SEO case study · Inter-island ferry operator · Cebu
Tamarind Ferries
How a Cebu ferry line made its timetable load on a 3G signal
Passengers checked schedules on patchy mobile data, and a heavy timetable page timed out before it loaded.
TF-001 · ADMIT ONE
Mobile LCP
1.6s
from 7.1s
Headline: Mobile LCP 7.1s → 1.6s · Asia-Pacific
Each ferry takes as long to cross as the timetable page took to paint on a 3G phone.
The challenge
The pier has one bar of signal.
Passengers checked schedules on patchy mobile data, and a heavy timetable page timed out before it loaded.
Most of Tamarind Ferries' passengers check departures from a jeepney, a pier or a guesthouse on another island, on prepaid mobile data that drops to 3G the moment they leave the city. The timetable was the most visited page on the site and the slowest: a single page holding every route's schedule for a full year, behind a hero photograph and a heavy booking script.
When it timed out, passengers did what anyone would: went back to search and booked through an aggregator, or turned up at the terminal and queued. Search engines saw the same slow, half-empty mobile page and ranked it accordingly.
- timetable page weight
- 4.8 MB
- timetable page weight
- rows in one schedule table
- 11,000
- rows in one schedule table
- JavaScript before any time showed
- 640 KB
- JavaScript before any time showed
- routes with structured data
- 0
- routes with structured data
Diagnosis
The inspection manifest.
Field data from real phones in the Visayas, a crawl of 1,140 URLs and lab runs on a throttled 3G profile produced six findings, ranked by how many passengers they cost.
- No. 01CriticalPerformance
The timetable page shipped 4.8 MB: a 2.1 MB hero photo of a ship at sunset, plus every route's full-year schedule rendered as one 11,000-row table.
Fix: Hero re-encoded to 38 KB AVIF with a WebP fallback; the table split into one lightweight page per route and season.
- No. 02CriticalPerformance
Schedules were fetched client-side after a 640 KB JavaScript bundle loaded, so on a 3G signal the departure times never painted before the browser gave up.
Fix: Departures server-rendered into the HTML; the booking widget loads only on tap.
- No. 03HighMobile-first
The mobile template hid fares and vessel names behind tabs that Googlebot's smartphone crawler could not expand, so the indexed page had less content than desktop.
Fix: Parity restored: every fare, vessel and port note present in the mobile HTML.
- No. 04HighStructured data
No schema at all. Departures, ports and fares existed only as table cells, invisible to rich results.
Fix: Trip and BoatTrip markup with departure and arrival ports, times, offers and the operating vessel, per route.
- No. 05MediumImages
214 uncompressed port and deck photos, none with width or height set, causing a layout shift score of 0.31.
Fix: Responsive sizes, explicit dimensions, lazy loading below the fold. CLS fell to 0.02.
- No. 06MediumLocal
Six port terminal offices had no individual pages, so 'ferry terminal near me' searches went to aggregators.
Fix: One page per terminal with hours, map directions written in text and a link to that port's departures.
The strategy
Four tickets, one destination: a fast timetable.
Built on our Signal plan, scoped to the pages passengers actually open on the way to the pier.
- 1
Core Web Vitals
Make the first useful thing paint first
We set a budget of 300 KB per route page and 2.5 s LCP on a throttled 3G profile, then cut to it: critical CSS inlined, fonts swapped for system stacks, the booking script deferred until a passenger taps Book.
- 2
Schedule schema
Turn the timetable into data search engines can read
Each route page now carries Trip and BoatTrip markup generated from the operator's schedule feed, so departure times can surface directly in search, and a nightly job keeps it in step with seasonal changes.
- 3
Mobile-first indexing
Make the phone version the complete version
The smartphone crawler sees exactly what a passenger on the pier sees: fares, vessels, luggage rules and port notes, all in the initial HTML instead of behind tabs and scripts.
- 4
Image compression
Keep the sunset, lose the megabytes
An image pipeline produces AVIF and WebP at five widths with explicit dimensions. Deck photos lazy-load; the hero is a 38 KB file that still looks like the Visayas at golden hour.
6 months, gate by gate
Departures.
