Every January I do a thing I used to find tedious and now find clarifying: I go through my credit card statement, find every recurring software charge, and decide whether I would notice if it disappeared. The ones I would notice stay. The ones I would not, I cancel. The exercise has, over five years, taken my software spend from genuinely unreasonable down to something I can defend without flinching, and improved my actual working setup at the same time.
What follows is the result of the 2026 round, written up honestly because the genre of 'these are the apps I love' essays is almost always corrupted by the fact that the writer got most of the apps for free in exchange for the essay. I pay for everything on this list. The amounts are accurate as of last month. The opinions are unpaid and, for what it is worth, mine.
The keep list
Notion · $10/mo
I have tried, every six months, to leave Notion for something more focused. Every time I come back. Not because the alternatives are bad — Capacities is interesting, Anytype is impressive, Obsidian is genuinely better for some things — but because the gravitational pull of the existing six years of notes is real. The cost of moving is higher than the value of switching, and Notion has been good enough to not force the issue. The $10 is for the personal Pro tier, which gets me unlimited file uploads and AI tools I rarely use. Worth it for the storage alone.
Claude Pro · $20/mo
Of the two AI subscriptions I have paid for in the last year — ChatGPT Plus and Claude Pro — Claude is the one I would keep if I had to pick. It is better at the specific shape of conversation I have most often (writing-adjacent thinking, code review, long-document analysis). ChatGPT was better at a different shape (image generation, voice, the long tail of toy use cases). I cancelled ChatGPT in March and have not missed it once.
Spotify Family · $17/mo
Split with two siblings, so my actual cost is closer to $6. I have made the same complaint about the Spotify catalog rotation for six years and made it again last month. I still pay for it, because the alternative is a music library I built myself, and I tried that for a year in 2022 and missed the discovery enough to come back. The complaint is real; the cancellation is not, evidently. I am part of the problem.
iCloud+ · $3/mo
200GB. This is the cheapest line item on the list and the one I would defend most strongly. Photos and device backups all live here. The day I started paying for this is the day I stopped having storage anxiety, and the absence of storage anxiety has been worth far more than $36 a year. The principle generalizes: there is a category of subscription whose value is psychological rather than functional, and those are often the best deals.
1Password Families · $5/mo
I switched from a free open-source password manager three years ago, after a setup-friction incident on a friend's borrowed laptop convinced me that the time I was spending fighting my own security was more expensive than the subscription. 1Password is not perfect — the redesigns are routinely controversial, the desktop app keeps getting bigger — but it has not lost a password yet, and that is the only metric that actually matters.
Figma Professional · $15/mo
I do not, professionally, do that much design. I do enough that paying for the editor tier is worth it for the version history alone. The free tier is genuinely generous; the upgrade is for the workflow features that the company has deliberately put behind the wall, and the wall is, in my case, correctly placed.
Kagi Search · $10/mo
The most surprising entry on the list. I have paid for a search engine for two years now. The results are better, in the small-but-significant way that good tools tend to be better — faster, less cluttered, more obviously written for humans rather than for SEO. The ad-free quality is the explicit pitch. The quieter pitch is that paying for the search engine changes the economic relationship between you and the company, and that change is felt in everything from the result ranking to the support quality. I would not have predicted I would keep this subscription. I would now defend it at length.
Readwise + Reader · $10/mo
Highlights from books, articles, and PDFs, with a reading app that has slowly become the place I do almost all my long-form reading. The integration with my notes is the part I would miss most. The reading app is good; the highlight-management is the reason I pay.
The watching list
These are the subscriptions I have kept this year but am no longer confident in. Each is on probation. If the trajectory does not improve before next January's round, they will not survive the next pass.
GitHub Copilot · $10/mo
Was $10/mo well-spent in 2023. Has been slowly losing ground to the integrated AI in Cursor and the chat-first workflows I have moved most of my code into. I still use the autocomplete. I am no longer sure I use it enough to justify the line item. Probably gone next year unless the product evolves meaningfully.
A pomodoro / focus app · $5/mo
I will not name the app, because the specific app is not the point. The category is the issue. I have been paying for some version of a focus-timer subscription for four years. I open it less every quarter. The behavior change I was paying for has either stuck (in which case I do not need the app) or has not (in which case the app is not helping). Either way, the subscription is on borrowed time.
Cloudflare Pro · $20/mo
For my personal site, which gets about four hundred visitors a month. This is genuinely indefensible. I keep it because I configured it once, three years ago, and have not been bothered to revisit. The next time I touch the site, this goes. I include it in the list because everyone has at least one of these — a subscription that survives on inertia alone, and is worth catching.
The drop list
Cancelled this year, and what I learned in each case.
- ChatGPT Plus — cancelled March 2026. Claude does the same job better for my work. Saved $20/mo.
- Dropbox Plus — cancelled February 2026. iCloud is the right backend for my Apple-heavy setup, and the cross-platform sharing case had withered. Saved $12/mo.
- Bear Notes — cancelled May 2026. I had been keeping it as a backup writing app for two years and had not opened it in nine months. Saved $3/mo.
- A second password manager I had as a backup — cancelled in January after I realized I had not used it once in eighteen months. Saved $5/mo.
- Adobe Creative Cloud — cancelled in August 2025. Figma covers design, Photopea covers occasional image work, DaVinci Resolve covers video. The Adobe stack was costing me real money for a small number of use cases I could solve with free alternatives. Saved $60/mo, which is the single biggest line item I have cut.
The total
Adding it up, the active stack runs to about $100/month, or $1,200/year. That sounds like a lot until you realize it represents most of the software I rely on for both work and life. The same stack in 2022 was about $250/month. The drop is not because the apps got cheaper — most of them got more expensive over the same period. The drop is because the criteria for keeping a subscription got stricter, and because I stopped letting old subscriptions persist by default.
The most useful insight from doing this exercise consistently for five years: the apps you really need are usually clearer to you than you think. The clutter comes from the apps you signed up for once, used twice, and then accepted onto your card. The annual cleanup is mostly about removing those, and the removal is almost always frictionless. You will not remember most of the apps a month after you cancel them.
Recommendations to other people doing this
- 1Do the audit annually, on the same day every year. (I do it the first Saturday in January, with coffee.) The discipline beats the intelligence.
- 2Look at the actual statements. The app stores hide subscriptions. The credit card does not.
- 3For each line item, ask: 'Would I notice within a month if this disappeared?' If the answer is no, cancel. You can always re-subscribe.
- 4Do not optimize for the cheapest possible stack. Optimize for the stack that you trust. A $20/mo tool you reach for instinctively is a better deal than a $5/mo alternative you have to remember to use.
- 5Keep one indulgent subscription that you would never recommend to anyone but that you love. Mine is Spotify, which I have justified above with a story I do not entirely believe. The indulgent subscription is the proof that the audit is calibrated, not puritanical.
The meta-point
The reason this exercise has value beyond the dollars is that it forces you to be honest about which tools you actually use. The list above is the list of apps that, when I open my laptop in the morning, I genuinely reach for. The apps that did not make the list are the ones I had convinced myself I cared about, evidence to the contrary. The truth is in the credit card statement, not in the bookmarks bar.
If you have not done this in a while, the first audit will be slightly painful. It is also, in my experience, the cheapest hour-long project on the calendar. The savings are real, the clarity is real, and the cleaned-up stack is the stack you will use better, because the absence of clutter is itself a form of focus.
If you publish your own version of this list, send it in. We collect these in a folder for the year-end roundup, and the variety is consistently the most interesting reading we do — every personal stack is different, and the differences are usually more informative than the overlaps.
Tags