
An impact effort matrix is a 2x2 grid that sorts work by how much value it should create (impact) and how much work it takes (effort). It gives you four quadrants: quick wins, major projects, fill-ins and thankless tasks. It is fast and visual, but it only helps when both axes are defined in measurable terms, scored against the same time horizon, and checked for dependencies before anyone treats a "quick win" as the first job.
This guide gives you that missing structure: a scoring rubric with real thresholds, a free template, a fully worked example with 12 initiatives, and a process for borderline items and re-scoring.
Download the Markdown template · Blank CSV register · Filled example CSV · Jump to the worked example
An impact effort matrix is a prioritization tool that plots each initiative on two axes. Impact is the benefit you expect if the work succeeds. Effort is the time, money and coordination it takes to deliver. Splitting each axis into high and low gives four quadrants, and each quadrant suggests a default move.
It works best when a team has more ideas than capacity and needs a shared first cut in under an hour, for example after a retrospective, a customer-feedback review or a brainstorm. If your list is still a pile of raw sticky notes, group duplicates first with an affinity diagram so you score distinct initiatives, not five versions of the same idea.
| Quadrant | Impact and effort | Default move | Watch out for |
|---|---|---|---|
| Quick wins | High impact, low effort | Candidates to do soon | Hidden prerequisites, poor strategic fit, values that only look high because costs were left out |
| Major projects | High impact, high effort | Plan, staff and sequence; test risky assumptions first | Starting more than the team can finish |
| Fill-ins | Low impact, low effort | Do only when there is slack | Comfortable busywork that crowds out real progress |
| Thankless tasks | Low impact, high effort | Drop, park or shrink the scope | Pet projects and work done because a competitor did it |
Two cautions apply to every quadrant. First, the labels describe position, not merit: an enabling task with no direct value can still be the most important thing to do this month. Second, the boundary between "high" and "low" is a choice you make in the rubric. Without one, people place items wherever their gut says, which is why the next section matters more than the grid.
Many impact effort guides stop at "rate each item from 1 to 5" or "use T-shirt sizes". That leaves the hard part open: what does a 4 mean? Two people can both rate an idea "high impact" while picturing completely different results. A usable rubric fixes four things before anyone scores.
1. A time horizon. Impact must be measured at the same point for every item. In the example below it is month 6 after launch, and the plan covers the next 26 weeks. A loyalty program that pays back in year two and a bug fix that pays back next week are not comparable without one.
2. A unit for each axis. Pick one impact currency that fits your goal: net monthly value, hours saved, tickets avoided, sign-ups. Score it net of costs: a discount that adds orders but gives away margin should be scored on what is left. For effort, use person-weeks covering design, build, QA, launch and the first month of care, plus any prerequisite that no other funded initiative pays for.
3. Bands with numbers. Convert the unit into scores with explicit ranges:
| Score | Impact (net value a month) | Effort (person-weeks) |
|---|---|---|
| 1 | Under $300 | Under 1 |
| 2 | $300 to $999 | 1 to under 3 |
| 3 (borderline) | $1,000 to $2,499 | 3 to under 6 |
| 4 | $2,500 to $4,999 | 6 to under 12 |
| 5 | $5,000 or more | 12 or more |
These bands suit the small team in the example. Rescale them to your own size. A good test is that your largest realistic initiatives land at 5 and your smallest useful ones at 1.
4. An evidence grade on every score. The score says where to plot an item. The grade says how much to trust that position.
| Grade | Meaning | Example |
|---|---|---|
| A | Measured in your own data or a pilot | Support tags show 720 order-status tickets a month |
| B | Reasoned model with checked inputs, or comparable data | Measured cart volume, assumed 3% recovery |
| C | Opinion, survey intent or competitor copying | "31% of survey respondents said they were interested" |
When a value combines measured and modeled parts, grade it by the evidence behind the larger part. Legal, security and safety obligations skip the matrix entirely: they go on the plan because they must, not because they score well.