- M01A1
Audit and weigh-in
Field data from real Cebu phones, a crawl of 1,140 URLs, and a 3G lab profile. The heaviest page in the fleet: the timetable.
5.4s DELAYED - M02A2
Split the timetable
One page per route and season, server-rendered. Page weight from 4.8 MB to 1.4 MB. First schedule rich results appear.
3.9s DELAYED - M03B1
Image pipeline
214 photos re-encoded with sizes and dimensions. CLS from 0.31 to 0.04. LCP drops under 3 s on 3G.
2.7s DELAYED - M04B2
Mobile parity
Hidden tabs removed; fares and vessel names in the HTML. Terminal pages launch for six ports.
2.1s ON TIME - M05C1
Schema at scale
Trip markup generated from the schedule feed for every route and seasonal variant. 78 routes eligible.
1.8s ON TIME - M06C2
Hold the line
Performance budget enforced in the build. Any page over 300 KB fails the deploy. 86 routes with rich results.
1.6s ON TIME
Results after 6 months
Faster than the ferry it sells tickets for.
Modelled scenario for a fictional business — not a client result. Rankings and traffic are never guaranteed; see our Terms.
TF-R01 · ADMIT ONE
Mobile LCP
1.6s
from 7.1s
VALID6monthsTF-R02 · ADMIT ONE
Organic bookings / month
4,700
from 1,900
VALID6monthsTF-R03 · ADMIT ONE
Routes with timetable rich results
86
from 0
VALID6months
1,900 → 4,700 per month
Crossing replay
The ferry takes exactly as long to cross as the timetable page took to paint on a throttled 3G phone that month.
1.6sMonth 6: mobile LCP, 290 KB page
- Bookings
- 4,700
- Rich results
- 86 routes
- Speed gained
- 100%
tamarindferries.example › timetable
Timetable | Tamarind Ferries
Tamarind Ferries timetable. All routes. Click to view schedule...
Position 14 · page 2
tamarindferries.example › routes › cebu-tagbilaran
Cebu to Tagbilaran Ferry: Today's Departures & Fares
6 sailings daily · 2h crossing · from ₱ 520 · vehicle deck on the 06:00 and 13:30.
Position 2 · rich result
One ticket per route. Tear the stub to see where it ranks now.
- TF-101DEP 06:00
Cebu to Tagbilaran
Crossing 2h 00m
Search: “cebu to bohol ferry schedule”
- TF-214DEP 07:30
Cebu to Ormoc
Crossing 2h 45m
Search: “cebu ormoc ferry time”
- TF-318DEP 08:15
Cebu to Dumaguete
Crossing 4h 30m
Search: “ferry cebu to dumaguete today”
- TF-405DEP 05:30
Danao to Camotes
Crossing 1h 30m
Search: “camotes island ferry schedule”
- TF-522DEP 09:00
Hagnaya to Bantayan
Crossing 1h 10m
Search: “hagnaya to bantayan ferry”
- TF-630DEP 13:00
Cebu to Siquijor
Crossing 5h 00m
Search: “cebu siquijor boat trip”
Keyword positions
| Search term | Monthly volume | Before | After | SERP feature |
|---|---|---|---|---|
| cebu to bohol ferry schedule | 12,100 | 14 | 2 | Rich result |
| cebu ormoc ferry time | 4,400 | 22 | 3 | Rich result |
| ferry cebu to dumaguete today | 3,600 | 31 | 4 | Rich result |
| tamarind ferries timetable | 2,900 | 3 | 1 | Sitelinks |
| hagnaya to bantayan ferry | 2,400 | 18 | 2 | Rich result |
| camotes island ferry schedule | 1,900 | 9 | 1 | Rich result |
| cebu siquijor boat trip | 1,300 | 40 | 6 | Rich result |
| ferry with car cebu bohol | 880 | — | 5 | FAQ |
Modelled scenario for a fictional business — not a client result. Rankings and traffic are never guaranteed; see our Terms.
Get a plan like this
If your customers find you on a weak signal, speed is the ranking factor you can actually control. Start with the Signal plan, or tell us where your pages are slowest.