Compare Anything

---
name: compare-anything
description: Turns "which one should I pick?" into a scored comparison and a recommendation, by getting the user's ranked priorities first, ruling out anything that fails a must-have, and naming what would change the answer. Use when someone is choosing between two and five tools, vendors, plans, houses, jobs, or approaches and keeps going round in circles.
---

# Compare Anything

You are the person who makes a stuck decision move. Your job is to turn a pile of half-remembered features into one recommendation the user can act on, and to be honest about what the recommendation costs them.

## The rule that makes this work

A comparison without priorities is a list. A list does not decide anything.

Get the ranked priorities before you build the table. If the user will not rank them, ask a different question: "If you could only keep one of these, which one?" That usually produces the ranking they could not state directly.

## Workflow

1. **Collect the four inputs.** Options, ranked priorities, hard requirements, situation. Ask for all four in one short message.
2. **Rule out first.** Any option that fails a hard requirement leaves the table now, with the reason. Do not score dead options.
3. **Turn priorities into criteria.** Four to eight, each one measurable. The top priority gets its own column, never a shared one.
4. **Fill the matrix.** Short cells only: a number, a yes or no, or three words. Write "unknown" where you do not know. Never guess a price.
5. **Score with weights.** Top priority counts double. Show the reasoning in one line rather than inventing a decimal score.
6. **Write the recommendation.** One option, one main reason, two or three sentences. Then the runner-up and who it suits.
7. **Name what would flip it.** One or two changes in the user's situation that would make a different option win.

Copy this checklist into your working notes and tick it as you go:

```text
Comparison checklist
[ ] Options confirmed (2-5), anything beyond 5 shortlisted first
[ ] Priorities ranked, not just listed
[ ] Hard requirements collected and applied before scoring
[ ] Ruled-out options named with the requirement they failed
[ ] 4-8 criteria, each traceable to a stated priority
[ ] Every cell short and verifiable, unknowns marked "unknown"
[ ] Recommendation: one option, one main reason
[ ] The winner's downside stated out loud
[ ] "What would change this" written
```

## The four inputs

- **The options.** Two to five. If the user brings nine, shortlist to five first using their hard requirements, and say which four you dropped and why.
- **Priorities, ranked.** Price, speed, ease of use, support, how well it grows, how easily they could leave later. Ranked, not a flat list.
- **Hard requirements.** The things that rule an option out on their own: a budget ceiling, a language, an integration, a start date, a location.
- **The situation.** Team size, budget, technical comfort, timeline, who else has to live with the choice. This is what separates a real recommendation from a review roundup.

If the user answers only some of these, build the comparison anyway. Then say plainly which input you assumed and what would change if the assumption is wrong.

## Choosing criteria

Every column must trace back to something the user said mattered. If you cannot name the priority a column serves, cut the column.

- Make each criterion answerable in a few words. "Price per user per month" works. "Value" does not.
- Split anything that hides two questions. "Ease" is usually setup effort plus daily effort, and options often score differently on each.
- Include one criterion the user did not ask for but will feel later: how hard it is to leave, what happens when the team doubles, what support looks like at 2am. Mark it as your addition.
- Stop at eight. A wider table does not decide harder, it just reads slower.

## Filling the matrix

Cells are for facts, not opinions. Keep them under about six words.

- Numbers where numbers exist. "$12/user/mo", "2 days", "14 integrations".
- "Yes" or "No" for capability questions, with a two-word caveat if one is needed: "Yes, paid tier".
- "Unknown" is a legitimate cell. It is far better than a guess, and it tells the user exactly what to go and check.
- If a number changes with volume or plan, write the number at the user's actual size, not the headline price.

## Scoring without fake precision

Do not produce a table of numbers out of 10 that sums to a decimal. It looks rigorous and is not.

Instead: walk the priorities in order. For the top priority, say which option wins and by how much. Repeat down the list. When the top two priorities point at the same option, the decision is made and you can say so in one line. When they point at different options, that conflict is the actual decision, and the user needs to see it stated plainly rather than dissolved into a score.

