Flux Full Circle
•
4 min
Turning booking-page guesswork into tested, validated wins
Confidentiality notice
Client names are abbreviated and commercially sensitive figures have been generalised.
CHALLENGE
Across a ~40-client luxury travel portfolio, booking conversion decisions (CTA placement, copy, page structure) were being made on instinct, not evidence, including a live rates page with no call to action or route into the booking journey.
STRATEGY
RESULTS
CHALLENGE
Flux runs CRO for a portfolio of roughly 40 luxury travel clients, lodges, hotels, and tour operators, each with a website built to drive direct bookings over third-party channels. Before this program, most CTA and layout decisions were made on convention: a "Book Now" button went wherever it looked good, hero sections followed whatever the template shipped with, and nobody had tested whether any of it actually converted better than the alternative.
The clearest evidence of the gap: a full mobile audit of Client JSL's site surfaced rage clicks concentrated on the cookie consent banner, dead clicks on non-interactive images across multiple lodge pages, and mobile users rarely scrolling past 10-20% of any page. Separately, a review of Client JSL's rates page (the page a visitor lands on specifically to check pricing before booking) found it had no call-to-action at all. A live, shipping page with zero path to convert.
The problem wasn't one broken page. It was the absence of a system for finding pages like it.
STRATEGY
Part one: the CTA testing program, three tests, one lever
Rather than optimise pages one at a time, I used three client experiences as a testing ground for the same underlying CTA hypothesis. We wanted to test how much impact the primary booking CTA had on conversion. This section is specifically about that lever, placement, alignment, and copy, tested and validated independently. The redesigns in Part Two go further and apply broader UX fixes beyond CTA testing alone.
01: Placement
Client ENT
Testing a CTA placed in the hero section against the same CTA placed in the site navigation.
Control: CTA in navigation only. Variant: CTA placed directly in the hero.
The hero placement converted at 20% versus 8.33% for the navigation control, a 140% relative lift. Statistical significance came in at 63%, below the 95% threshold a rigorous test would want before rolling out unconditionally, so I treat this as a strong directional signal rather than a settled result on its own.
02: Alignment
Client SAM
Centre-aligned CTAs produced a 78.9% relative lift, with 97% confidence.
Control: default-aligned CTA. Variant: centre-aligned CTA, tested across 422 (control) and 418 (variant) visitors.
Across 422 control visitors (9.72% conversion) and 418 variant visitors (17.46% conversion), centre-aligned CTAs won by 78.9%, at 97% statistical significance. This is the most rigorous result in the program, real sample sizes, a clean split, and a confidence level well above standard thresholds.
03: Copy
Client 21N
Both tests moved in the same direction: +16.6% and +24.3%. The consistency suggested the wording was worth carrying forward, although the evidence was less rigorous than the alignment test.
Same button, same placement, different word. Tested twice, on two different pages, same result both times.
Both replications moved in the same direction: +16.6% and +24.3%. Testing the same hypothesis twice and getting the same answer both times is what turns a single result into a validated pattern.
Three different levers, three independent clients, one consistent finding: don't guess at the primary CTA,
test it.
Part two: applying the findings, two full redesigns
Client LAB: a persistent prompt with measurable commercial impact
“Book Directly With Us and Get a Discounted Rate”, positioned under the hero on load, then pinned to the top of the screen as the visitor scrolls, staying visible for the entire session rather than being seen once and forgotten.
Unlike the CTA tests above, this wasn't a refinement of an existing CTA, it was a new, persistent element designed to stay in view for the entire browsing session rather than being seen once and scrolled past. Comparing a 30-day window before and after:
Bookings increased by 174% alongside a 128.5% increase in sessions. The result suggests the persistent prompt may have contributed to stronger booking intent, although the before/after comparison doesn't establish causation.
ALIGNMENT
97%
statistical significance on the strongest CTA test (Client SAM)
X
X%
increase in booking sessions following Client JSL's mobile audit
DIRECT BOOKINGS
174%
increase in direct bookings following Client LAB's persistent booking prompt
X
X
a high-leverage conversion element on the booking page
