Slow pages, fixed
at the actual cause.

LCP, INP and CLS diagnosed from what real visitors experience, not from a lab score you can game. We measure, fix the causes in order of size, and hand you before-and-after numbers. Prepaid in Hour Blocks of 6, 12 or 24 hours - a Starter Hour Block usually covers a full audit with a costed fix list.

LCP Inp Cls Field data, not lab scores Before & after numbers

Within two business days

Your message goes straight to the studio. No sales queue, no account manager in between.

Prepaid Hour Blocks

Buy 6, 12, or 24 hours up front at $30/hr. Unused hours never expire.

10 live products

Documented Core Web Vitals work on this very site that you can read in its source.

You own the code

Full source, in your repository, with the reasoning written down. No lock-in to us or anyone else.

Three metrics,
a dozen real causes.

Core Web Vitals are three numbers, but each one fails for specific, findable reasons. Here is what we look for and what the fix looks like, so you can tell this apart from a report generated by a plugin.

Largest Contentful Paint

LCP should be under two and a half seconds, and it breaks into four parts: server response time, the delay before the browser even discovers the important resource, how long that resource takes to download, and the delay before it renders. Diagnosing which part dominates is the whole job - the fix for a slow server has nothing in common with the fix for a hero image that is lazy-loaded and therefore discovered late. Render-blocking CSS, unhinted fonts and a JavaScript-injected hero element are the usual suspects.

  • LCP broken into its four sub-parts
  • Lazy-loading removed from the LCP element
  • Fetch priority and preload used correctly
  • Render-blocking CSS reduced or inlined
  • Font loading that does not hide text
  • Server response and caching at the origin

Interaction to Next Paint

Interaction to Next Paint replaced First Input Delay in March 2024 and is considerably harder to pass: it measures the full path from any interaction to the next painted frame, for every interaction, not just the first. Failures come from long tasks blocking the main thread, event handlers doing layout-triggering work synchronously, oversized DOM trees making every style recalculation expensive, and third-party scripts that were never budgeted. The fix is breaking up work and deferring what does not need to be immediate.

  • Long task identification on real interactions
  • Handlers split so the frame can paint
  • Layout thrashing removed from hot paths
  • DOM size and style recalculation cost
  • Third-party script budget and deferral
  • Immediate visual feedback on slow actions

Cumulative Layout Shift

The metric users feel most viscerally, because it is the one that makes them click the wrong thing. Almost always caused by the same handful of omissions: images and videos without dimensions, ad and embed slots with no reserved space, web fonts swapping in at a different size, banners and notices injected above existing content, and content inserted after load without space held for it. Each fix is small and mechanical; finding all of them is the work.

  • Explicit dimensions or aspect ratios on media
  • Reserved space for ads, embeds and iframes
  • Font fallback metrics matched to reduce swap shift
  • Banners that do not push content down
  • Skeletons sized like the real content
  • Shift attribution back to the exact element

The scripts you did not write

Tag managers, analytics, chat widgets, heatmaps, ad pixels and consent banners. Each one was added by someone who did not have to pay for it in milliseconds, and together they are frequently the single largest cost on the page. The answer is rarely deleting them; it is loading them at the right time. This site defers its tag manager until the first real user interaction, with a five-second idle fallback, which keeps analytics working for human visitors while moving the scripting cost out of the window that blocking-time measurement cares about.

  • Inventory of what every third party costs
  • Deferral to interaction or idle
  • Consent banners that are not the LCP element
  • Duplicate and abandoned tags removed
  • Self-hosting where licensing permits
  • An honest cost-per-tag conversation

What the browser has to download

Images served at three times their display size, modern formats never enabled, CSS bundles carrying rules for pages this one does not use, and JavaScript shipped for interactions that never happen on a phone. Bundling decisions matter both ways: this site merged three render-blocking stylesheets into one because three separate blocking requests cost more than the bytes saved by splitting them. There is no universal rule, only measurement.

  • Responsive images in modern formats
  • Request count versus payload, measured
  • Unused CSS and JavaScript identified
  • Compression and cache headers verified
  • Preconnect and preload where they pay
  • Critical path reduced to what paints

When it is the server, not the page

Sometimes the front end is innocent and the server is taking eight hundred milliseconds to produce the first byte. That is a database and application problem: unindexed queries on the template, an N+1 in a loop, no page or object caching, a plugin running a remote call during render. Because we work on the backend as well, the diagnosis does not stop at the boundary where a front-end specialist would have to hand you a ticket for somebody else.

  • Time to first byte broken down
  • Slow query and N+1 identification
  • Page, fragment and object caching
  • Remote calls removed from render
  • Compression, HTTP version and TLS setup
  • Fixes implemented, not just recommended

Numbers, not adjectives.

Performance work is easy to fake, because everything feels faster to the person who just changed it. Everything below exists so you can check our claims rather than trust them.

A measured baseline

Field data for the templates that matter plus lab traces, captured before anything changes. Without a baseline, an improvement is an opinion.

A prioritised fix list

Every finding with its estimated impact and its estimated cost in hours, ordered by the ratio. You can hand it to your own team and never hire us.

The fixes implemented

Not just a report. Changes made in your codebase as reviewable commits, on a branch, verified on a staging copy before they ship.

Before-and-after comparison

The same measurements re-run after the work, side by side with the baseline, including anything that did not improve as much as hoped.

A way to not regress

A performance budget and a check you can run before a deploy, so the next theme update or tag addition does not quietly undo the work.

Written explanation

Why each change works, in language you can forward to a marketing team that wants to add another script next month.

