Speed services · Buyer's guide
Hiring a Core Web Vitals agency? Demand this deliverable list
Speed work is easy to sell and easy to fake. A serious optimization service delivers a measured audit, an itemized fix list, and before/after proof per template — not a plugin install and a screenshot of one good run.
The deliverables that matter
| # | Deliverable | Why |
|---|---|---|
| 1 | Per-template audit: LCP/INP/CLS + weight budget | Home, service and post templates fail differently |
| 2 | Itemized fix list (images, fonts, JS, CSS, third parties) | You can verify each line shipped |
| 3 | Before/after lab medians, same URL + throttling | Single screenshots prove nothing; medians do |
| 4 | Field-data confirmation plan (28 days, CrUX) | Lab wins must survive real devices |
| 5 | No-regression note: what changed, what to avoid | Stops the next plugin from undoing it all |
Questions to ask before you pay
- Which templates will you measure, and what are their current numbers?
- What exactly will you change — file by file, script by script?
- How will you prove the lift, and what happens if field data disagrees?
- What should we not install afterward to keep the gains?
Red flags
- Quotes without auditing ("$X for Core Web Vitals fix" sight unseen).
- Installs a cache/optimization plugin and calls it done.
- Reports desktop scores when your traffic is 80% mobile.
- Refuses per-template before/after numbers.
- Optimizes for the test (one URL, one run) instead of templates + field data.
What good looks like (example)
On this site the audit found a 332 KB PNG served at two sizes, six render-blocking font weights, document-level mousemove JS and a blurred sticky header. Fixes: right-sized WebP (3.3 KB + 28 KB), three async font weights, deleted tilt JS, solid header, one LCP preload. Result: 68 → 95+ Performance with identical design. Full weights and code: the before/after teardown.
FAQ
What should a speed service include?
A measured per-template audit, itemized fixes, before/after lab medians, field-data confirmation, and a no-regression note.
How do I verify the work?
Same URL, same throttling, three-run medians before and after — then Search Console field data 28 days later.
How much does it cost?
Single-site fixes typically run $200–$1,200 depending on templates and third-party mess. Anything quoted without an audit is a guess.
Want the audit on your URL?
Send your link on LinkedIn — you get per-template numbers and a fixed scope. No forms, no queue.