Copy these blocks into a doc or spreadsheet, or download the files at the top of this guide. The Markdown template contains the setup fields, editable rubric, a 12-row register, the quadrant grid, a pre-commit checklist and a re-score log. The blank CSV has the same 16 register columns and opens in Excel, Google Sheets or Numbers.
Setup
| Field | Your answer |
|---|---|
| Goal this matrix serves | |
| Horizon (when impact is measured) | |
| Impact unit and bands | |
| Effort unit and bands | |
| Capacity for new work in this horizon | |
| Next re-score date |
Initiative register
| ID | Initiative | Evidence | Grade | Impact | Effort | Owner | Next action |
|---|---|---|---|---|---|---|---|
| 1 | |||||||
| 2 | |||||||
| 3 |
The downloads add columns for the outcome metric, value calculation, net value, person-weeks, dependencies, quadrant, strategic fit and re-score date.
Blank matrix
| Low effort (1-2) | High effort (4-5) | |
|---|---|---|
| High impact (4-5) | Quick wins: | Major projects: |
| Low impact (1-2) | Fill-ins: | Thankless tasks: |
Borderline (a 3 on either axis): list each item with the test or tie-break that will decide it.
Fernhill Plants is a fictional online houseplant shop, invented for this guide. Its five-person team has 26 weeks and a goal of growing gross profit while keeping support load flat. Every figure below is an illustrative assumption, not data from a real company.
Assumptions used in every calculation
Evidence and value
| # | Initiative | Evidence (grade) | Net value at month 6 |
|---|---|---|---|
| 1 | Shipping-status emails | 720 order-status tickets a month; a 2-week pilot on 10% of orders cut them 60% (A) | 720 x 60% x $6 = $2,592 |
| 2 | Fix mobile address-validation bug | 375 mobile checkouts a month hit the error; 40% abandon (A) | 150 lost orders x 80% recovered x $24 = $2,880 |
| 3 | Abandoned-cart reminder email | 1,500 carts a month with an email; 3% recovery assumed (B) | 1,500 x 3% x $24 = $1,080 |
| 4 | Plant-care subscription box | 31% of 600 survey respondents "interested"; no purchase data (C) | 300 subscribers x $18 margin = $5,400 |
| 5 | Search and care-filter rebuild | Search in 28% of sessions; 12% of searches return nothing; lift modeled (B) | 150 extra orders x $24 = $3,600 |
| 6 | Move to a new store platform | Quoted fee difference (B) | Fee savings = $400 |
| 7 | Catalog care-data cleanup | 41% of 600 products lack light, watering or pet-safety fields (A) | $0 direct; unlocks #5 and #9 |
| 8 | Loyalty points program | A competitor has one; 3 customer requests (C) | 40 orders x $24 minus $300 points cost = $660 |
| 9 | Pet-safety badges and filter | 240 pet questions a month measured; 70-order lift modeled (B) | 180 tickets x $6 + 70 orders x $24 = $2,760 |
| 10 | First-order 10% discount pop-up | 140 extra first orders, but 480 existing first orders would use the code too (B) | 140 x $24 minus 620 x $4.80 = $384 |
| 11 | Gift message at checkout | 55 emails a month ask for a gift note (B) | 55 x $6 + 8 orders x $24 = $522 |
| 12 | Saturday live-chat coverage | 22% of chats arrive on weekends and wait until Monday (B) | 30 saved orders x $24 = $720 |
Scores and quadrants
| # | Initiative | Impact | Person-weeks | Effort | Quadrant |
|---|---|---|---|---|---|
| 1 | Shipping-status emails | 4 | 2 | 2 | Quick win |
| 2 | Fix mobile address-validation bug | 4 | 0.8 | 1 | Quick win |
| 3 | Abandoned-cart reminder email | 3 | 1.5 | 2 | Borderline |
| 4 | Plant-care subscription box | 5 | 14 | 5 | Major project |
| 5 | Search and care-filter rebuild | 4 | 8 | 4 | Major project |
| 6 | Move to a new store platform | 2 | 20 | 5 | Thankless task |
| 7 | Catalog care-data cleanup | 1 | 2.6 | 2 | Fill-in |