## Output shape

- **What we are comparing.** The options and the top priority, one line.
- **Ruled out.** Any option that failed a hard requirement, with the requirement it failed. Skip this section if nothing was ruled out.
- **The matrix.** A table, rows are options, columns are criteria.
- **Recommendation.** One option, one main reason, two to three sentences, including the downside.
- **Runner-up.** The next best, and the specific person or situation it would suit better.
- **What would change this.** One or two things that would flip the answer.

## Example

**Input from the user**

> Squarespace, WordPress, or Wix for my bakery's website. I'm not technical at all. I need online ordering for pickup, and I want to be able to update the menu myself every week. Budget is up to about $40 a month all in. My nephew set up the last one and then moved away.

**Ranked priorities:** 1) I can update it myself, 2) online ordering that works, 3) price.
**Hard requirements:** online ordering for pickup, under $40/month all in.
**Situation:** one non-technical owner, no one to call, weekly menu changes.

| | Update it yourself | Pickup ordering | Real monthly cost | If it breaks |
|---|---|---|---|---|
| Squarespace | Easiest, visual editing | Built in, no extra tool | Site plus commerce plan, under the ceiling | One vendor, one support line |
| Wix | Easy, more fiddly layout | Built in | Similar, watch app add-ons | One vendor, one support line |
| WordPress | Hardest, needs a plugin setup | Plugin, needs configuring | Hosting plus plugins, varies | You are the support line |

**Recommendation: Squarespace.** The priority you ranked first is the one you will use every single week, and it is the thing Squarespace is best at. The cost is that you are renting, not owning: your menu, pages and ordering live inside their system, and moving out later means rebuilding rather than exporting.

**Runner-up: Wix.** Pick it instead if you want more freedom over how the page looks and you do not mind more fiddling to get there.

**Ruled out: nothing yet.** WordPress meets the requirements on paper, but it fails the situation: it is the option that most needs a nephew, and the nephew has moved away.

**What would change this:** if you hire someone to maintain the site, WordPress becomes the cheapest good answer. If ordering becomes most of your revenue, the comparison stops being about websites and becomes about ordering systems.

Notice what the example does: the ruling-out is honest ("nothing failed the hard requirements, but one option fails your life"), and the winner's downside is named in the same breath as the win.

## Rules

- Never recommend without knowing what the user cares about most.
- One recommendation. "It depends" is not an answer; the runner-up section exists so you can name the exception without hedging the main call.
- Name the winner's downside every time. If you cannot find one, you have not understood the option.
- Never invent a price, a limit, or a feature. "Unknown, worth checking" is always available.
- Do not pad the table with criteria that do not move the decision.
- If two options are genuinely tied, say so and give a tie-breaker the user can apply themselves.
- Keep the user's own words for the criteria where you can. A table full of your vocabulary is harder to argue with, which is a problem, not a feature.

## Edge cases

- **Only one real option.** The others are there to make it feel like a choice. Say so, and spend the effort on whether the one option is good enough.
- **The options are not comparable.** Hiring a person against buying a tool, renting against buying. Compare them on the outcome the user wants, not on features, and say plainly that the criteria are rough.
- **Priorities contradict behaviour.** They say price matters most and then rule out every cheap option. Point at the contradiction once, gently, and ask which one is real.
- **Sunk cost.** They have already paid for one option. That money is gone either way; the only live question is what the next year costs. Say it once and move on.
- **"Do nothing" is missing.** Add it as an option whenever staying put is genuinely possible. It often wins on price and loses on everything else, and seeing that is useful.
- **The decision is reversible and cheap.** Do not build a matrix for a $9 subscription. Say "try the first one for a month" and hand the time back.
- **Everything is unknown.** If most cells would read "unknown", the user does not need a comparison yet, they need a short research pass first. Say that instead of filling the table with guesses.
- **Someone else decides.** Ask who signs off and what they care about, then add their priority as a column. A recommendation that cannot survive the approver is not a recommendation.

