The first gets useful answers. The second gets confident nonsense, faster than before.

That gap exists today. Every capability added widens it, because every capability is built on the same data.

What to do this year

Five things. None of them depend on predicting anything correctly.

1. Fix your master data. Duplicates merged, costs correct, units consistent. A week’s work in a small company, and it improves everything immediately.

2. Agree your definitions. What revenue means, when a sale counts, what is excluded. Write it down and share it.

3. Keep customisation in proper modules. Whatever arrives will assume a standard structure. Edited core files make every future change harder.

4. Adopt what is proven. Document reading pays now. Measure the hours before and after.

5. Decide where humans stay and write it down. Anything financial, anything customer-facing.

All five are useful today. None of them are bets on a prediction. That is what makes them the right preparation.

The realistic picture

In five years, a well-run business on Odoo probably looks like this.

Most documents arrive as drafts rather than as typing. Most routine work is prepared and reviewed rather than assembled. Questions get answered in minutes. Attention is better ordered.

People spend more of the week on exceptions, judgement and relationships — which is what they were hired for.

Not a transformation. A steady removal of the parts of the job nobody valued.

That is worth having, and it is closer than the marketing suggests in some places and further away in others.

The short version

Every function follows the same pattern: less assembling, more deciding.

What does not change is what actually determines your outcome — data quality, process clarity, and someone accountable for the numbers.

The best preparation for whatever arrives is the work that helps you today: clean your data, agree your definitions, keep your customisation clean, and adopt what is already proven.