Desktop performance from 53 to 95, in three weeks

The brief

Healing Hands Hospital is a multi-speciality hospital in Nagpur. Patients arrive at the site to do one of three things: understand whether the hospital treats their condition, look up a consultant, or book an appointment.

The site did all three badly. It was not broken in a way anyone would report – it simply cost the hospital appointments quietly, which is the most expensive kind of broken.

What we found

We audited the live site before quoting. The performance numbers were the headline, but the functional faults were more urgent.

• The appointment button did nothing. The primary call to action on the home page did not link to the booking section. Every visitor who acted on the main conversion path hit a dead end.
• Mobile performance scored 47 out of 100, with a largest contentful paint of 17.9 seconds. On a mid-range phone the hero was still resolving well after most visitors would have left.
• Desktop scored 53, with 460ms of total blocking time and a speed index of 3.8 seconds – the page appeared, then ignored you.
• Empty sections on the home page left large gaps that broke the reading flow and made the site feel unfinished.
• Spelling and capitalisation errors in the doctor’s profiles. On a hospital site, a misspelt speciality is not a typo – it is a credibility problem at the exact moment a patient is deciding whether to trust you.
• An age-restricted YouTube embed that refused to play in place and pushed visitors off to YouTube instead.

What we changed

Three weeks, fixed scope. The functional faults were fixed first, because a fast site with a broken booking button is still a site that takes no bookings.

• Appointment button wired to the booking flow, and the booking path tested end to end on mobile and desktop
• Home page rebuilt to remove the empty sections and tighten the sequence a patient reads
• Image pipeline reworked – correct formats, correct sizes, and the hero prioritised instead of competing with everything else
• Render-blocking scripts deferred and unused work removed, which is where the 460ms of blocking time went
• Doctors’ profiles proofread, with specialities spelled and capitalised consistently
• Restricted video embed replaced so the content plays in place
• Technical SEO and best-practice issues cleared as part of the same pass

The numbers

Measured on Lighthouse before the work and again on 15 Dec 2025, on the same page.

The desktop result is the clean one: a 95, with blocking time at zero and the largest element painting in 1.2 seconds. Best practices and SEO both reached 100.