What I shipped:
- Two new tools live: Resize Image and Crop Image, both on libvips natively
- A long-form guide, "How to Compress Images Without Losing Quality," written to feed my compress JPG/PNG/WebP pages with real internal links and real benchmarks from 24 of my own iPhone photos
- A site-wide "Image Tools" block (inline section + footer column) so every page now cross-links to the new tools
- Squashed a nasty non-deterministic segfault in the libvips Rust bindings. It was a double-free: the binding derives Clone over the raw image pointer, so cloning copied the pointer without bumping the refcount and Drop freed the same object twice. Replaced the clone with an explicit copy, disabled the op cache, and the crashes plus the G_IS_OBJECT warnings are gone. Happy to write up the war story if anyone wants it.
The numbers (last 7 days):
- Clicks doubled, up 100% week over week. Still small in absolute terms, but real and trending the right way for the first time in weeks.
- Impressions also up 100%, as the new pages finished getting indexed.
- Average position is creeping up, into the mid-40s on the most recent day, from around 51 the week before.
- A brand-new long-tail query, "sony hif file to jpg converter," is already sitting at position 6. That is exactly the kind of low-competition query I am betting the whole strategy on.
The honest read:
Clicks and impressions both doubled, which is great, but the absolute numbers are still small and capped by domain authority. I have around 15 referring domains; the big converters have thousands. On competitive terms like "avif to png" (KD 42) I cannot rank without backlinks, and that is a months-long game. So the play stays low-KD long-tail: pages that can reach the top 10 even at low DR. That position-6 result on a niche query is the early proof it works.
Next week:
- Concentrate internal links on the low-KD pages that already have impressions
- Start background removal (rembg sidecar) research
- Keep backlink outreach slow and quality-first
convertifyapp.net
