Listing Page CRO for a Karachi Property Portal

The Client

The client in this engagement is a Karachi-based property brokerage operating a dual-language listings portal alongside a 28-agent sales floor. The portal carried roughly 6,000 active listings at any time — resale apartments, houses, and commercial units concentrated in DHA Karachi, Clifton, Bahria Town Karachi, Gulshan-e-Iqbal, and North Nazimabad — and the brokerage’s growth model depended on it: Google Ads and Meta campaigns drove searchers onto listing detail pages, listings generated enquiries, and agents worked those enquiries toward site visits and mandate agreements.

Monthly portal traffic sat around 140,000 sessions, of which about 52,000 paid sessions landed directly on listing detail pages. That paid layer cost roughly PKR 1.15M a month, and by the time the brokerage asked WeProms Digital to intervene, leadership had a specific complaint: the portal “gets traffic but not enquiries.” The sales floor had a second complaint, which was that the enquiries they did receive were thin, duplicated, or attached to the wrong property. Both complaints were correct, and both traced back to the same under-instrumented funnel.

What made this engagement different from a typical landing-page project was the inventory structure. A developer sells a handful of projects and can hand-craft each page. A brokerage portal renders thousands of listings from agent-entered data, so conversion is a data quality problem as much as a design problem: the same template shows a complete, well-photographed, honestly priced listing and a three-line stub with no price and one blurry photo. Any conversion programme that only redesigned the template would simply make the bad listings prettier. The work therefore ran as a full conversion rate optimization programme covering the listing template, the enquiry form, listing data standards, and the routing layer between the portal and the sales floor. The client is described here in anonymized terms, and the numbers that follow are illustrative outcome ranges drawn from common patterns in this vertical.

The Problem

Six issues were suppressing enquiries on the portal:

• The enquiry form punished commitment. Nine fields on a single mobile screen — name, phone, email, city, locality, budget (typed free text), purpose, preferred property type, and a message box. Field-order analysis later showed that budget and purpose, the two most useful fields for agents, sat at positions seven and eight, after most mobile users had already abandoned.
• A third of form starters never finished. Of roughly 2,400 people who started the form each month, only about 825 submitted — a 34% completion rate. Session recordings showed users tapping the budget free-text field, being confronted with a numeric keyboard and no guidance on format, and leaving.
• 38% of resale listings hid the price. “Contact for price” dominated the resale inventory because agents feared undercutting themselves in negotiation. Whatever the reason, the effect was measurable: price-hidden listings received about 61% fewer enquiries per session than comparable priced listings.
• A quarter of the inventory was effectively unlisted. Forty-four percent of listings had fewer than three photos, and 31% were duplicate entries of the same unit uploaded by different agents. Buyers saw the same apartment three times at three different prices and distrusted all three.
• Junk enquiries polluted the funnel. Twenty-two percent of submitted enquiries were duplicates, spam, or referred to a property the enquirer had not actually viewed. Agents spent the first hour of every day sorting real interest from noise, and response times to genuine enquiries suffered.
• Nothing was measured below the page level. There was no form-field analytics, no template-level conversion comparison, and no unified view of enquiry-to-site-visit progression. Portal decisions were made by opinion, and the two teams blamed each other.

By the numbers: 1.6% of paid listing sessions produced an enquiry, cost per qualified enquiry had drifted to roughly PKR 7,400, and only 155 genuinely qualified enquiries (with budget, timeline, and intent captured) reached the sales floor each month.

Phase 1 — Funnel Diagnostic and Listing-Quality Audit (Weeks 1-2)

The first fortnight was forensic, and it followed the sequence of a structured UX audit and conversion analysis: instrument everything, then let evidence rank the problems.

Funnel instrumentation. GA4 events were wired across the full enquiry path — listing view, photo gallery engagement, filter use, CTA clicks, form start, per-field interaction, step completion, and submission — segmented by template type, device, locality, and traffic source. Field-level drop-off tracking was the critical addition: for the first time the portal could see the exact field where abandonment happened, not just that it happened.

Session evidence. Recordings and heatmaps on the top 200 paid-traffic listings showed three repeating behaviours: rage taps on the price area of “contact for price” listings, users bouncing between duplicate listings of the same unit, and mobile users abandoning the form at the free-text budget field.

Listing-quality audit. Every active listing was scored on five dimensions: price present, photo count, description length, freshness (updated within 60 days), and duplicate status. The distribution explained the conversion problem better than any design critique — paid traffic was being spent almost evenly across good and stub listings alike.

