Slow main content
Hero images and fonts, render-blocking CSS and JS, server response time and caching.
Performance engineering services
Performance engineers who baseline, fix and prove the gains.
Epsilon Delta Wizard is a performance engineering service. Our engineers baseline your site, fix what slows it down and report the measured gain. Fixed term, fixed cost, and every gain measured on real users.
- minimum improvement
- 30%*
- what most customers see
- 50%+
- of customers come back
- 80%
Engagement · shop.example.comSample engagement
- Week 1Audit and baseline Done
- Weeks 2–4Implement fixes In progress
- Week 5Report impact Planned
LCP p75 so far
Baseline3.6 s
Now2.4 s
▼ 33%
Engineer's note · Hero image preloaded and two tag scripts deferred. LCP is already past the 30% target.
How it works
Five weeks from baseline to proof.
Every engagement has a fixed term and a fixed cost, agreed before we start. You always know what is being worked on and what it has changed.
-
Week 1
Audit and baseline
We record real-user Core Web Vitals for your key pages with Gemini, or Orion for apps, then audit the code, scripts, images, fonts and servers behind them.
You get Baseline report and a fix list ranked by impact
-
Weeks 2–4
Implement fixes
Our engineers work with your team on the biggest wins first, through pull requests or pairing, and check each change against the baseline as it ships.
You get Shipped changes, each with its measured effect
-
Week 5
Report impact
We compare the same pages and the same real-user metrics with week 1, so the gain is measured, not estimated.
You get Impact report: before, after and what drove it
-
After
Coaching and support
Complimentary follow-up support and long-term coaching, so new releases keep the speed you paid for.
You get Guardrails and alerts for future releases
What we fix
We start where real users feel it most: the slowest metric on the pages with the most traffic.
Slow clicks and taps
Long main-thread tasks, heavy third-party and tag-manager scripts, costly re-renders.
Layout jumps
Images, ads and banners without reserved space, late fonts and injected content.
Slow servers
CDN and cache rules, redirects, backend calls and edge rendering.
Too much JavaScript
Bundle size, unused code, hydration cost and script loading order.
Mobile app speed
Cold start, screen load, jank and API time on Android, iOS and Flutter, measured with Orion.
Impact report
The report you get in week 5.
Real-user p75 values for the same pages, before and after. These numbers are a sample, not a customer result.
shop.example.com · /product/:idSample data · real-user p75
| Metric | Baseline | After | Change |
|---|---|---|---|
| LCP p75 | 3.6 s | 2.2 s | ▼ 39% |
| INP p75 | 340 ms | 190 ms | ▼ 44% |
| CLS p75 | 0.21 | 0.06 | ▼ 71% |
| TBT | 820 ms | 410 ms | ▼ 50% |
| Gemini Score | 58 | 81 | ▲ 23 pts |
What drove the LCP gain
Insight · Two changes delivered 70% of the gain. They shipped in week 2.
Why teams hire Wizard
Improved site speed
A faster site that delights users and passes Core Web Vitals, which Google uses in search ranking.
Positive return on investment
Faster pages lift engagement and conversions. A fixed cost means you know the price before the gain.
Continuous support
Complimentary support and long-term coaching after the engagement, so the gains last.
- minimum improvement we commit to
- 30%*
- improvement most customers see
- 50%+
- of Wizard customers hire us again
- 80%
- typical engagement, fixed term and cost
- 5 weeks
*[Confirm the terms of the 30% commitment: which metric, which pages, and how it is measured.]
Wizard compared with a one-off audit
| Capability | Wizard | Lighthouse or PageSpeed report | One-off consultant audit |
|---|---|---|---|
| Based on | Real users on your site | One simulated load in a lab | Usually lab tests |
| Fixes implemented | Yes, with your team | No, a list of suggestions | Sometimes, billed by the hour |
| Gain measured | Before and after, on real users | Lab score only | Often not measured |
| Cost | Fixed, agreed up front | Free | Usually time and materials |
| After the project | Coaching, support and alerts | None | New engagement needed |
Lighthouse and PageSpeed Insights are useful lab tools; we use them too, alongside real-user data.
Questions about Wizard
What is a website performance optimization service?
Engineers who find what makes your site slow for real users, fix it with your team and measure the result. Wizard does all three in a fixed term at a fixed cost.
How long does a Wizard engagement take?
A typical engagement is five weeks: one week to audit and baseline, three to implement fixes and one to report the impact. [Confirm typical range for larger sites.]
How is the improvement measured?
With real-user p75 values for the same pages, before and after, collected by Gemini for websites or Orion for apps. We commit to a minimum 30% improvement; most customers see 50% or more.
Do I need Gemini or Orion to use Wizard?
We use Gemini or Orion to baseline and measure the gain, and set it up as part of the engagement. [Confirm whether it is included in the price.]
Do your engineers need access to our code?
Yes, read access is enough for the audit. Fixes go through your normal review, as pull requests or by pairing with your developers.
How much does Wizard cost?
Each engagement has a fixed cost agreed before we start, based on the size of the site or app. Talk to us for a quote.
Last updated 7 Nov 2026 · Reviewed by [Name], [Role]