Skip to content
Get Ranked On TopIllustrative case studySignal plan

A fictional business. The figures are a modelled scenario, not a client result.

WAHAT KEYS · 300-APARTMENT HOLIDAY-HOME OPERATOR · DUBAI MARINA

How a Dubai Marina holiday-home operator got Google to see 300 apartments in four languages

From brand-only to +210% organic bookings

10 months · Middle East & Africa · Signal plan

  • Server-side rendering
  • Hydration budget
  • Crawlable URLs
  • Hreflang & schema

Listing pages Google could read · month 10

1,180of 1,200

A barjeel catches the Gulf breeze and sends it down into every room below. Server-side rendering does the same for content.

01THE CHALLENGE

Beautiful rooms. No air reaching them.

The listing site rendered only in the browser. Googlebot saw empty shells: 300 apartments in four languages were soft 404s, and only the brand name ranked.

Two years earlier Wahat Keys had rebuilt its booking site as a single-page app. In a browser it looked superb: swipeable galleries, live availability, prices in dirhams, euros and roubles. But every URL returned the same almost empty HTML document and waited for 2.3 MB of JavaScript to build the page. Google indexes what it can render, whenever it gets round to rendering it; faced with 1,200 near-identical shells, it mostly didn't. Guests searching for a Marina apartment in English, Arabic, Russian or German found portals and competitors instead.

What any listing URL returned

<title>Wahat Keys</title>
<div id="root"></div>
<script src="/static/js/main.4f9c1e.js">
  • 1.9 KB of HTML
  • 0 words of content
  • 2.3 MB of JavaScript before anything shows
  • 1,200 pages sharing one title

02DIAGNOSIS

What the audit found behind the walls.

RAW HTML

0 words

Fetched without JavaScript, every one of the 1,200 listing URLs returned a 1.9 KB shell: a logo, an empty root div and a script tag. Titles, prices, descriptions and reviews existed only after the bundle had run.

Finding 1 of 4

SEARCH CONSOLE

1,140

1,140 listing URLs were reported as soft 404 or “Crawled – currently not indexed”. Google saw near-identical thin pages and dropped them; only 60 language versions, mostly older English listings, made it into the index.

Finding 2 of 4

JAVASCRIPT

2.3 MB

One 2.3 MB bundle loaded every feature on every page. Mobile LCP at p75 was 5.8 s, and the rendering service sometimes timed out before the availability API had answered.

Finding 3 of 4

ROUTING & HEAD

1,200 hash URLs

Apartments lived at hash URLs (#/apartment/2312), which Google treats as a single page; internal links were click handlers, not <a href>. Hreflang, canonicals and titles were injected after hydration, if at all.

Finding 4 of 4

03THE STRATEGY

Four fixes, one direction of airflow: server first.

Built on our Signal plan: the technical engagement for sites where crawl, rendering and architecture decide the outcome.

  1. 01

    Server-side rendering

    Listing pages moved to server rendering, with all 1,200 URLs pre-rendered and regenerated whenever a price, photo or description changes. Every response now carries the full content as HTML: about 640 words, 38 crawlable links and the main photo with explicit dimensions.

  2. 02

    Hydration budget

    Only three blocks need to be interactive: the photo gallery, the availability calendar and the map. They hydrate as islands, the calendar only when it scrolls into view; everything else stays plain HTML. JavaScript per listing fell from 2.3 MB to 380 KB.

  3. 03

    Crawlable URLs

    Hash routes became real paths per language and neighbourhood (/ar/dubai-marina/…). A small script forwards old #/ links, internal navigation uses real <a href> links, and one XML sitemap per language lists every live apartment.

  4. 04

    Hreflang & schema

    Hreflang for en, ar, ru and de plus x-default, self-referencing canonicals and unique titles are written into the server HTML. Each listing carries VacationRental structured data with address, capacity and rating; delisted units now return 410 instead of an empty 200.

04RENDER VS RAW

Switch JavaScript off. Which rooms does Googlebot actually receive?

