Same traffic, double the enquiries — in six weeks.
How a conversion-first website rebuild turned Vertex Real Estate's existing visitors into 2.4× more enquiries — without spending one extra dirham on traffic — and what twelve weeks of testing added on top.
Plenty of traffic. A website that wasted it.
Vertex Real Estate is a mid-sized brokerage: thirty-eight agents, around 2,400 active listings, and a marketing budget that had bought real visibility — portal placements, Google Ads, and a solid social presence pushed roughly 60,000 visitors a month to the website. The leadership's instinct was to buy more. The audit said the opposite: Vertex didn't have a traffic problem, it had a conversion problem, and buying more visitors meant paying twice for the same leak.
The numbers made the case brutally. Sixty thousand visitors were producing under 500 enquiries a month — a 0.8% conversion rate in a market where 2% is ordinary. On mobile, where 71% of the traffic arrived, the rate was 0.4%. Every campaign the brokerage ran, on any channel, terminated in the same underperforming asset. Fixing the website wasn't one option among many; it was the multiplier on every dirham already being spent.
Seven seconds, thirty fields, zero trust.
Session recordings told the story better than any analytics chart. Mobile visitors waited 6.8 seconds for a homepage dominated by an auto-playing video they immediately scrolled past. The property search — the only thing most visitors came to do — hid behind a hamburger menu, reset its filters on every back-button press, and re-loaded the entire page for each refinement. Users who survived to a listing found gorgeous photography and then a wall: an enquiry form demanding name, email, phone, nationality, budget, and message before revealing whether the agent would even respond.
Trust signals — the currency of high-value transactions — were nearly absent. Agent profiles had no photos or track records. There were no market reports, no transaction histories, no reviews. For a purchase where the average visitor was committing a million dirhams or more, the site offered less reassurance than a food-delivery app.
And measurement was broken in a way that flattered everyone: phone calls (60% of real enquiries, per the front desk) weren't tracked at all, form submissions fired duplicate events, and nobody could connect a lead to the campaign that produced it. The brokerage was flying a plane where the fuel gauge read full regardless of the tank.
Rebuild around the visitor's job: find, trust, enquire.
A property website has one job chain — help visitors find the right property, trust the people selling it, and raise a hand with minimum friction. Everything in the rebuild either served that chain or didn't ship. Then, instead of declaring victory at launch, twelve weeks of measured testing squeezed the remaining gains.
Session recordings, heatmaps, funnel analysis, and interviews with the agents who take the calls. Every design decision in the build traced back to an observed behaviour, not a stakeholder preference.
Mobile-first search on the first screen, instant filtering, listing pages restructured around trust, enquiry friction cut to two fields plus WhatsApp. Launched behind the old URL structure so rankings carried over intact.
A/B tests on enquiry CTAs, agent placement, and listing layouts — one variable at a time, called only at significance. Call tracking finally connected leads to campaigns, reshaping the ad budget.
What the rebuild actually changed.
Search on the first screen, results in milliseconds
The new homepage opens with the search bar — community, budget, bedrooms — because recordings showed 83% of visitors trying to search within ten seconds. Filtering became instant: results update as filters change, no page reloads, and the back button behaves like a sane person expects, preserving every filter. Saved searches and a "notify me" option catch the majority of buyers who aren't ready to enquire on day one — an audience the old site simply lost. Mobile search-to-listing progression tripled in the first month.
Listing pages rebuilt around trust
Each listing now answers the questions buyers actually weigh: real service-charge figures, community context with commute times and school proximity, comparable recent transactions, and honest photography ordering (the actual view before the lobby render). The agent moved from a footnote to a card with photo, languages, years in the community, and recent sales — because a million-dirham enquiry is sent to a person, not a form. Time-on-listing rose 74%, and agents reported a change they could feel: callers arrived pre-qualified, already knowing the service charge and the comps.
Two fields and a WhatsApp button
The thirty-field-feeling enquiry form became two fields — name, phone — plus one-tap WhatsApp with the listing reference pre-filled. Agents feared junk leads; the data said otherwise. Enquiry volume rose 2.4× and qualified viewings rose 31%, because friction wasn't filtering out time-wasters — it was filtering out busy, serious people who wouldn't type an essay on a phone keyboard. The qualification moved to where it belongs: a human conversation within minutes, not a form demanding trust the site hadn't yet earned.
Performance engineering, not performance theatre
Getting 6.8 seconds down to 1.2 was a demolition job: the auto-playing video went, four redundant script bundles went, and 2,400 listings' worth of images moved to a CDN with responsive sizing and lazy loading. Search runs against an indexed API instead of rendering the whole catalogue. The speed shows up twice on the dashboard: mobile bounce fell 64%, and Google's crawler — which had been timing out on listing pages — indexed the full inventory for the first time, lifting organic listing traffic 28% with no new content at all.
Twelve weeks of disciplined testing
Post-launch, we ran one test at a time and let significance decide. Winners included a sticky mobile enquiry bar (+22% enquiries), agent WhatsApp response-time badges (+11%), and a mortgage-estimate line under the price (+9% on mid-market listings). The most instructive loser: a homepage hero video the brokerage loved, which cut search engagement by 15% and died in a week. The testing rhythm mattered as much as any single result — it converted website decisions from taste debates into a number both sides could read.
Measurement that follows the money
Call tracking numbers per channel, WhatsApp click attribution, and CRM integration finally connected each lead to its source campaign — and the picture redrew the media plan. One portal partnership, consuming a third of the budget, was producing 6% of qualified viewings; two Google Ads campaigns dismissed as underperformers were quietly driving the cheapest viewings in the account. Reallocating that spend, with no budget increase, is where the final +31% in qualified viewings came from. The website rebuild paid for itself; the measurement rebuild kept paying.
The screenshots behind the numbers.
Google Analytics and Search Console data straight from the client's own accounts — plus the work itself.
Your numbers could be next.
Get a free audit of your website's conversion funnel, speed, and tracking — delivered in 48 hours, with the quickest wins ranked.
The traffic never changed. Everything else did.
Every improvement here compounded channels the brokerage was already paying for: the ads converted better, the portals converted better, organic converted better. That's the defining economics of conversion work — it doesn't compete with your marketing budget; it multiplies it.
Three lessons any lead-driven business can steal.
Doubling conversion is mathematically identical to doubling traffic — and dramatically cheaper, because you pay for it once instead of every month. Any business about to raise its ad budget should first ask what percentage of current visitors leak out, and why.
The long form wasn't protecting agents from time-wasters — it was protecting time-wasters from agents, because serious buyers on phones gave up first. Cut the form, answer fast, and qualify by talking. Enquiries rose 2.4× and quality rose with them.
The beloved hero video died not because an agency said so, but because forty recordings showed real buyers scrolling past it and a test measured the damage. Watching ten real users is worth ten opinions from any meeting room — including ours.