Impact Effort Matrix 2026: How to Build and Use One

What an impact effort matrix sorts

The grid sorts a closed list of ideas that already exist. It produces an order, and twelve ideas in the order they were shouted out carry no order at all. Two axes impose one in about fifteen seconds, which is the cheapest ranking device available under time pressure.

Its place in a case is after generation. You brainstorm, you end with a list, and the list needs a front. The grid gives the front and the argument for it in the same move.

Firms publish the step that calls for it. Deloitte's five step approach to a case ends on the consequence: its fifth step is to "Outline next steps and expected results/impacts", checked 30 September 2026 at deloitte.com. Expected impacts are a quantity, and a quantity per idea is one of the two axes.

Oliver Wyman's published six step method includes using numbers as a step of its own, and instructs the candidate to show why an approach beats the alternatives, checked 30 September 2026 at oliverwyman.com. Beating the alternatives is a comparison across a list, which is what the grid performs.

The matrix has no established author in our files. We looked for an originating publication on a primary domain and found none, so this page attributes it to nobody. The quadrant names below are the names in common use, and the dividing lines in the worked example are ours.

The two axes, and how to score each one

Impact is the annual value an idea returns. Effort is what the idea consumes: one time dollars, weeks of delivery, or both. Score each axis one of two ways.

The first way is ordinal. Impact is high or low, effort is high or low, or each axis runs from one to five. Scoring takes seconds and the result divides into nothing. Two ideas worth 5,500,000 dollars and 1,320,000 dollars a year both read as low against a six million dollar line, and they differ by a factor of 4.17.

The second way is a unit. Impact becomes dollars of annual profit. Effort becomes dollars of one time cost and weeks to deliver. A unit on both axes makes return per dollar available, and return per dollar is the number that ranks a list.

Published cases carry effort in units already. Oliver Wyman publishes a supermarket pharmacy case whose client "is tight on capital and requires a payback period of two years or less", checked 30 September 2026 at oliverwyman.com. The case publishes 10,000 monthly store visitors, a 66 percent prescription share of the US population, a 100 dollar average monthly drug bill, a 25 percent margin and a 4,000,000 dollar initial investment.

Run it. 10,000 visitors at 66 percent is 6,600 pharmacy customers a month. At 100 dollars each that is 660,000 dollars of monthly revenue, and 25 percent of that is 165,000 dollars of monthly profit. Divide 4,000,000 by 165,000 and payback is 24.2 months. The case concludes that the investment "just falls within the desired payback period window", and the division returns 24.2 months against a limit of 24. Both readings stand on the figures the firm publishes.

The four quadrants of the impact effort matrix

The two dividing lines cut the grid into four quadrants, and each quadrant carries a name and a standing instruction.

Quick Wins and Thankless Tasks are the two easy calls. The real argument happens between Major Projects and Fill-ins, because that pair asks whether the organization has the capital and the patience for a big move or the discipline to bank small ones.

Two facts about the quadrants decide how much to trust them. The first is that nothing fixes the two dividing lines. No publication we found on a primary domain sets the impact threshold or the effort threshold, so every drawn matrix carries lines its author chose, and the author usually leaves them unstated.

The second is that a quadrant is a pair of binary judgments. Two ideas inside one quadrant are ranked equal by the picture, however far apart their numbers sit. The worked example below puts a 5,500,000 dollar idea and a 1,320,000 dollar idea in the same box.

An impact effort matrix example, priced in dollars

Deloitte publishes a retail case with enough figures to price several ideas, checked 30 September 2026 at deloitte.com. The client has 11 billion dollars of revenue, 5 percent of it online, 200 owned stores and 20 million annual website visitors.

Start with the sanity check the case makes possible. Online revenue is 5 percent of 11 billion dollars, which is 550 million dollars, spread over 20 million visitors. That is 27.50 dollars of revenue per visitor. The case's own one dollar uplift per visitor is 1 over 27.50, a lift of 3.64 percent, which is a credible number rather than a heroic one.

Idea 1 is the firm's own. The case states that improving the website costs 10 million dollars once, that each visitor would spend an additional dollar, that the margin on website purchases is 50 percent, and that 20 million people visit the site annually. So 20 million visitors at one dollar is 20 million dollars of incremental revenue, and 50 percent of that is 10 million dollars of profit. Against a 10 million dollar one time cost the return per dollar is exactly 1.00 and payback lands at twelve months.

