The framework · 3W Factory

Where to start with AI:
prioritizing your levers.

The question is not "which AI tool to adopt", but "which lever to install first". Too many initiatives start from the flashiest toy and die for lack of payoff. The right entry point is chosen on three criteria.

Impact × effort × adoption maturity. In that order.

In short — To prioritize your AI levers, you assess each initiative on three criteria: impact (how much margin or time it recovers), effort (data, integration, complexity) and adoption maturity (is the team ready to change the way it works). You always start with the lever that is high impact × low effort × high maturity: a measurable quick win on a single function, one that proves the value before you extend. You avoid the trap of the visible-but-unadopted lever — the one that will never be used. The rule: enter through proof, not through ambition.


Why do most AI projects
start badly?

The common reflex: start with the most visible use case — an assistant, a generator, the topic everyone is talking about. It is spectacular in a demo, and invisible six months later. The lever was chosen on its hype, not on its value.

The other, symmetrical mistake: aiming too big. You launch the project that "would transform everything", before having proven anything. It burns budget for months and ends up as a report. In both cases, the problem is the same: the priority order was set on instinct, not with a grid.

The right question is not "which tool".
It is which lever, first.

On which 3 criteria do you prioritize an AI lever?

Three axes, assessed for each initiative. It is their combination — not any single one — that names where to enter.

Criterion 01 · Impact

How much it recovers

The direct business value: margin recovered, time freed, incidents avoided. A high-impact lever touches a flow that truly matters — not a cosmetic detail.

The right question If this lever works, what does it change in the P&L or in the team's time?
Criterion 02 · Effort

What it costs to install

The real difficulty: data available or to be built, integration into the existing stack, complexity of the flow. Low effort means a scope where the raw material is already there.

The right question Does the data already exist, even imperfect — or does it have to be created first?
Criterion 03 · Maturity

Whether the team will adopt it

The criterion everyone forgets. A lever no one uses has a real impact of zero. Adoption maturity measures whether the team is ready to change the way it works.

The right question Who will use it every day — and accepts changing their routine for it?

Impact without adoption produces nothing.
Adoption without impact weighs nothing. You need all three.


The prioritization matrix
impact × effort

Plot each lever on two axes — impact and effort — then filter by adoption maturity. Four quadrants, one clear decision.

High impact · Low effort

Quick win — start here

The ideal entry point: a lot of value, little friction, data already there. This is the first lever to install to prove quickly, on a single function.

Verdict: install first.
High impact · High effort

Foundational project — later

The real long-term return, but heavy to install. Keep it for when a quick win has already proven the value and unlocked confidence — not as an entry point.

Verdict: after the first proof.
Low impact · Low effort

Bonus — opportunistic

Easy, but marginal. Do it in the background if the chance arises, never as a stated priority. It does not mobilize the first wave.

Verdict: opportunistic, no priority.
Low impact · High effort

Trap — to avoid

Costly for little value. Often the "impressive" project chosen for its hype. This is the quadrant that drains budgets and kills AI's credibility internally.

Verdict: do not launch.

Then filter: a quick win lever the team will not adopt becomes a trap again. Maturity breaks the ties.


How to apply the matrix in four moves

A simple method to go from a fuzzy list of ideas to a defensible order of priority.

01

List the levers

Inventory every AI initiative you are considering, framed as a business outcome — not as a tool.

02

Score impact & effort

Place each lever in a quadrant. Be honest about effort: missing data inflates it fast.

03

Filter by maturity

Discard the levers the team will not adopt. An unused quick win is not a quick win.

04

Enter through proof

Install the "quick win" quadrant lever first. Proven value unlocks the foundational projects.

Prioritizing is not launching everything in parallel.
It is choosing where to prove first.


Prioritizing your AI levers, concretely

Where should you start when bringing AI into your company?

With the lever that has the highest impact, the lowest effort and the highest adoption maturity — that is, a measurable quick win on a single function. You assess each initiative on these three criteria, place the levers in an impact × effort matrix, then filter by adoption. You install the first proven lever before tackling the foundational projects. The rule: enter through proof, not through ambition or the topic of the moment.

What is an impact × effort prioritization matrix?

It is a two-axis grid — business impact on one side, installation effort on the other — that sorts levers into four quadrants: high impact / low effort (the quick win, to do first), high impact / high effort (the foundational project, later), low impact / low effort (opportunistic bonus) and low impact / high effort (the trap to avoid). For AI, you add a third decisive filter: adoption maturity.

Why add adoption maturity to the two classic axes?

Because a lever no one uses has a real impact of zero, whatever its theoretical value. Adoption maturity measures whether the team is ready to change the way it works to use it. It is the most often forgotten criterion, yet the one that sinks the most AI projects: you install a flawless tool that no one adopts. Maturity breaks the tie between two levers equally well placed on impact and effort.

Should you aim for the most ambitious project from the start?

No. The high-impact-but-high-effort lever is a real long-term return, but a poor entry point: it burns budget for months without proving anything, and risks ending up as a report. You start with a quick win — high impact, low effort — that proves the value on your own data. That proof unlocks the confidence and the budget for the foundational projects. This is the logic of entering small, proving, then extending.

Which lever should you avoid above all?

The one in the low impact / high effort quadrant: costly for little value. It is often the most impressive project in a demo, chosen for its visibility rather than its return. This quadrant drains budgets and damages AI's credibility internally — after a costly failure, no one wants to hear about the subject anymore. The matrix exists first and foremost to rule out this trap.

How does 3W Factory help prioritize the levers?

The Audit phase of the 90-day method maps your levers and scores them by impact × effort, factoring in each team's adoption maturity. It names the quick win to enter through — the first intra-team loop to install. For a fast first reading, the Pilotage Score™ situates your priority levers in 2 minutes, with no email. We then install a capability, not a dependency.