I cold-email shop owners about the broken things on their sites. Last week my own scanner crashed on those exact sites.

Quick follow-up to last week's post about cold-emailing strangers.

The play is simple: scan a shop's site, find the one real accessibility thing that's broken, email the owner leading with their problem — not my tool. So I've been sending. Then I went to actually generate the reports for a few of those shops, and my own scanner fell over.

Not on toy sites. On the real ones. A typical small WooCommerce shop turned out to be 350KB of HTML, 61 images, 103 scripts. My scanner loads every page in a real headless browser — it has to, because accessibility depends on the rendered result: computed colors, menus that only exist after the JS runs. On my cheap 512MB server, those heavy shops ate all the memory and the whole thing 502'd. The heavier the shop, the harder it died.

Sit with that for a second: the exact customers I'm chasing — real shops with lots of product images — were the ones my tool couldn't handle. Light sites scanned fine. My actual prospects didn't.

The fix was more memory, which means money. And at $0 revenue every instinct screams don't pay. So I spent hours trying to stay free: blocking images during the scan, timeouts so one stuck scan couldn't wedge the whole server, scrolling the page to wake lazy-loaded images, chromium flags. It helped — some borderline sites started passing. But the heaviest, most real shops still died.

In the end I paid. $25/mo for enough RAM. The reframe that finally moved me: this isn't "scaling." A scanner that can't scan the sites I'm emailing about makes the whole outreach a lie. It wasn't growth spend — it was making the core promise true. The moment I upgraded, every shop that had been crashing scanned in 8–15 seconds.

Still $0 from strangers, to be clear. This didn't make me a cent. It just stopped me from selling something that didn't work on the people I was selling to.

The question I keep turning over: where's your line between "stay scrappy, stay free" and "the product has to actually work for the customer in front of you"? When did you spend money before a single sale — and was it the thing that mattered, or just fear dressed up as progress?