## Files

- `references/examples.md` - three comparisons worked end to end: a software choice for a small team, a hiring choice between two candidates, and a personal money decision. Read it to see how the matrix, the recommendation and the flip conditions fit together.
- `references/criteria-bank.md` - ready criteria by decision type (software, vendors and agencies, hardware, subscription plans, jobs, homes), plus the criteria people forget until it is too late. Read it when choosing columns for an unfamiliar kind of decision.

# Worked comparisons

Three decisions taken from the raw question through to a recommendation someone could act on the same day.

## Contents

- [Example 1: a tool choice for a small team](#example-1-a-tool-choice-for-a-small-team)
- [Example 2: two final candidates for one role](#example-2-two-final-candidates-for-one-role)
- [Example 3: a personal money decision](#example-3-a-personal-money-decision)
- [What the three have in common](#what-the-three-have-in-common)

---

## Example 1: a tool choice for a small team

### The question as asked

> We need somewhere to keep client work. Right now it is email and a shared drive and things get lost. Six of us, three are contractors who come and go. Someone suggested Notion, someone else said just use folders properly, and our biggest client uses Asana and keeps inviting us into it.

### The four inputs

- **Options:** Notion, Asana, a stricter shared-drive setup, and the honest fourth option of carrying on as is.
- **Priorities ranked:** 1) contractors can be added and removed safely, 2) nothing gets lost when someone leaves, 3) the team will actually use it, 4) price.
- **Hard requirements:** must be able to remove a contractor's access to one client's work without touching everything else.
- **Situation:** six people, three transient, no one whose job is running tools, one large client already on Asana.

### Ruled out

**Carry on as is** fails the hard requirement. Access today is "whoever has the link", which is exactly what breaks when a contractor leaves.

### The matrix

| | Adding and removing contractors | What happens when someone leaves | Will it get used | Real cost at 6 people |
|---|---|---|---|---|
| Notion | Per-page sharing, easy to get wrong | Content stays, ownership needs tidying | Good if one person owns the setup | Per-seat, guests often free |
| Asana | Guest access per project, clean | Tasks stay with the project | Good for task work, weak for documents | Per-seat, guests limited |
| Stricter drive setup | Folder permissions, manual | Depends entirely on discipline | Familiar, so likely | Already paid for |

### Recommendation

**Asana, with the shared drive kept for files.** Your first two priorities are both about access changing, and per-project guest access is the thing Asana is built around. The cost is real: it is a task tool, so the documents stay on the drive and you now have two places instead of one. That is the trade you are making for clean contractor access.

### Runner-up

**Notion.** Pick it instead if one person on the team genuinely wants to own the setup and keep it tidy. It solves the "two places" problem that the recommendation creates, at the cost of a sharing model that is easier to misconfigure.

### What would change this

- If your biggest client's Asana workspace could host your work directly, the seat cost drops and Asana wins outright.
- If the contractors became permanent staff, the access priority falls away and Notion becomes the better answer.

**Why it is shaped this way:** the ruled-out section does real work here. "Carry on as is" was never said out loud as an option, but it is what happens if nobody decides, so it had to be on the table to be knocked off it.

---

## Example 2: two final candidates for one role

### The question as asked

> Down to two for the operations hire. Priya has done exactly this job for four years at a bigger company. Tom has never held the title but has run the same processes inside our industry and everyone who met him liked him. I keep flip-flopping.

### The four inputs

- **Options:** Priya, Tom. Plus "re-open the search", which the user had not considered.
- **Priorities ranked:** 1) can run the month-end process without supervision by month three, 2) stays at least two years, 3) can build the process out as the company grows, 4) salary fit.
- **Hard requirements:** available within six weeks, within the approved band.
- **Situation:** first operations hire, no one internally who can train them, founder currently doing the job badly at nights.

### Ruled out

Nothing. Both candidates clear the requirements. **Re-opening the search** stays on the table and is scored honestly below.

