Back to blog
How-To GuideLast-Mile Delivery

How to reduce failed deliveries in Saudi Arabia: 4 steps

A share of KSA deliveries fail before the driver is ever assigned, because standard geocoders can't place the address at all. The fix is a data-layer change, with no checkout flow to rebuild.

July 16, 20267 min read
The delivery journey · where it's actually lost

The failure is logged at the doorstep. It happened before the van moved.

01CHECKOUT
Customer writes their address
"Near the blue water tower, left after the petrol station, white building."
how people describe home
02GEOCODEFAILURE POINT
Standard geocoder returns nothing
Not in the National Address system, not parseable. Just a blank, not an error.
still recoverable here
03DISPATCH
Routed to an approximation
The job is assigned anyway, to a district centroid that may be blocks away.
looks routable in the log
04DOORSTEP
Delivery fails
Logged as "address not found" and attributed to the field, not the data.
status: reattempt
The address exists and the customer described it clearly. What failed is the geocoding layer, and that's recoverable before a driver is assigned.
The failure surfaces at step 04 but occurs at step 02, where it is still recoverable.

Failed deliveries in Saudi Arabia frequently trace back to a structural problem in how addresses are recorded. In 2022, Jeddah Municipality classified 64 of the city's 210 districts as informal settlements. An Amnesty International review of a Jeddah Municipality document from the same year identified more than 558,000 residents across 60 of those neighborhoods, and that figure covers only the districts targeted for demolition, not the full informal footprint of the city. Residents of these areas navigate by landmarks, mosque names, and alley descriptions rather than street numbers, and any geocoding tool built on Saudi Arabia's National Address system (Wasel) returns nothing for those addresses.

The fix is to pass checkout addresses through a geocoding API built for KSA before route assignment, with no change to your checkout flow required. A MEA-specific database covers informal settlements and accepts free-form text, converting landmark-based instructions into delivery coordinates before a driver is dispatched.

Key takeaways
In 2022, Jeddah Municipality classified 64 of its 210 districts, nearly a third of them, as informal settlements sitting outside the National Address system (Jeddah Municipality).
Standard geocoders that rely on National Address records return no usable coordinate for addresses in informal and newly developed areas. Not an error, just a blank.
No checkout flow change is required: Blassa converts free-form address text, including landmark-based descriptions, directly to coordinates.
Running addresses through Blassa's geocoding API before dispatch catches failures before a driver is assigned.
Blassa's MEA-specific database covers informal settlements, peri-urban expansions, and newer districts global platforms have not indexed.

Jeddah classified 64 of its 210 districts as informal settlements. Standard geocoders have no data for any of them.

The National Address system assigns formal street names, building numbers, districts, and 4-digit zip codes. In established urban areas such as central Riyadh, commercial Jeddah, and downtown Dammam, coverage is high and standard geocoders perform reliably. The gaps are structural, not accidental:

Jeddah's informal settlements · Jeddah Municipality & Amnesty International, 2022
64/210
of Jeddah's districts classified as informal settlements in 2022
~⅓
of the city's districts, nearly one in three
558K+
residents across 60 of those neighborhoods, counting only demolition-targeted districts
These residents give directions by landmark, mosque name, and alley, not building number. For a geocoder built on National Address records, those directions are unresolvable.
Figures: Jeddah Municipality (2022) and Amnesty International's review of a Jeddah Municipality document.
KSA address coverage by area type

Where National Address coverage ends, geocoding reliability falls with it.

Area type
National Address coverage
Expected geocoding reliability
Established urban
Riyadh centre, Jeddah downtown, Dammam
High
High
Peri-urban & new developments
Residential districts built ahead of surveying
Partial
Moderate
Informal settlements
Ashwaiyyat, built outside formal planning
Low or absent
Low
Agricultural & rural areas
Farms, villages, remote districts
Low
Low
Formal coverage and geocoding reliability decline together across area types.

These residents give delivery drivers directions the way anyone does locally: landmarks, mosque names, alley descriptions. For a geocoder built on National Address records, those directions are unresolvable. In Jeddah, that gap covers nearly a third of the city's districts. The result is a failed delivery, not because the address doesn't exist, but because the geocoding layer can't place it.

Why free-form address text breaks standard geocoding tools

A standard geocoder such as Google Maps, HERE, or an OpenStreetMap-based tool converts address text to coordinates by matching against a structured database. That works when addresses are formatted consistently and the database is complete. In KSA, two things break that assumption.