Diagnostic	Finding	Conversion effect
Price hidden	38% of resale listings	~61% fewer enquiries per session vs priced listings
Photos under three	44% of listings	Bounce 18 pts higher than 5+ photo listings
Duplicate unit listings	31% of inventory	Duplicated units converted 40% worse overall
Form field drop-off	Budget + purpose fields	Positions 7-8; 43% of starters abandoned before them
Junk enquiry rate	22% of submissions	Agent response time to real enquiries averaged 9+ hours
Mobile LCP	4.4s median on 4G	78% of sessions mobile; slow first paint on image-heavy pages

Baseline lock. Before any change went live, the baseline was frozen in a shared dashboard: 34% form completion, 1.6% session-to-enquiry, 155 qualified enquiries, PKR 7,400 cost per qualified enquiry. Every later number in this case study is measured against this lock, which is the discipline that separates a CRO programme from a redesign.

Phase 2 — Form and Listing Template Rebuild (Weeks 3-5)

Phase 2 rebuilt the two surfaces where buyers decided whether to enquire: the form and the listing template.

A progressive three-step form. The nine-field wall became a three-step flow with a visible progress indicator. Step one captured intent — purpose (buy or rent), property type, and a budget band selector (tappable ranges formatted in the crore and lakh figures Karachi buyers actually think in, from “under PKR 1 crore” upward). Step two collected contact details — name, phone with inline validation for Pakistani number formats, and optional email. Step three showed a confirmation summary of what would happen next, plus a one-tap WhatsApp handoff to the assigned agent. Counter-intuitively, the form kept more fields than a minimalist redesign would have; the difference was sequencing. Asking intent before contact details did two things: it committed the user psychologically before the “personal information” moment, and it filtered junk at step two rather than in an agent’s inbox.

A listing template that answers the first three questions. The redesign put price (formatted as “PKR 2.4 crore” with a per-month rental equivalent where relevant), covered area with a sq yd / sq ft / marla unit toggle, and locality position on a map above the fold on mobile. Photo galleries were restructured into a swipeable hero with a visible photo count, and a freshness stamp (“updated 3 days ago”) replaced the undated listings that had sat unchanged for seasons.

Deduplication and the price policy. A matching pass consolidated duplicate unit listings into a single canonical record, keeping the strongest photos and the freshest price and routing enquiries to the listing owner. Alongside it, a price-display policy was introduced: resale listings could show a band (“PKR 2.3–2.5 crore”) where an exact figure was sensitive, but “contact for price” was removed as an option for any listing that wanted paid traffic. Price coverage rose from 62% to 89% of active listings within four weeks.

Performance work. Listing pages loaded 4.4-second LCP on median 4G, driven by uncompressed agent photography. An image pipeline — WebP conversion, responsive sizes, lazy loading below the hero, and a cache layer on the template shell — brought mobile LCP to 2.1 seconds, which helped both conversion and the portal’s organic visibility.

Routing and the SLA. Every enquiry was auto-assigned to the owning agent with a 15-minute response target during business hours, and the thank-you step told the buyer exactly that: “Your agent will call within 15 minutes.” Response-time data flowed back to the dashboard weekly.

Phase 3 — Experimentation and Routing Optimization (Weeks 4-8)

With the rebuild live, weeks four through eight ran a structured test programme across the templates. Each test ran to a pre-decided sample before a winner was declared.

Experiment	Variant A	Variant B	Winner
Step order	Contact first, intent second	Intent first, contact second	Intent first (+4.1 pts completion)
Price display	Exact figure only	Band where exact unavailable	Band fallback (kept coverage high)
CTA copy	”Submit enquiry"	"Get price and visit details”	Specific CTA (+1.8 pts)
Photo threshold for paid traffic	Any listing eligible	5+ photos required	Quality gate (+0.7 pts session-to-enquiry)
Confirmation step	Simple thank-you	Thank-you + WhatsApp handoff	WhatsApp handoff (visit rate up)

Two findings shaped the programme more than the rest. First, asking intent before contact details outperformed the reverse by a clear margin — buyers would happily answer three property questions while still deciding whether to share a phone number. Second, the photo threshold test pointed toward a structural insight: paid traffic converted better when it was only sent to listings that met a quality bar, which became the foundation of Phase 4’s gating system rather than a one-off win.

Routing optimization ran in parallel. Enquiries that sat unanswered past the 15-minute SLA escalated to the team lead and then to a shared floor queue, and the junk rate was attacked at the source — deduplication, step-two filtering, and phone validation cut duplicate and spam submissions from 22% to under 6%. Agents began trusting the portal’s enquiries again, which showed up where it mattered: the share of qualified enquiries converting to booked site visits rose from 39% to 41% even as volume grew.

Phase 4 — Quality Gating and Compounding (Weeks 8-12)

The final phase turned the wins into a system the brokerage could run without us.