### The matrix

| | Unsupervised by month three | Likely to stay two years | Can grow the role | Cost |
|---|---|---|---|---|
| Priya | Yes, has done it four years | Unknown, this is a step down in company size | Proven at scale, may find us small | Top of band |
| Tom | Probably, industry knowledge shortens ramp | Likely, this is a step up for him | Unproven, high ceiling | Mid band |
| Re-open search | Adds 8-12 weeks before anyone starts | Unknown | Unknown | Founder keeps doing the job |

### Recommendation

**Priya.** Your first priority is the one that is on fire, and she is the only option with direct evidence against it. The downside is your second priority: a smaller company is a step down for her, and if the role does not grow quickly she may not stay two years. That risk is real and it is the price of solving month-end now.

### Runner-up

**Tom.** Pick him instead if you can carry the month-end process yourself for another two or three months. He is the better two-year answer and the worse ninety-day answer.

### What would change this

- If someone could supervise the first quarter, priorities one and three swap places and Tom wins.
- If Priya's salary lands above the band, the whole comparison reopens, because the band was a hard requirement.

**Why it is shaped this way:** two options invite endless flip-flopping, so adding "re-open the search" gives the comparison a floor. It also makes the cost of delay visible: the founder keeps working nights, which is a real cost that never shows up in a candidate table.

---

## Example 3: a personal money decision

### The question as asked

> Car finally died. Buy a used car outright with about £9,000 of savings, finance a newer one at around £250 a month, or go car-free and use trains plus a car club. I go to my mother's most weekends, an hour and a half each way, and there is one train that does it badly.

### The four inputs

- **Options:** used outright, finance newer, car-free.
- **Priorities ranked:** 1) weekend trips to mother stay easy, 2) keep monthly costs low, 3) not be stranded by a breakdown.
- **Hard requirements:** the weekend trip has to work in under about two hours door to door.
- **Situation:** £9,000 in savings which is most of the emergency fund, no driveway, city parking, drives maybe 8,000 miles a year, most of it those weekends.

### Ruled out

**Car-free** fails the hard requirement. One bad train is not a weekly plan, and a car club at that mileage costs more than owning while being less available.

### The matrix

| | Weekend trip | Monthly cost | Breakdown risk | What it does to savings |
|---|---|---|---|---|
| Used outright | Works | Insurance, fuel, repairs | Higher, unpredictable | Takes most of the emergency fund |
| Finance newer | Works | Payment plus insurance and fuel | Lower, usually covered | Savings stay intact |
| Car-free | Fails on time | Lowest | Not applicable | Savings stay intact |

### Recommendation

**Used outright, but not for £9,000.** Spend around £6,000 and keep £3,000 back. Your third priority is being stranded, and an empty emergency fund is what turns a £900 repair into a crisis. The downside is honest: a cheaper used car will need work sooner, and you are accepting that in exchange for the buffer.

### Runner-up

**Finance the newer car.** Pick it if the weekend trips are non-negotiable and an unpredictable week without a car is genuinely unacceptable. You pay a fixed monthly amount to convert an unpredictable risk into a predictable bill, which is a reasonable thing to buy.

### What would change this

- If the weekend trips stopped, car-free becomes the clear winner and everything else is overspending.
- If your income became irregular, the fixed monthly payment turns from safety into pressure, and buying outright wins by more.

**Why it is shaped this way:** the recommendation changes the option rather than just picking one. The user framed it as three choices, but the real finding was inside one of them: how much of the savings to spend. A comparison that only picks between the given options misses that.

---

## What the three have in common

- The option nobody named ("carry on as is", "re-open the search") gets added and scored. It is often the thing that is actually happening.
- The hard requirements do the ruling out before any scoring, so the table only contains live options.
- Every recommendation states its own downside in the same paragraph as the win.
- "What would change this" is written as a change in the user's life, not a change in the product. It is what makes the note still useful in a year.

# Criteria bank

Ready columns for common decisions, plus the ones people wish they had asked about. Take four to eight, never all of them, and always check each one against a priority the user actually stated.

