Diagnostic brief · Law Firm Sites
Web hosting for lawyers — Dividing a slow page into the half we own and the half you do
Before anyone changes host, someone has to establish whether the first byte or the last paint is what is slow, because they have different causes and different owners.
The short answer
Split the measurement before you diagnose anything: TTFB is a server number and LCP is a visitor number, and a slow page is one, the other, or both. Responsibility follows a clean seam — hosting owns the floor, because nothing can render before the first byte arrives, and everything above that floor belongs to the page: image discipline, script weight, and space reserved so nothing jumps while it loads.
Firms with a good first number and a poor second one are extremely common, and changing host will not move them. HostingFast owns the floor with NVMe storage behind server-level LiteSpeed caching from a couple of dollars a month, with free SSL, free migration, mailboxes on your domain and a human on support at any hour.
By the HostingFast team · Reviewed 24 August 2026
99.9%
Uptime commitment, monitored
24/7
Humans on support
Free
SSL, every plan
NVMe
Storage class on every plan
A law firm site is a set of practice-area pages, a few solicitor profiles and an enquiry route. Structurally simple, and yet the commonest complaint is that it feels slow — a description that cannot be acted on until it is split into numbers.
The split is worth doing carefully, because the two halves have different economics. The floor is cheap to fix and mostly a matter of whether the page is being cached at all. The ceiling is a build decision: a hero image, a theme framework, a chat widget, a font set. Confusing them produces a migration that changes nothing and a bill that changes something.
What follows is the diagnostic in order, and the shape of the evidence a partner is likely to accept at the end of it.
Establish the floor first
Run `curl -s -o /dev/null -w '%{time_starttransfer}' https://yourdomain.example` five times against a practice-area page and keep the median. Google's good boundary for the first byte is 800 ms. A cached page coming off LiteSpeed on NVMe, from a UK connection to our London origin with a 5–15 ms round trip, should sit a long way inside it.
If it does not, establish whether the page is cached before drawing any conclusion about hardware. Firm sites frequently run a live-chat plugin or a case-results feed that sets a no-cache header sitewide as a side effect, which turns every visit into a full PHP render. That is a configuration finding, not a hosting one, and it is the cheapest fix on this page.
Then budget the practice-area template
Because the pages share a template, they share a budget. Load the busiest one throttled to a mobile profile, note the total transfer and identify the LCP element, then set a ceiling that keeps LCP at or under 2.5 seconds on that connection. Once that ceiling exists as a number, a new practice-area page either fits it or does not, and nobody has to adjudicate whether the site feels heavy.
The usual findings are unremarkable and expensive: a stock photograph of a courtroom at four times its displayed size, a chat widget loading on every page including the ones with no chat, and a second font family used only in the footer. None of those is a hosting decision; all of them are on the critical path for a client comparing three firms on a phone.
The stages before your server even starts
Two costs sit before the first byte and are frequently ignored. Redirect chains: a request to the bare domain that goes to http, then https, then www, then the page has paid for three extra round trips before anything useful happens. Collapse it to one redirect and the saving is real on a mobile connection.
And the handshake. `curl -w` reports namelookup, connect and appconnect separately, so you can see DNS, TCP and TLS as distinct line items rather than lumping them into a vague slowness. SSL is included and renews itself here, so the certificate never becomes the thing that lapses and frightens a client away — but the handshake still costs time, and a redirect chain multiplies it.
The evidence a partner will accept
Quote what you measured, not what a host claimed. Record median TTFB from curl and the Core Web Vitals field data from PageSpeed Insights, both dated, and attach the method. Google's published boundaries — 800 ms for the first byte, 2.5 seconds for LCP at the 75th percentile — give the reader a yardstick they already recognise.
The 75th percentile is the part people skip, and it is why a site can look fine on a partner's laptop and fail in the report: three quarters of visitors must be inside the threshold. Present both numbers, say which one is failing, and the decision about whether to spend money on hosting or on the build makes itself.
Our half of the seam
The floor is ours and it is specified rather than promised: NVMe storage on every tier, LiteSpeed caching inside the web server, free SSL that renews itself, and a 99.9% uptime target monitored around the clock.
The ceiling is yours, and staging plus SSH with Git and Composer are there so you can work on it properly — change, measure, keep or revert, without a client meeting the experiment.
