78% → 84% first-attempt. No room full of callers

What everyone assumes

In most of the GCC, an address is a sentence, not a code. That much everyone agrees on.

What everyone assumes next is the expensive part. Incomplete addresses are a last-mile problem, so you fix them with last-mile people: phone the recipient before dispatch, write down a better address, hit 85–88% first-attempt the way competitors already do. It works. It is also a permanent headcount line that scales with volume, sold as a premium service at a premium price.

Hold volume constant. Who gets more first attempts? The desk that calls harder. Equal call effort: the team with better scripts. Equal scripts: the one that somehow gets cleaner addresses from merchants. Keep changing the variable and the market's answer stays the same — buy a room full of callers.

This aggregator was landing 78% of parcels on the first attempt. The missing fifth were not lost. They were re-attempted, held, called about, re-routed, and eventually delivered — each step costing a van slot, an agent's time, and a customer's patience. Support fielded the calls. Drivers made judgement calls at the kerb. Neither was positioned to fix the record at source.

So the question was not whether address quality could be fixed. It was whether it could be fixed without buying a call centre.

The variable that actually moved

The variable was not better callers, better merchant portals, or a post-failure review queue. It was when the address gets corrected.

XB AddressIQ sits in front of the aggregator's OMS. Every inbound address is parsed, resolved and corrected before the order record is created — not flagged for someone to review, not queued for a human, not caught after a failed attempt.

That placement is the whole design. An address corrected after dispatch has already cost you the attempt. An address corrected inside a review queue has already cost you a person. Correcting it upstream of the OMS means every system that reads the order downstream — routing, driver app, tracking, support — reads the good version, and nobody had to be told.

Same shippers. Same order flow. No new format, no portal for merchants to learn, no integration work on their side. Change only where correction happens, and the call centre stops being the required answer.

In markets without a dependable postcode, many "bad addresses" are not wrong — they are uncheckable against the tools built for ZIP and postcode worlds. Treating them as data-entry failures sends you to phones. Treating them as a resolution problem before the OMS exists sends you to a different cost curve.

What they can sell now

First-attempt delivery moved from 78% to 84% — six points, on the aggregator's own numbers, across its GCC lanes.

That lands a point under the 85–88% competitors reach with people on phones — near enough to compete on service, reached a completely different way. Their band is a team that grows as volume grows, and a premium charged to cover it. This aggregator's 84% is a fixed cost with no one added to the payroll: the same class of result as a property of the pipeline, not as a service someone performs.

Six points sounds modest until you count what a first attempt is worth. Every avoided re-attempt is a van slot returned to the schedule, a support call that never happens, and a recipient who does not have to be home twice. Those savings compound across every lane, every day, without anyone doing anything.

What they can sell now is competitive rather than operational: quote against the premium-service operators without carrying their cost base — call-centre delivery rates without the call centre.