Ivo

Check the current numbers for your country before quoting them. They moved recently and they aren't the same everywhere.

Underrated: paper and a scan. Scanning apps do OCR now and you end up with searchable PDFs anyway, for zero subscription. I went iPad, went back to paper, and the only thing I actually miss is infinite page size.

There is no universal number because the constraints come from individual issuers rather than from some global rule. What most people I know settle on is roughly one card every three months as a default, which is slow enough that nothing looks frantic and fast enough to be worth doing.

The more useful habit is order rather than spacing. Some issuers care a lot about how many new accounts you have opened recently across all lenders, so people apply for those first while their record is quiet, and leave the relaxed issuers for later. Work out which of your targets is the fussiest and go there first.

Rules change constantly, so confirm current terms before you apply rather than trusting a two year old guide, including this comment.

At four customers, funnel analysis is astrology. You have four rows. Open them, find out who those people are, email all four, and ask what they were doing the day they signed up. You will learn more in one afternoon than a year of reconciling dashboards, and the sample size is the same either way.

Match the block to the shape of the task rather than picking one length for everything. Code review is naturally chunked, one file or one PR, so short timed blocks fit it well. Writing has a warm up cost of maybe fifteen or twenty minutes before anything good comes out, which means a twenty five minute block is almost entirely warm up and then a bell. I run about forty five to fifty for writing, open ended when it is going well, and twenty five for anything I am avoiding. Two lengths, chosen by task, not one universal number.

Going to disagree with the top answer slightly. At a genuine $110 with the same real parts, take the prebuilt, especially for a first machine. The value in building is that you choose every part and you learn the machine, not that you save money, because the saving mostly evaporated years ago. The one thing I would check is whether opening the case voids the warranty, because often it does not, and if you can add RAM and drives yourself without losing cover then the prebuilt is a perfectly good starting chassis.

I'll defend the truly boring option: one long text file, appended to, with the date at the top of each entry. Two years in. Search finds everything because it's one file. There is no app to migrate out of, ever. It isn't for everyone but the ceiling is much higher than people assume before they try it.

One small thing that moved my number more than it should have: telling people what the connection will and will not do, right next to the button, in plain words. Something like "we read your last 90 days, we never post anything, you can disconnect any time and we delete within a day".

I had all of that in a privacy page nobody opened. Moving three sentences of it next to the button was maybe forty minutes of work and it visibly changed the drop-off at the consent screen.

'Follows you across any change' is worth considerably more to me than the last 10% of features.

That's an upgrade I had no idea I was getting.

Set yourself a rule: you may install a plugin only after writing the note that made you want it. Sounds childish, worked immediately. My plugin count has been flat for a year and my note count hasn't.

Lunch in the main compartment with a laptop is how you eventually get yoghurt on a laptop. Separate bottom pocket or a properly sealed box. Learned that one expensively.

Time saved is real but not where I expected. It didn't save me the note-taking, it saved me the twenty minutes after every call where I used to write up actions. What it produces is a good action list and a mediocre summary. I read the actions every time and the summary about once a fortnight.

Demote it to a reference is a good reframe. I think I was treating finishing it as the goal.

Rebuilding from memory sounds miserable but so does forgetting everything, so I will try it for a fortnight.

Twenty customers at 200, and it is not close given your 25 hours.

Support load does not scale with revenue, it scales with the number of humans holding your product. Two hundred hobbyists will generate more tickets than twenty businesses, and the tickets will be harder to answer because the users are less expert and the stakes are lower for them. Twenty business customers you can know by name, call, and learn from.

The other thing nobody mentions is churn. Consumer subscriptions at 19 churn constantly, so you have to keep refilling the bucket, which means marketing forever. Business customers at 200 stick around because switching costs them a workflow, so month 14 revenue is mostly the same people from month 2.

The honest downside is that getting the first twenty is much slower and involves sales conversations, which most technical founders hate.

The reverse effect too: it made me pick up my phone far less, because I could see the buzz was nothing without opening a device full of other things. That was the actual life change.

Fair, and worth weighing. Though I would still start high and discount rather than start low and try to raise later, which is miserable.

Ask whether the diagnostic fee comes off the repair if you go ahead. Both shops near me do it and neither offers unless you ask first.

This is why 'fill the roof' advice aged so badly. Under net billing, oversizing is how you lose money slowly and feel good about it.

The thing that built my pan wasn't oven coats, it was six months of frying onions and bacon in it. Base coats stop rust and give you something to cook on. Everything after that is use. You cannot shortcut use with an oven cycle, however many times you run it.

The built-in notes app on my phone, four years, and I feel faintly embarrassed admitting it in this room. Gave up: linking, tags that mean anything, structure of any kind. Still annoys me: search is mediocre and it gets slow past a few thousand notes. I keep it because it opens in half a second, and that turns out to be the only feature I use every single day.

That maps onto my week precisely. My good afternoons are always review afternoons and I have been counting them as failures.

Every day reading as a failure is exactly the effect and I had not connected it to the size of the block.

Someone answering a specific question about their own workflow in writing. That costs them effort now, which is the only thing that predicts effort later.