A 90-Day Business Efficiency Plan: What to Do at Each Stage and What to Measure

Most efficiency programmes fail for the same reason: they start with a solution. A plan that works starts with two weeks of measurement, picks exactly one process, and measures again using the same method - so you can tell whether anything genuinely changed.

The structure: three 30-day cycles

This beats one long project, because it produces a proven result within two months - which is what keeps the team willing to run the next cycle.

Days 1-14: measure and change nothing

The temptation to fix immediately is strong, and worth resisting. In these two weeks you record four simple things for each main process:

1. How often it happens per week.
2. How long it takes each time, roughly.
3. How many people touch it.
4. How often something goes wrong, and what exactly.

The record does not have to be precise, it has to be consistent. One spreadsheet with four columns is enough, and what matters is that the same person records it the same way across the two weeks.

Days 15-30: choose one process

The choice comes from a simple multiplication: frequency times duration. The process that scores highest is usually not the one that feels most irritating - it is something small that happens twenty times a week.

It is also worth weighting a third column: how many errors it produces. A process that takes little time but generates a weekly error reaching a customer costs more than the clock shows. The full calculation method is in what manual work costs.

Days 31-60: change one process

The change starts by mapping the process as it genuinely runs rather than as written down. Then come four questions, in this order:

• What can be eliminated? The cheapest, and the least examined.
• What can be merged with something else happening anyway.
• What can be simplified - fewer fields, fewer approvals, fewer handoffs.
• And whatever remains - what is worth automating.

The order matters: automating an unnecessary process is the most expensive way to preserve it, which is exactly what automating the wrong things describes.

Days 61-90: measure again and choose the next

Re-measurement must use the same method over the same duration. If you measured for two weeks, measure for two weeks; if you counted in a spreadsheet, count in the same spreadsheet. Comparing a careful measurement with a sloppy one produces an "improvement" that does not exist.

What to check: whether frequency fell, whether time per occurrence fell, and whether errors fell. At least one of the three should move clearly. When none does, the change did not work - which is a useful conclusion, provided you admit it rather than carrying on regardless.

What sinks plans like this

• Changing several processes at once, then not knowing what worked.
• Skipping the first measurement and relying on impression.
• Picking the most irritating process rather than the most expensive one.
• Not assigning an owner to the change itself.
• Buying a tool before the process is understood.

The last is most common in businesses with a budget, and it is also the most expensive: a new tool requires implementation, and implementing on a broken process cements it.

What happens after 90 days

You run another cycle with the next process on the list. The advantage by then is that measurement already exists and the team knows how it works, so the second cycle is shorter than the first.

It is also worth documenting what was done - what changed, why, and the numbers before and after. It takes half an hour and serves three purposes: an explanation for a new employee, a basis for the next decision, and evidence when someone asks whether it was worth it. The wider frame for all of this is in how to improve business efficiency.

What happens when the change does not stick?

It happens, and it is usually not a failure of the change but of its ownership. A new process holds as long as somebody is watching it; when nobody is, the team gradually returns to the old way - not in rebellion but because it is familiar.

Three things make a change stick: that it is easier than the old way, not merely more correct; that somebody checks in the first fortnight whether it is genuinely happening; and that it is written down somewhere you can show a new employee. When one of those three is missing, the change will probably fade within a month or two.

The early sign is specific: somebody is still maintaining the old version "just in case" - a file, a notebook, a side process. While it exists, the team has not really moved, and it is worth understanding why it still feels necessary before forcing it away.

What not to include in the plan

• Changes that depend on an outside party - a supplier, an authority, a large customer. They stretch without your control.
• System projects - replacing an ERP or CRM is not a 90-day plan.
• Organisational restructuring - it changes the very processes you measured.
• More than three cycles - better to finish three well than to start six.

The second matters especially: a decision about a new system can absolutely come out of the plan, but implementing it is a separate project with its own timeline - and confusing the two is what turns a short plan into a year of work.

Sources

• Israel Tax Authority - information for businesses
• How to improve business efficiency