On this page
Not the expensive ones. The tool that costs the most is usually the one you use the most, which is why you noticed its price. The waste is in the ones that renew without ever crossing your mind — and those are, by definition, the ones you will not think of when you try to make this list from memory.
So do not make it from memory. Twenty minutes, your bank statement, and three columns.
The audit
Open the last three months of card statements and write down every recurring charge. Then add two columns:
| Tool | €/month | Last opened | Replaces |
|---|---|---|---|
Last opened is the one that does the work. Not "do I use it" — when did you last actually open it. If the honest answer is a month you have to think about, that is the answer.
Replaces is the check against false economy. A tool that replaces four hours a month is cheap at almost any price. A tool that replaces nothing is expensive at €5.
Sort by the third column, not the second. Price and waste are close to unrelated.
What usually comes out of the first pass?
Three things, in my experience of doing this and in the pattern everyone describes.
The overlapping pair. Two tools doing one job, adopted months apart, each for one feature the other lacked. You use one of them. The second is still billing.
The trial that converted. Signed up to test something, it worked, the trial ended, the charge started, and the project it was for finished in April.
The one you cannot identify. A line on the statement you have to search for to recognise. That recognition problem is the whole argument — if you cannot name it, it is not doing anything you would miss.
The whole list
67 tools, what each costs and when you actually need it. Updated twice a year.
What about things you use twice a year?
Not automatically waste, and this is where people over-correct.
A tool you open twice a year is worth keeping if what it does on those two days is worth the annual cost. Something that saves a day of work twice a year at €9 a month is €108 against two days — fine.
The question is not frequency, it is what happens on the day you need it. If the answer is "I would find a free alternative in twenty minutes", cancel it and find the alternative on the day. If the answer is "the whole job stops", keep it.
What should you check before cancelling?
Two things, and both are cheaper than discovering them afterwards.
Export anything you would miss. Some services delete on cancellation, some keep data for a grace period, and the difference is not advertised. If there is anything in there you would want, get it out first.
Check what depends on it. Automations are the common trap: a tool cancelled in isolation can be a step in a chain you set up a year ago and have not looked at since. The failure is silent, which is the worst kind.
Which switches are still on when you launch? is about the same category of problem from the other direction — things set up once that nothing reminds you about.
When is cancelling the wrong move?
When the subscription is the cheap half of the problem.
A €20 tool that saves you an hour a week is not where your money is going. If the audit is being driven by the feeling that costs are too high, check what the costs actually are first — what AI tools actually cost per month has the realistic numbers, and the total is usually lower and more boring than the anxiety suggests.
Cancelling four small things to feel in control, while the actual problem is that nothing is selling, is a well-trodden path and it does not lead anywhere.
How often?
Quarterly, on a calendar reminder.
Monthly is too frequent to see a pattern and becomes a chore you drop. Yearly lets twelve full months of something unused go through. A quarter is long enough that the "last opened" column is informative and short enough that nothing runs for a year unnoticed.
What this article is missing
A list of the ones I cancelled.
That would be the stronger version of this piece, and I am not going to invent it. The method above is what I would hand someone; the personal list is a different article and it needs the actual statements in front of me rather than a plausible reconstruction.
The one real example I can give is from the other side of the ledger: a tool was removed from the list on this site — not cancelled, but dropped as a recommendation — after its links stopped resolving where they claimed to. The lesson transferred: check that the thing still does what it did when you adopted it, because neither tools nor your use of them stay still.
Where this came from
Method rather than measurement, and deliberately so. The three recurring findings are a pattern rather than a statistic, and the audit itself is the part worth copying.
The AI Mastery Playbook covers the quarterly review this belongs in.
Tools in this piece
| Tool | What it does | Price |
|---|---|---|
| Plausible | Simple, privacy-friendly visitor stats with no cookie banner. | ~€9/mo |
| Umami | Same idea, free if you host it yourself. | Free self-hosted |
| Zapier | The most integrations, the highest price. | Free tier, then ~$20+/mo |
| Make | Visual automation between tools, lower barrier than n8n. | Free tier |
| n8n | Workflow automation you can host yourself. Plenty of control. | Free self-hosted, cloud ~€20/mo |
| Canva | Thumbnails, covers and anything that has to look good quickly. | Free tier |
| Figma | Design and screens, including showing a client something before you build. | Free tier |
Every tool has its own page with the price, who should skip it and what to check before paying. Some links are affiliate links.
Questions people ask
How do you decide which subscriptions to cancel?
Work from your bank statement rather than from memory, sort by what you opened in the last month rather than by price, and cancel anything you cannot name a use for. The expensive ones are usually fine; the forgotten ones are the problem.
Why start from the bank statement?
Because memory systematically omits the ones that matter. You remember the subscription you think about and forget the one that renews silently, and the silent one is precisely the candidate for cancellation.
Should you cancel the most expensive tool first?
Rarely. Price and waste are almost unrelated. A €40 tool you use daily is excellent value and a €9 one you have not opened since March is pure waste, and only the second is a decision.
What about tools you use occasionally?
Paying for something you use twice a year is not automatically waste — it depends on what it replaces when you need it. The question is whether the annual cost is less than the cost of not having it on the day you do.
Does cancelling break things?
Sometimes, and that is the risk worth checking before you cancel rather than after. Export anything you would miss, and check whether something else depends on it, especially automations.
How often should you run this?
Quarterly. Monthly is too often to see a pattern and yearly lets a full twelve months of an unused subscription go through. A calendar reminder beats intention.