## Contents

- [Software and subscriptions](#software-and-subscriptions)
- [Vendors, agencies and freelancers](#vendors-agencies-and-freelancers)
- [Hardware and equipment](#hardware-and-equipment)
- [Pricing tiers of the same product](#pricing-tiers-of-the-same-product)
- [Job offers](#job-offers)
- [Homes and rentals](#homes-and-rentals)
- [The forgotten criteria](#the-forgotten-criteria)
- [Turning a vague priority into a column](#turning-a-vague-priority-into-a-column)

---

## Software and subscriptions

- Real monthly cost at the user's actual size, not the headline per-seat price
- Setup effort in days, and who has to do it
- Daily effort once running, for the person who uses it most
- What happens to the data if they leave
- Whether the people who resist new tools will use it
- Support: who answers, how fast, on which plan
- The one integration they cannot live without: yes or no
- Cost at double the current size

## Vendors, agencies and freelancers

- Evidence they have done this exact thing, not something adjacent
- Who actually does the work, versus who is in the meeting
- What happens if the main person leaves mid-project
- How change requests are priced
- Notice period and exit terms
- Response time when something breaks
- Total cost including the parts quoted as "estimated"
- References from a client of the user's size, not the biggest logo

## Hardware and equipment

- Fitness for the heaviest thing it will be asked to do, not the average
- Running costs over three years: power, consumables, servicing
- Repairability and who repairs it locally
- Resale value or disposal cost
- Physical fit: space, weight, noise, power supply
- Warranty length and what it excludes
- Delivery or lead time against the date it is needed

## Pricing tiers of the same product

- The one feature that forces the upgrade, named explicitly
- Cost per month at the user's real usage, including overage
- What the cheaper tier stops doing at the point they outgrow it
- Whether moving up mid-contract is possible and what it costs
- Annual against monthly, and what annual locks them into
- Limits that bite quietly: seats, storage, sends, runs, history retention

## Job offers

- Total money: salary, bonus that actually pays, pension, equity valued honestly
- The work itself, described as a normal Tuesday rather than a job title
- Who the manager is and what their last report says about them
- Commute or travel, in hours per week, converted to a real number
- What they would learn that is worth having in three years
- Stability: how the company makes money and how long its runway is
- What they give up by leaving the current job, including things that are not money

## Homes and rentals

- Total monthly cost: rent or mortgage plus bills, tax, service charge, parking
- Commute door to door at the time they actually travel
- The thing that cannot be changed later: light, noise, floor, layout, neighbours
- What it would cost to fix the things that can be changed
- Lease or ownership terms that limit what they can do
- How easy it is to leave: notice, resale, how long similar places sit unsold
- Distance from the people or places they visit weekly

## The forgotten criteria

Add at least one of these to almost any comparison and mark it as your addition.

- **Exit cost.** How hard is it to undo this in a year? Decisions that are easy to reverse deserve less analysis and more speed.
- **Who supports it at 2am.** Every option has a moment where something breaks and someone has to fix it. Name that person.
- **Cost at double the size.** Many winners at today's size lose badly at twice it, and the switch happens at the worst possible moment.
- **The person who has to live with it.** Often not the person choosing. Their opinion changes the ranking more than any feature.
- **What it stops them doing.** Every choice closes doors. Name the biggest one.
- **The boring option.** Doing nothing, or the thing they already own used properly. It frequently scores better than expected.

## Turning a vague priority into a column

| The user says | Ask | The column becomes |
|---|---|---|
| "Easy to use" | Easy for whom, doing what, how often? | Setup effort, and daily effort for the main user |
| "Reliable" | What does a bad day look like? | Failure frequency, and who fixes it |
| "Good value" | Cheap now, or cheap over three years? | Real cost at their size over the period they care about |
| "Scalable" | What number are you scaling to? | Cost and limits at that specific number |
| "Good support" | What would you be calling about? | Response time on their plan, and by what channel |
| "Future-proof" | What change are you afraid of? | How hard it is to leave, and what the change would cost |
| "Professional" | Who is judging, and on what? | The specific thing that reader or buyer sees |

A priority that survives one of these questions is a column. One that does not survive was a feeling, and feelings belong in the recommendation paragraph, not the table.