- First byte answered from cache, not from PHP
- NVMe storage on the cheapest plan as well as the dearest
- Free SSL issuing and renewing automatically
- 99.9% uptime as the target, monitored round the clock
Why HostingFast
Standard on every plan
A floor stated as a number
Cached responses off LiteSpeed and NVMe, measurable with one curl command against Google's 800 ms boundary. No interpretation required.
A certificate that never interrupts
Free SSL on every plan, reissued long before expiry, so an interrupted browser warning is never what a prospective client meets first.
Staging for the build half
Clone the site, remove the chat widget, measure the difference. Two numbers settle a question that two opinions cannot.
Command-line access on shared plans
SSH with Git and Composer, so WP-CLI runs and a change is scripted and reversible rather than clicked and half-remembered.
Uptime with a remedy attached
A 99.9% target monitored around the clock; if a month falls short through a fault on our side, the terms provide a pro-rated credit.
Support that will trace it
Humans at any hour, and PHP X-Ray on CloudLinux Pro on the Nitro tier when the answer needs to be a function name rather than a theory.
Quick Start
From order to online
- 1
Measure the floor
Five curl runs against a practice-area page, median kept, compared to the 800 ms boundary. Also check the cache status header while you are there.
- 2
Measure the ceiling
Throttled mobile load of the same page: total transfer, LCP element, INP. Now you have two numbers with two different owners.
- 3
Fix the one that is failing
A failing floor is usually a caching header or a migration. A failing ceiling is images, scripts and fonts. Doing both at once tells you nothing about either.
Built In
Loaded onto every plan
- Median TTFB from five curl runs on a practice-area page, dated
- Cache status header checked on a logged-out request to the same page
- Total transfer and LCP element captured on a throttled mobile load
- A byte budget written for the practice-area template and defended
- Redirect chain collapsed so the bare domain reaches the page in one hop
- curl timing read stage by stage: namelookup, connect, appconnect, starttransfer
- Chat and case-feed widgets checked for setting no-cache headers sitewide
- Free SSL confirmed as issuing and renewing itself on every plan
- Field data quoted at the 75th percentile, not from a partner's laptop
- The failing half named explicitly before any money is spent
Frequently Asked
What people ask us most often
Our first byte is quick but the page still feels slow. What now?
That is the commonest pattern there is, and it means hosting has already done its part. Move to the ceiling: load the page throttled to a mobile profile, identify the LCP element, and read the total transfer. It is nearly always an oversized hero image, a chat widget loading everywhere, or a second font family. None of those change if you change host.
How much does a redirect chain actually cost?
Each extra hop is a full round trip before any content arrives, and on a mobile connection that is tens to hundreds of milliseconds each. A request that goes bare domain, then http, then https, then www has paid three of them. Collapse it so the canonical destination is reached in one redirect, then confirm with curl that the chain is gone rather than assuming the configuration took.
What is the 75th percentile and why does it keep failing us?
It means three quarters of your real visitors must be inside the threshold, not the average visitor. A site that is fast on office broadband and dreadful on a mid-range phone on mobile data can average well and still fail, because the slow quarter is the quarter being graded. Measure on a throttled mobile profile if you want your lab runs to predict the field report.
Can you show where a slow request spends its time?
Yes, and it is the fastest way to end a disagreement. curl reports the stages separately, so DNS, TCP, TLS and server processing are individual line items. On the Nitro tier, PHP X-Ray on CloudLinux Pro goes further and traces a slow request down to the function, which turns an argument about hosting into a line number in a plugin.
Keep reading
Restaurants — Hosting Guide
A lunchtime spike, and the one route a page cache can never answer.
Online Magazines — Hosting Guide
Publishing at volume: archive queries, image weight and what a reader waits on.
SSL Certificate (Glossary)
The certificate, the handshake, and how to confirm the renewal actually happened.
Plesk Reseller Hosting — Built for Law Firms
White-label accounts on the same measured platform, when the sites belong to clients.
Plesk Hosting
The same NVMe and LiteSpeed stack behind a Plesk panel instead of cPanel.
Changing hosts? Run through our checklist first.
A straightforward sequence for a switch your visitors never feel: which files move first, how to shift email across without losing a single message, the right moment to repoint DNS, and the two mistakes behind almost all the downtime we get asked to rescue.
Name the failing half before you spend.
NVMe and server-level LiteSpeed for the floor, free SSL that renews itself, and staging plus SSH for the half that is yours.
View Plesk Reseller Hosting — Built for Law Firms plans