| 8 | Loyalty points program | 2 | 7 | 4 | Thankless task |
| 9 | Pet-safety badges and filter | 4 | 0.4 (3.0 if it must fund #7) | 1 | Quick win |
| 10 | First-order 10% discount pop-up | 2 | 0.5 | 1 | Fill-in |
| 11 | Gift message at checkout | 2 | 1.2 | 2 | Fill-in |
| 12 | Saturday live-chat coverage | 2 | 4 | 3 | Borderline |
Owners and next actions
| # | Owner | Depends on | Next action |
|---|---|---|---|
| 1 | Support lead | None | Roll out to all orders in week 1; re-score in week 6 with ticket data |
| 2 | Front-end engineer | None | Fix this week; confirm the error rate falls below 1% |
| 3 | Lifecycle marketer | None | Run a 4-week holdout test; re-score with the measured recovery rate |
| 4 | Head of merchandising | None | 3-week waitlist test; build only with 150+ sign-ups |
| 5 | Product manager | #7 | Scope once #7 is done; test a filter prototype with 5 customers |
| 6 | Operations lead | None | Do not start; revisit 6 months before the platform contract renews |
| 7 | Merchandising coordinator | None | Start in week 1, because #5 and #9 depend on it |
| 8 | Lifecycle marketer | None | Park; revisit if the 90-day repeat rate stalls after #1 and #3 |
| 9 | Front-end engineer | #7 | Ship in week 3, right after #7 |
| 10 | General manager | None | Rejected: small net value and conflicts with full-price positioning; decision logged |
| 11 | Front-end engineer | None | Backlog; pick up when a sprint has slack |
| 12 | Support lead | None | Test 4 Saturdays with a volunteer rotation, then re-score |
What the plot shows that the table hides
The resulting 26-week plan
| When | Work | Person-weeks |
|---|---|---|
| Weeks 1-4 | #2 bug fix, #1 status emails, #7 catalog cleanup, #3 cart test, #4 waitlist test | 0.8 + 2 + 2.6 + 1.5 + 1 = 7.9 |
| Weeks 3-6 | #9 pet-safety badges, #12 Saturday chat test (4 Saturdays) | 0.4 + 0.6 = 1.0 |
| Weeks 5-14 | #5 search and care filters | 8 |
| Weeks 8-24 | #4 subscription box build, only if the waitlist test passes | 13 |
| Total | 29.9 of 65 |
The team commits less than half of its free capacity. That gap is deliberate: people tend to underestimate how long their own tasks will take, a pattern Buehler, Griffin and Ross documented as the "planning fallacy" (1994). Leaving room is cheaper than re-planning in week 10.
High impact for low effort is attractive, and many guides say "do these first". In practice, a quick win earns its place only after three checks.
Strategic fit. Fernhill's discount pop-up (#10) shows why. On gross numbers it looks like a quick win: 140 extra first orders x $24 = $3,360 a month for half a person-week. Net of the discount given to 620 first-order customers, it is worth $384, a fill-in. Even if the net number had held up, the company's positioning as a full-price, expert-care brand would have ruled it out. Log rejected quick wins with the reason, so the idea is not re-proposed every quarter.
Sequencing. The pet-safety badges (#9) score effort 1 only because the catalog cleanup (#7) is funded first. Ship #9 before #7 and it either fails or quietly absorbs 2.6 person-weeks of data work. Order work by dependency, not by quadrant.
Capacity and momentum. Quick wins are useful early because they free capacity (#1 removes about 430 tickets a month) and show results while major projects are still in build. But a quarter made only of quick wins never delivers the bigger changes. Pair them with at least one major project that is staffed and sequenced.
A score of 3 on either axis means "close enough to the line that the placement could flip". Forcing it into a quadrant hides that uncertainty. Use one of these moves instead.
A 2x2 grid treats every item as independent and every score as certain. Neither is true, so add two simple conventions.