Each room is one content block of a listing page. Air is HTML that arrives in the server's first response, before any JavaScript runs. Turn JavaScript off to see what a crawler's first pass reads, then move rendering to the server and watch the air reach every room.

JavaScript

Rendering

  • Head: title, hreflang, schemaBuilt by the browser
  • Title & priceBuilt by the browser
  • Photo galleryBuilt by the browser
  • DescriptionBuilt by the browser
  • AmenitiesBuilt by the browser
  • Availability calendarBuilt by the browser
  • Guest reviewsBuilt by the browser
  • Map & neighbourhoodBuilt by the browser
  • Air: content in the server HTML
  • Hydrated: interactive once JavaScript runs

Furnished, but airless. Rooms with air: 0 of 8.

Rooms with air: 0 of 8

Furnished, but airless

In a fast browser the bundle builds every room after 5.8 s. The HTML response, though, has not changed: Google has to queue the page for rendering, fetch 2.3 MB of scripts and hope the availability API answers in time.

Words in the HTML response
0
HTML response
1.9 KB
JavaScript run
2.3 MB
Links in the HTML
0
LCP (p75, mobile)
5.8 s
The HTML response for this URL
<head><title>Wahat Keys</title></head><body><div id="root"></div><script src="/static/js/main.4f9c1e.js"></script></body>

Identical to the JavaScript-off response: the rooms on screen were built afterwards, in the browser.

05THE TIMELINE

10 months, room by room

Listing URLs indexed, of 1,200
  1. Month 1

    Render audit

    Raw and rendered HTML compared for all 1,200 listing URLs; log files and Search Console exports matched to each apartment.

  2. Months 2–3

    URLs & routing

    Hash routes replaced by real paths per language, a forwarding script for old links, one XML sitemap per language.

  3. Months 4–5

    Server rendering

    English listings switched to server rendering first, in month 4; Arabic (right-to-left), Russian and German followed in month 5.

  4. Months 6–7

    Hydration budget

    Gallery, calendar and map split into islands and the 2.3 MB bundle retired. Core Web Vitals monitored per language folder.

  5. Months 8–10

    Head & schema

    Hreflang, canonicals, unique titles and VacationRental data in every response; delisted units return 410. Then: measure and refine.

06RESULTS

Fresh air in all 1,200 rooms.

Organic bookings a month

+210%

150 → 465 bookings a month

Listing pages indexed (of 1,200)

60 → 1,180

Listing URLs across four languages

Soft 404s

1,140 → 0

Delisted units now return 410

Non-brand keywords in the top 10

4 → 380

Brand queries held at #1 throughout

Modelled scenario for a fictional business — not a client result. Rankings and traffic are never guaranteed; see our Terms.

Organic bookings a month, brand and non-brandNon-brand searchesBrand searches
SSR live0150300450M1M2M3M4M5M6M7M8M9M10465150

Index coverage by language, of 300 listing URLs each

  • Indexed
  • Crawled, not indexed
  • Soft 404

English

Month 134
Month 10299

Arabic

Month 19
Month 10293

Russian

Month 111
Month 10296

German

Month 16
Month 10292

Core Web Vitals and JavaScript, field data (p75, mobile)

LCP5.8 s → 1.6 s
INP420 ms → 160 ms
JavaScript per listing2,300 KB → 380 KB

The vertical rule marks the “good” threshold (for JavaScript, our 500 KB budget).

Queries that found the apartments

QueryLanguageMonth 1Month 10
dubai marina holiday apartmentEnglish—#4
dubai marina apartment sea viewEnglish—#7
jbr beach apartment rentalEnglish—#9
شقق فندقية دبي ماريناArabic—#3
شقة للإيجار اليومي دبي ماريناArabic—#6
апартаменты дубай марина посуточноRussian—#5
квартира в дубае у моряRussian—#11
ferienwohnung dubai marinaGerman—#4
wahat keysEnglish#1#1

— : outside the top 100. Non-brand keywords in the top 10: 4 → 380.

Let the air in

Running a JavaScript-heavy site that Google doesn't seem to see? Start with the Signal plan, or tell us what your pages look like with JavaScript switched off.

← All case studies