The database has gaps

Informal settlements aren't in the National Address system, new developments take months to be indexed, and rural areas are sparsely covered. A geocoder can only match what's in its database.

no formal recordunindexed developmentssparse rural
The input doesn't match the format

"Near the blue water tower, left after the petrol station, ring the intercom" is a real instruction. A geocoder built for "14 King Fahd Road, Al Olaya" can't parse it, and returns nothing.

landmark-baseddirectional cuesmixed AR–EN
Customers write exactly what their neighbors and local drivers understand. The geocoding layer is the part that isn't keeping up, not the customer.
Coverage gaps and unstructured input: two independent failure causes that both hit KSA addresses at once.
Want to see what percentage of your current address database falls into this gap? Blassa can run that comparison against your shipment records.

How Blassa geocodes what standard tools miss

Blassa's geocoding API is built for the two specific problems above, coverage gaps and free-form text, and resolves them in a single call before any driver is assigned.

Same free-form address · two geocoders
INPUT"Jeddah ashwaiyya — near the blue water tower, white building, ring the intercom"
STANDARD GEOCODER
No result

Nothing in the National Address–based database matches. The dispatcher gets a blank and falls back to a district centroid.

coordinate: null · confidence: —
BLASSA API
Usable coordinate

A MEA-specific database covers the settlement; free-form parsing reads the landmark cues and returns a structured point with a confidence score.

coordinate: 21.54, 39.19 · confidence: high
Coverage
Informal settlements, peri-urban areas, and developments global platforms haven't indexed.
Free-form text
Mixed Arabic–English, landmarks, directional cues, partial addresses, with no reformatting on your end.
Coordinates are illustrative. Whatever the customer types at checkout goes into the API as-is.

4 steps to fix the problem without changing your checkout flow

Each step slots into an existing dispatch workflow. Together, they move address-quality problems from the field to the data layer, before a driver is assigned.

1
Keep your existing checkout flow

Customers write addresses however they do today. Blassa handles the conversion, and informal-settlement addresses have no National Address code to enter anyway.

no UX change · no friction
2
Pass addresses through Blassa's API before route assignment

One geocoding call between checkout and dispatch returns a coordinate and confidence score. High-confidence results route directly.

checkout → API → dispatch
3
Route low-confidence results to a dispatcher review queue

A low score triggers a short human check before dispatch, replacing the silent fallback to an approximate centroid.

low confidence → review
4
Track geocoding hit rate by district, monthly

Measure the share returning a usable coordinate per area. It shows where coverage was weak and confirms improvement over time.

hit rate · by district · monthly
The integration is a single API call inserted into an existing dispatch workflow, typically a day's work, not a platform migration.
Four steps, none of which touch the customer-facing checkout.

Standard geocoding tools weren't built for KSA's address conventions. The markets they were optimized for don't have nearly a third of a major city's districts classified as informal settlements, with residents navigating by mosque names and alley descriptions instead of building numbers. Blassa's MEA-specific database and free-form text geocoding were built for that context specifically. For more on the structural reasons standard tools miss MENA addresses, see Why informal addressing systems in MENA break standard geocoders.

Curious how your current tool's match rate compares to Blassa's in KSA or Egypt? Let's show you the data.
Frequently asked
Do customers need to enter addresses differently at checkout?

No. Blassa's API accepts free-form text input, including landmark-based and informal descriptions. The geocoding layer handles the conversion, so customers write exactly as they do today.

What is the scale of informal settlements in KSA?

Jeddah is the best-documented case. In 2022, the city classified 64 of its 210 districts as informal settlements. Amnesty International's review of a Jeddah Municipality document from the same year identified more than 558,000 residents across 60 of those neighborhoods, though that figure covers only districts targeted for demolition under the Jeddah Central Project, not the full informal footprint. Riyadh has documented informal settlements as well; a 2023 study in MDPI Remote Sensing mapped them across multiple districts of the capital.

Why does this matter more in KSA than in other markets?

Most global geocoders are trained on markets where street numbering is complete and consistently formatted. In KSA, informal areas are physically large, densely populated, and give addresses by landmark rather than street number. Global tools weren't built for this convention; Blassa's MEA-specific database and free-form text processing are.

How long does the Blassa API integration take?

The integration is a single API call inserted into an existing dispatch workflow. For teams with a developer on hand, this is typically a day's work, not a platform migration.