What you’re actually buying
You could build: The same shape as any optimiser: compress on upload, generate modern formats, bulk-process the back catalogue, keep the originals.
What you’d still be missing: Identical to EWWW, and for the same reason — the compression runs on their servers and is metered in credits because it genuinely costs them per image. Rebuilding the plugin gets you a client with nothing behind it; matching the product means paying for compute either way.
Is it worth building?
| If your time is worth | Year one | Each year after | Sites to break even |
|---|---|---|---|
| $25/hr | $350 | $200 | 4 |
| $50/hr | $700 | $400 | 8 |
| $100/hr | $1,400 | $800 | 15 |
| $150/hr | $2,100 | $1,200 | 22 |
Year one only, and it assumes the 70% above is worth nothing to you. That is the assumption most likely to be wrong — read the “still missing” line again before you decide.
The prompt
Build a self-contained WordPress plugin that covers the core of ShortPixel. I want a working first version I can install and test, not a scaffold. Requirements: - Compression on upload with a lossy/lossless choice - Bulk optimisation of an existing library without timing out - WebP and AVIF generation and delivery - Originals retained so it can be undone - Credit or quota accounting the site owner can see Constraints: - Plain PHP against the WordPress APIs. No build step, no framework, no Composer runtime dependencies. - Follow WordPress coding standards. Prefix every global function, class and option. - Escape on output, sanitise on input, nonce + capability check every write. Prepare every SQL query. - Translation-ready with a single text domain. - Uninstall routine that removes options and tables it created. - Target the current WordPress and PHP 8.1+. Deliver: the plugin files, a short README covering setup and limitations, and an explicit list of what you did NOT implement. Context for scope: the paid product is from $8.33/mo. Do not try to match it feature for feature — build the part I would actually use, and tell me where the gap is.
Don’t feel like building it?
Self-reported and unverified — one tap per person, roughly. Treat it as a mood ring, not a statistic.
Why this list exists
“Can I just vibe-code this?” has become a reasonable question to ask before renewing a licence. For most SaaS the answer is a straight yes or no. For WordPress plugins it usually isn’t, because a premium licence is rarely paying for source code alone.
Ask an agent for a security plugin and you’ll get a firewall and a file scanner in an afternoon. What you won’t get is the rule that shipped this morning for the vulnerability disclosed last night — that’s a research team, and it’s the entire reason the product works. Ask for a caching plugin and you’ll get page caching, minification and lazy-loading. What you won’t get is the eight years of exclusion rules that stop it breaking somebody’s checkout.
So instead of a verdict, every plugin above is split in two: the share an agent could plausibly build for you, and the share that only exists because a vendor is running something on an ongoing basis — a live data feed, servers doing the work, a legal review, or an integration matrix kept alive against other companies’ breaking changes.
How to read the score
The percentage is the answer to one question: what would you still be missing the day after the agent finished? A low score means the product really is mostly code, and building it yourself is a fair trade. A high score means you’d finish the build and still need to buy the thing that made it useful.
The baseline is one competent developer with a good agent and a couple of weekends — not a team, and not a year. Scores are editorial judgement rather than measurement, and they’re deliberately arguable. If you think a call is wrong, that’s the interesting conversation.
Ranking, prices and live-site counts come from the same daily data as the rest of the site — see the premium plugin ranking for how those are gathered.