Quality-score gating. The Phase 3 finding became policy: paid campaigns now route only to listings that clear a quality score (price present, 5+ photos, fresh within 60 days, not a duplicate). Listings that fail are still live on the portal for organic discovery, but ad spend no longer subsidizes them. This single rule linked the agents’ data-entry discipline to the marketing budget, and listing quality coverage became a weekly reviewed metric rather than an aspiration.

Rollout beyond resale. The winning template and form variants were extended to the rentals and commercial sections, which had inherited the old single-column form. Rentals converted differently — rental enquirers abandoned at the move-in date field — so the field order was retested rather than copied blindly, and the rental completion lift settled slightly above the resale result.

Dashboard and cadence. The shared dashboard consolidated completion rate by template and device, cost per qualified enquiry by source, junk rate, SLA compliance, and enquiry-to-visit progression. A weekly experiment review kept one test running at all times, and a monthly quality review walked the sales floor through which listing improvements moved which numbers — closing the loop between the people who enter listing data and the people who work the enquiries.

Handoff. The instrumentation, test playbook, and quality-gating rules were documented and handed to the portal’s in-house team, so the programme compounds under their ownership rather than depending on an agency’s calendar.

Final Results

Measured against the Phase 1 baseline lock, at the 90-day mark:

Metric	Baseline	90 days	Change
Enquiry form completion rate (of starts)	34%	51%	+51% relative
Listing session to enquiry conversion	1.6%	2.6%	+62% relative
Monthly enquiry submissions	~825	~1,330	+61%
Junk / duplicate enquiry rate	22%	under 6%	-16 pts
Monthly qualified enquiries	155	225	+45%
Cost per qualified enquiry	PKR 7,400	PKR 5,100	-31%
Monthly site visits booked from enquiries	61	92	+51%
Listings displaying a price	62%	89%	+27 pts
Mobile LCP on listing pages	4.4s	2.1s	-52%

Every number traces to a phase: completion and junk-rate gains to the Phase 2 form and deduplication work, the session-to-enquiry and quality-coverage gains to the template and price policy, the incremental test wins to Phase 3, and the durability of the whole stack to Phase 4’s gating and handover. Ad spend was held flat throughout, so the enquiry growth is a conversion outcome, not a spend outcome.

What Made This Work

1. Listing data quality was treated as a conversion lever. The largest single lift came not from the form but from the price policy and deduplication — proof that on a brokerage portal, the inventory is the interface. Treating “contact for price” as a conversion bug rather than a negotiation tactic changed the economics of the whole portal.

2. The form sequenced commitment before disclosure. Asking property questions first and contact details second outperformed a shorter contact-only form, because buyers invest before they disclose, and the intent answers doubled as pre-qualification for the sales floor.

3. Quality gating aligned agents with marketing. Once paid traffic only flowed to listings that met the bar, agents had a direct, personal reason to enter prices and upload photos. Coverage rose from 62% to 89% in a month — a governance outcome, not a design one.

4. The routing SLA made enquiries worth answering. Cutting junk from 22% to under 6% and promising a 15-minute callback rebuilt the sales floor’s trust in portal leads, which raised the enquiry-to-visit rate at the same time as volume grew.

5. Performance work served conversion directly. Bringing mobile LCP from 4.4s to 2.1s on image-heavy listing pages removed a tax on the 78% of sessions that arrive on phones on Pakistani mobile networks.

What Teams Can Apply

1. Audit your inventory before your template. On any listings-driven portal — property, vehicles, classifieds — conversion is capped by the worst listing your ads touch. Score listings on price presence, photos, freshness, and duplicates, and fix the data before the design.

2. Gate paid traffic behind a quality bar. Let weak listings live on organic; stop buying clicks for them. This one rule converts your data-quality policy into a budget decision, which is the only version of data governance most organizations actually enforce.

3. Sequence your form for commitment. Put intent and qualification questions before contact fields, use tappable band selectors instead of free-text budget fields, and validate phone formats inline. In the Pakistani market, crore/lakh-formatted ranges convert where free-text fields fail.

4. Kill duplicates at the source. The same unit listed three times at three prices burns trust and budget alike. A matching pass with a canonical record is unglamorous infrastructure with an outsized conversion return.

5. Report enquiry quality, not just volume. Track junk rate, SLA compliance, and enquiry-to-visit progression alongside completion rate. When the sales floor sees lead quality improving, follow-up improves with it, and the funnel’s back half stops leaking.

The same sequence — audit inventory, rebuild the enquiry path, test with discipline, gate spend on quality — applies to brokerages and portals beyond Karachi. WeProms Digital runs it across the Pakistani property vertical; you can read more sector context in our overview of real estate marketing in Pakistan, and the engagement pattern above is representative rather than a single named client’s audited results.