home/services/speed
$ speed

Make the mobile score stop lying

Video grids, thumbnails everywhere, ad scripts and a theme nobody profiled. Speed work here is not installing a cache plugin — it is finding which of the thirty things on the page is actually holding the paint hostage.

Lab score and field data (CrUX) — a 100 that real users never see is worthless

lighthouse — mobile run
$ lighthouse --preset=perf --emulated-form-factor=mobile performance 69 · lcp 4.8s · cls 0.31 render-blocking css .......... 480 ms webfont swap shifts layout ... 0.19 cls jpeg thumbnails .............. 1.9 MB $ xsoft speed --fix critical-css,fonts,webp,cache performance 88 · lcp 1.9s · cls 0.02 field data follows in ~28 days
what we do

The list, in the order it pays off

Critical CSS

Above-the-fold styles inlined and generated from the real stylesheet, so it stays correct after you change the design instead of drifting into a broken first paint.

Images

WebP and AVIF companions for the whole library, correct sizes per breakpoint, lazy loading below the fold and an explicitly preloaded hero. Usually 30–60% less bandwidth.

Fonts

Self-hosted, subset, preloaded, font-display: swap with metric-compatible fallbacks so the swap does not shove the layout sideways.

Cache layers

Page cache, object cache and CDN that agree with each other. Also the trap nobody debugs: a cache that froze the old image format and serves jpeg to browsers asking for avif.

CLS hunting

Layout shifts rarely reproduce in your browser. We chase them with local Lighthouse traces until the element responsible has a name and a reserved box.

Server side

TTFB is the ceiling everything else lives under: PHP pools, slow queries, missing indexes, and cron jobs that run on visitor requests.

receipts

Recent runs

69mobile before
88after
4.8slcp
1.9slcp
0.31cls
0.02cls

three sites on one server, same week · your ceiling depends on your ad stack and hosting

honest limits

Ads cost points

Third-party ad and tracker scripts will always take a bite out of the score. We tell you exactly how many points each network costs so you can decide — instead of quietly removing your revenue to win a benchmark.

no cheating

We do not fake it

Serving a stripped page to testing tools is detectable and gets sites in trouble. Everything we do is served to real visitors, which is why field data moves too.

faq

Questions

Will speed alone bring traffic back?

Rarely on its own. Speed is a tie-breaker and a conversion lever; crawl and content problems are usually the bigger hole. The audit tells you which one you have.

Does this survive my next theme update?

That is why critical CSS is generated from your stylesheet by a script we hand over, rather than pasted in once. You regenerate it after design changes.

Can you work on a site that is not on your theme?

Yes. Most speed jobs are on other people's themes and page builders.

Send the URL, get the numbers

A written diagnosis with the specific blockers on your pages, before any invoice.