Ideas 2 to 4 are ours, with every assumption stated. Idea 2: one percentage point on 550 million dollars of online revenue is 5,500,000 dollars, at an assumed cost of 900,000 dollars. Idea 3: moving returns from 18 percent to 16 percent frees 2 percent of 550 million dollars, which is 11,000,000 dollars of returned value, and at an assumed 12 cents of processing cost per dollar returned that saves 1,320,000 dollars. Idea 4: 200 stores at an assumed 22,000 dollars of fit out is 4,400,000 dollars, against an assumed 6 million store shipped orders saving 1.90 dollars each, which is 11,400,000 dollars a year.

Where the four ideas land on the grid

Plotting needs two lines, and here are ours. Impact splits at six million dollars of annual profit, a round number inside the range the four ideas cover. Effort splits at 25 weeks, the midpoint of the 14 to 36 week spread. Both lines are our choices and both are stated so a reader can move them.

Four ideas fill two quadrants and leave two empty. Quick Wins has nothing in it and Thankless Tasks has nothing in it, which is the ordinary outcome when real figures replace adjectives. A drawn matrix with one token in each corner is a diagram of the template.

The emptiness also tells you the lines did the work. Drop the impact line from six million dollars to five million and idea 2 crosses it, landing in Quick Wins with the best return per dollar of the four. One threshold moved by a million dollars promotes an idea from a box labeled low value to the box everyone is told to do first.

Idea 2 and idea 3 share the Fill-ins box at these lines. Their annual profits differ by a factor of 4.17 and their returns per dollar by a factor of 1.85. The grid ranks them equal, so a candidate who stops at the picture hands the interviewer a tie that the numbers already broke.

Why both axes carry units

Units turn the grid into arithmetic, and arithmetic produces an order the picture cannot. Rank the same four ideas twice and both orders are defensible.

The two lists run almost backwards. The best idea by return per dollar sits third by absolute profit, and the largest absolute profit sits third by return per dollar. Idea 1, the one the case puts in front of the candidate, is last on one list and second on the other.

Choosing between the two orders is a question about the client, and it is the question worth asking out loud. A capital constrained owner ranks by return per dollar. An owner with money and a five year growth target ranks by absolute profit. The case says the client is pursuing "an aggressive +40% annual growth in direct-to-consumer digital channels over the next five years", which points at the second.

Then add the constraint every real prioritization carries. Say five million dollars of capital this year.

By hand: idea 4 plus idea 3 costs 4,400,000 plus 400,000, which is 4,800,000 dollars, inside the limit, and returns 11,400,000 plus 1,320,000, which is 12,720,000 dollars. Idea 4 plus idea 2 costs 5,300,000 dollars, over the limit by 300,000. The best affordable set is ideas 4 and 3, and it excludes idea 2, the idea with the best return per dollar of all four, because 300,000 dollars of capital stands between it and the shelf. One column and one subtraction found that answer. No quadrant drawing could.

How to build an impact effort matrix in six steps

Six steps, in this order, and the first one happens before the grid exists.

1. Generate the list first, without the grid. The matrix sorts and has no method for finding anything, so using it as a structure produces four boxes and no ideas.
2. Price each idea in annual profit, using figures already on the table. Then run one sanity check on the biggest number, the way the 27.50 dollars per visitor check above tests a one dollar uplift.
3. Cost each idea twice, in one time dollars and in weeks to deliver. Weeks matter when the binding limit is a team rather than a budget.
4. Set the two dividing lines out loud, and say where each came from. A round number inside the range for impact, the midpoint of the spread for effort, and a sentence naming both.
5. Plot, then read the quadrants in order. Quick Wins go first, Major Projects get staged and funded one at a time, Fill-ins wait for idle capacity, Thankless Tasks get dropped with a reason.
6. State the constraint and pick the best affordable set. Then say what you would drop first if the constraint tightened by a third, which is the sentence that shows the ranking was yours.

Steps 2 and 3 are where the exercise lives. A grid built on adjectives can be drawn in fifteen seconds and defended against nothing. A grid built on two priced columns survives the follow up question, because every box has a number behind it and the number came from the case.

What the grid does not do

Used as a case structure the matrix fails, because it contains no branches at all. It sorts, and it has nothing to sort until the analysis is done.