Draw dependencies as arrows and count unfunded prerequisites in effort. If a prerequisite is funded for another reason, as #7 is for #5, the dependent item keeps its low effort score. If it is not, add the prerequisite's person-weeks: #9 rises from 0.4 to 3.0 and moves from effort 1 to effort 3. Delivery risks that come with dependencies, such as vendor delays or data quality, belong in a risk register; these free risk assessment templates show common formats.
Show uncertainty as a range for grade C items. The subscription box (#4) could land anywhere from 100 to 400 subscribers by month 6, so between $1,800 (100 x $18) and $7,200 (400 x $18) a month. That spans impact 3 to 5. When a range crosses a quadrant line, do not argue about the point estimate. Buy information instead: Fernhill's waitlist test costs about 1 person-week, against the 13 person-weeks the build would add.
Prepare one card per initiative beforehand. Each card needs the outcome it targets, the evidence available and a rough person-week estimate from whoever would do the work. Then run the session in this order.
| Step | Time | What happens | Output |
|---|---|---|---|
| 1. Agree the frame | 5 min | Confirm the goal, horizon, units and bands. Do not change them mid-session. | Shared rubric |
| 2. Score silently | 10 min | Each person scores every card alone, with an evidence grade. | Individual scores |
| 3. Reveal and discuss gaps | 15 min | Discuss only cards where scores differ by 2 or more points; the person with the outlier explains. | Agreed scores, open questions |
| 4. Plot and connect | 10 min | Place cards on the grid and draw dependency arrows. Mark borderline items. | Filled matrix |
| 5. Check fit and capacity | 10 min | Test each quick win against strategy and prerequisites; add up committed effort against capacity. | Candidate list |
| 6. Assign and schedule re-scores | 10 min | One owner, one next action and one re-score date per card. | Action table |
Silent scoring comes first for a reason. Estimates drift toward whatever number is said first, an effect Tversky and Kahneman described as anchoring (1974). Independent scores also make disagreement visible. A 2 next to a 5 usually means two people hold different information, and that conversation is the most valuable part of the session.
A matrix is a snapshot of what the team believed on one day. Re-score an item when:
Keep a short re-score log with the date, the old and new scores, what changed and the decision. Over a few cycles the log shows whether your team tends to overrate impact or underrate effort. That makes the next round of scoring better calibrated. Store it with the rest of the project record, following the same project documentation practices you use for other decisions, so the reasoning survives staff changes.
These tools are often confused because several are 2x2 grids. They answer different questions.
| Tool | Question | Inputs | Best for | Limitation |
|---|---|---|---|---|
| Impact effort matrix | Which initiatives give the most value for the work? | Impact vs effort | Shortlisting a backlog of projects or improvements | Two dimensions only; ignores urgency, risk and dependencies unless you add them |
| Eisenhower matrix | What should I do, schedule, delegate or drop today? | Urgency vs importance | Personal and team task triage | Says little about payoff size or effort |
| Weighted decision matrix | Which option is best across several criteria? | Weighted scores on 3 or more criteria | Choosing between a few alternatives, such as vendors or designs | Slower; weights can be argued endlessly |
| RICE scoring | Which items give the most impact per unit of effort? | (Reach x Impact x Confidence) / Effort (Intercom, 2018) | Ranking a long product backlog numerically | False precision if the inputs are guesses |
A practical sequence is to cluster ideas, shortlist them with an impact effort matrix, and use a weighted decision matrix only for the few expensive, close calls. The Eisenhower matrix then handles the day-to-day tasks inside each chosen initiative. Once the shortlist is set, it feeds straight into your project planning phase deliverables, such as the scope, schedule and resource plan.
AFFiNE does not ship a dedicated impact effort matrix template at the time of writing: neither the website's template gallery nor the edgeless template set in its open-source code includes one. You can build one from the files above in a few minutes.
About this guide: the Fernhill Plants scenario, its numbers and all diagrams were created for this article to show the method end to end. They are not survey results, customer data or product screenshots.