We optimise real-user performance on the pages that carry your revenue, working from field data rather than a synthetic lab score. You get before-and-after numbers on your own URLs, and a written record of every change so nothing is a black box.
“Our score is green but the site still feels slow.”
Lab scores test one synthetic run. We work from field data, which is what Google actually ranks on. See the approach
“We failed Core Web Vitals in Search Console.”
The audit identifies which metric, on which template, for which device. Start with the audit
“Someone optimised it and it broke the layout.”
Every change is diffed visually and reverted if it costs more than it saves. See how we work
“We keep adding plugins to make it faster.”
Usually the fix is removing four of them. The audit says which. See the plans
These figures cover every Gatilab service, not this one alone. We’d rather say that plainly than imply a per-service record we haven’t published.
Brands we’ve delivered work for
Performance work goes stale the moment someone installs a plugin. The first two are projects. The third exists because most sites regress within a quarter.
For teams who know performance is a problem but not which part of it, or who need evidence before budget.
Up to 5 templates · 7 working days
The fee is credited against a full optimisation booked within 30 days.
For sites failing Core Web Vitals or losing conversions to load time, where someone needs to do the work.
Audit included · 3 to 4 weeks
If a target metric doesn’t improve, we keep working on it at no extra cost.
For teams releasing weekly, running many plugins, or managing multiple sites that drift back over time.
3-month minimum
If your site stays stable for two quarters, we’ll suggest you drop this.
Every stage produces evidence, because performance claims are easy to make and easy to check.
Days 1 to 3
We pull field data from Search Console and CrUX for your actual visitors, then profile each template in the browser. Lab tools come last, because they measure one run on one connection.
You leave with: A baseline you can re-measure yourself, per template and per device.
Days 4 to 7
Waterfalls, long tasks, layout shift sources, and server response time get attributed to specific plugins, scripts, images, and queries. Most sites have three changes worth 80% of the gain.
You leave with: A prioritised list with estimated impact and effort per item.
Weeks 2 to 3
Changes land on a clone and get visually diffed. Anything that saves 40ms but breaks a layout gets reverted. We keep the record of what we rejected too.
You leave with: A staging site you can compare side by side against production.
Week 4
Production deploy, then we wait for field data to refresh rather than declaring victory off a lab score the same afternoon.
You leave with: Before-and-after field numbers and a change log naming every edit.
Performance is the easiest service to fake with a screenshot. Two of these are reasons we decline.
Something not covered? Ask directly and you’ll get a straight answer, not a calendar link.
No, and be suspicious of anyone who does. A 100 is achievable by stripping functionality and gaming a lab test on one simulated connection. We optimise Core Web Vitals as measured on your real visitors, which is what Google ranks on. That usually lands in the 85 to 98 range with everything still working.
We won’t put a number on it before the audit, because it depends entirely on what’s currently wrong. What we will do is state expected impact per fix in the audit, before you commit to implementation. If the audit finds your site is already well optimised, we’ll tell you and you’ll have spent $600 instead of $2,400.
Often, yes. Plugin count itself isn’t the problem, but a plugin loading 400KB of JavaScript on every page to power one widget is. The audit names each one with its measured cost so it’s your decision with real numbers attached, not our preference.
Because Core Web Vitals field data is a rolling 28-day window. Anyone showing you improvement the same afternoon is showing you a lab score, which can move without a single real visitor being faster. We’d rather wait and show you the number that counts.
We work with what you have first. Hosting is occasionally the real bottleneck, usually shared hosting with a slow database, and if so we’ll show you the server response numbers that prove it. We don’t resell hosting, so we have nothing to gain from telling you to move.
Not in the one-time projects. Sites regress, usually within a quarter, when someone installs a plugin or uploads an unoptimised image. The retainer exists for that. If you ship rarely, buy the project and re-audit in a year instead.
Give us two or three URLs that carry your revenue and Search Console access if you have it. We’ll tell you roughly where the time is going before you pay for anything.
