Before I add another paid app to my business, I ask one question first: could I actually build this myself, and would owning it outright be better than renting it. Three real times this year the honest answer was yes. One time, it genuinely wasn't. Here's the real reasoning behind each one, and the actual test I run every single time this same question comes up again.
Two SEO tools I was handed and didn't install
I was given two separate SEO tools to try for this site's blog, both uploaded the same day I was standing up the real pipeline. One turned out to be nearly a duplicate of a tool I'd already reviewed that same day for a different blog. The other was built for comparison-page SEO on a SaaS product, which has nothing to do with a personal blog like this one. I didn't install either one.
What I actually kept from those two tools
Instead of installing them, I pulled five specific techniques out of the review and folded them directly into the blog pipelines I already run: mapping out the real follow-up questions a reader would actually ask about a topic, keeping every section able to stand on its own without needing the one before it to make sense, naming the real product or number in a sentence instead of a vague "this one," answering each section's own question before telling any story around it, and keeping a consistent, recognizable look on every post's featured image. All five, for free, with nothing installed and nothing subscribed to.
Pinterest pins that render themselves instead of costing a design-tool seat
Every pin across all three Pinterest pipelines gets its actual image built by code, rendered through a browser engine running quietly in the background, not through a paid design tool or template service. The same renderer serves Marie Gems, Your Kind Mind, and now this site, and building it once, in-house, means adding a fourth brand later costs nothing extra in tool fees, just more pins running through the same real system.
A knowledge base instead of a vector database subscription
Rather than pay an ongoing subscription for a hosted AI search tool, I built a local one on September 16th that searches my own Shopify catalog and blog files with no database bill at all. It's not fancier than what I could have bought. It's exactly as fancy as what I actually needed, running on files already sitting on my own computer, tested against thirteen real records before I trusted it with anything important.
Replacing a paid routine that didn't sound like me
I used to run social posting through a paid cloud routine that technically worked, real content, actually posted, but it came out in a voice that wasn't mine, emoji and all. Rather than shop around for yet another different paid tool, I rebuilt the whole thing myself in-house specifically so I could control the one part that mattered most to me: it had to actually sound like me, not like a template.
When does the answer become no, just pay for it?
My Higgsfield account for AI image generation is currently blocked at every model I've tried, and the honest answer there is to upgrade the plan, not to build around it. There's no real local substitute for AI image generation the way there was for search or scheduling. Sometimes the tool genuinely does something I can't reasonably rebuild myself, and pretending otherwise would just waste time I don't have to waste.
The real test I use
If I can define exactly what the tool needs to do, check its output myself, and keep everything it touches on my own files where I can see it, it's usually worth building. If it needs something I have no real path to building myself, like actual image generation from scratch, it's worth paying for instead. Most months this question comes up again for something new, and running it through the same test keeps me from either overbuilding everything myself or paying out of habit for a tool I never actually needed.
More on the businesses this reasoning actually runs is on the Brands page.