Read the source
of this page.

The most checkable proof we have for this particular service is the site you are on. Its performance work is documented in comments in its own templates, dated, with the reasoning kept next to the code.

LCP

The consent banner was the LCP element

The GDPR banner used to be injected by JavaScript after parsing, which made it the largest contentful paint on every page. It is now server-rendered immediately after the body tag so it paints with everything else. Its markup has to byte-match the script that wires it up, and that constraint is noted in the template so nobody breaks it later.

Blocking time

Third-party scripting moved out of the way

The tag manager, analytics and ads tags together measured around 1.3 seconds of scripting. They now load on the first genuine user interaction with a five-second idle fallback, rather than on window load, which fires too early on a page this light. Humans still get measured; the blocking window does not.

Render blocking

Three blocking stylesheets became one

Three separate render-blocking stylesheets, about fifteen kilobytes compressed in total, were flagged at roughly a hundred and seventy milliseconds. They are concatenated in the same cascade order so specificity is unchanged - a merge rather than a rewrite, which is why it was safe to do.

Two other relevant pieces of our own work: TechPrognoza is a content site moved from Hugo to Django with speed kept as a requirement, and NajNajNovije is a content-heavy aggregator tuned to stay fast on mobile where front-end weight is felt first.

Sounds like your project? Prepaid Hour Blocks · reply within two business days
Start a project

Four steps, no discovery call
you have to sit through.

Measurement first. A Starter Hour Block usually buys a full audit and a costed fix list, which is yours whether or not you continue.

1

Tell us the problem

A few sentences on what you have and what you need. Rough is fine - we ask the questions that matter and tell you honestly if it is not a fit.

2

Pick an Hour Block

Prepay a block of hours - 6, 12, or 24. The block is the budget: we work until it is used, or you top up.

3

Work in the open

Fixes in your codebase as reviewable commits, highest impact first, verified on a staging copy and re-measured against the baseline.

4

Handover and support

You get the before-and-after numbers, the performance budget, and a short support window after the changes ship.

Bought in
prepaid Hour Blocks.

No volume games and no agency margin on top. You buy the block of hours that fits the size of the job, so you know the number before we start rather than after.

Starter 6 hrs
$180

Typically enough for a focused fix, a small feature, or an integration you are stuck on.

Start with Starter
Pro 24 hrs
$720

Typically enough for a larger build with integrations, auth, or billing wired in.

Start with Pro

Every Hour Block is hours at $30/hr with no discount. It is a bucket of time, not a fixed-scope deliverable. Unused hours do not expire - roll them into the same project or a later top-up. If the work runs long, you buy another block; nothing is billed without you agreeing to it first.

Frequently asked.

Will this improve our Google rankings?

Page experience is a real but modest ranking input, and anyone promising you positions from a speed project is guessing. The honest case for this work is different and stronger: faster pages convert better, fewer people abandon before the page paints, and a failing Core Web Vitals report in Search Console is one fewer thing standing between you and whatever else you are doing for search.

What is an Hour Block, and what if the job takes more or fewer hours?

An Hour Block is a prepaid block of hours: Starter is 6 hours for $180, Standard is 12 hours for $360, Pro is 24 hours for $720. It is a bucket of time, not a fixed-scope deliverable. Unused hours do not expire and roll into the same project or a top-up. A Starter Hour Block is usually enough for an audit with a prioritised, costed fix list.

Our PageSpeed score is 40. Can you get it to 100?

Possibly, but that is the wrong target. The lab score is a synthetic run on a simulated slow device; what Google actually uses for page experience, and what your users actually feel, is field data from real visits over a rolling 28-day window. We optimise the field metrics. The lab score usually improves as a side effect, and we will tell you when a change would move only the score and nothing real.

What is INP and why did our score change without us touching anything?

Interaction to Next Paint replaced First Input Delay as a Core Web Vital in March 2024. It is much stricter: it measures the full time from an interaction to the next frame being painted, across all interactions on the page, not just the first one. Plenty of sites that passed comfortably on the old metric started failing on the new one without a single code change, because the old metric simply was not measuring what users feel.

Is this not just installing a caching plugin?

A caching plugin helps time to first byte and sometimes nothing else. It will not fix an image that is lazy-loaded when it is the largest element, a layout that shifts because an ad slot has no reserved height, or an interaction delay caused by a third-party script blocking the main thread. Those are the usual causes, and each one is a specific fix rather than a setting.

What platforms do you work on?

WordPress and WooCommerce, Laravel, Django, Vue applications, and plain static sites. The diagnosis is the same everywhere because the browser does not care what generated the HTML. Where the platform matters is in the fix, and we can implement it rather than handing you a report that tells your developer what to do.

How long before the improvement shows up in Search Console?

The lab measurements change the day the fix ships. Field data is a rolling 28-day window, so Search Console typically takes several weeks to reflect it fully and will move gradually rather than jumping. That lag is normal, and it is worth knowing in advance so nobody panics in week two.

How do we start?

Send the URL of a page that matters through the form on this page, along with what platform it runs on and what your Search Console report says if you have it. You get a reply from us directly, usually within two business days, with an honest read on what is achievable.

Tell us which pages are slow.

The URL, what platform it runs on, and what your field data or PageSpeed report says if you have it. A few sentences is plenty - we read every message and reply within two business days with an honest read of what is achievable.

Prefer email?

[email protected]
Please enter your name
Please enter a valid email
Which Hour Block?
Tell us a little about the project
Thanks - we have got it.

Expect a reply from us within two business days.

Hour Blocks from $180reply within two business days Start a project