A sale comes in during a campaign night. The province field says "Istanbul", the district field is empty, and the address line has everything: "Barbaros Mah. Çırağan Cad. No:12 D:4 Beşiktaş".
To your eye there's no problem, the district is obvious. But the system printing the label reads province and district from separate fields, and it sees the district field empty. It doesn't read the address line the way you do.
The annoying thing about address errors is that the cost multiplies depending on where the error is caught. At checkout, it costs nothing. At order import, a few seconds. At label creation, a minute. After the parcel has left, a redelivery, a support exchange and maybe a return.
This guide covers the four types of address error, which ones software can catch, which ones the customer has to be asked about, and where Shipink fits into the flow. We also write down its limits.
We covered the cost of a failed delivery after a bad address and the countermeasures in the failed deliveries guide, and which label fields trigger surcharges in the shipping label guide. Here we focus only on the part of the address before the label.
Address Errors Aren't One Error: They're Four
| Type | Example | Can software catch it? |
|---|---|---|
| Province/district spelling and format | "Istanbul", "SISLI", "Yuksekova" | Yes, automatically |
| Province/district buried in free text | District field empty, district in the address line | Largely, with AI |
| Street, building or apartment missing or wrong | "Atatürk Cad." with no number | No. Flagged, or you ask the customer |
| Phone empty or wrong | No recipient number | No. Flagged if empty, not verified |
The first two are mechanical errors: the right answer is known and a system can find it. In the last two, only the recipient knows the right answer. The distinction matters, because "AI corrects addresses" sounds like it solves all four, and it doesn't.
The Cost: Numbers From Elsewhere, Same Logic
We couldn't find a published, independent address-error rate for Turkey. The two sources we have are both from abroad and both are the seller's own data:
- Shippo says, based on its own shipment data, that about 2.1% of parcels have an address-related issue, that unvalidated addresses are 92% more likely to, and that the average address correction surcharge is $4.64 per package.
- ShippyPro, citing SmartRoutes' 2025 last-mile research, reports that address errors account for 45% of first-attempt delivery failures.
Don't read these as a Turkish rate. Correction fees also vary in Turkey by carrier and contract; check whether your own agreement has an address-correction or redelivery line. But the logic of the calculation is the same everywhere: the number of shipments with a problem address, times what one problem shipment costs you. You can find the first number from your own orders.
How Tools Abroad Do It
Sendcloud's address validation docs describe two layers: a check against the carrier's formatting rules, and a content validation that checks whether the address actually exists. The second corrects recognizable typos, for example "Amstredan" to "Amsterdam". When an address is valid but can't be matched with confidence, the docs recommend reviewing it manually before shipping.
The lesson to take: automatic correction should act silently where it's sure and call you where it isn't. A validator that guesses hides the error instead of catching it.
There's one difference. These tools are built around full-address matching. In Turkey, most order forms consist of a province, a district and a single free-text address line. The problem often starts not at the street number but with the province and district not landing in the right fields. So a validator's first job should be to resolve province and district.
The Flow in Shipink: Three Steps Before the Label
1. Match spelling differences as the order arrives. The province and district coming from the sales channel are matched against Shipink's location database. "Istanbul" and "İstanbul", "Yuksekova" and "Yüksekova" go to the same place. Close spellings resolve to the right name.
2. Extract province and district buried in free text. If a field is empty or doesn't match, AI reads the address line and extracts province and district (on Pro and Max plans). The corrected order shows the note "Address was corrected by AI", and the customer's original value opens from the marker next to the field. In our own testing, this model reaches 99.81% accuracy in extracting province and district. That figure covers province and district only, not street or door number.
3. Put what can't be resolved in front of you. If no step can match it with confidence, the order is flagged as "missing address". In the orders list you select them and complete country, province and district on one screen. The shipment creation screen also shows the province/district issue on each row and lets you fix it there. An address you correct by hand is not overwritten by later updates from the sales channel.
The result: the order shows as "missing" in the list, so you know before the label prints instead of hearing about it from the carrier.
What It Doesn't Do
- It doesn't verify that the street, building and apartment number exist. It doesn't check whether "Atatürk Cad. No:12" has a number 12.
- It doesn't verify the phone. It flags an empty one, but can't know the number works. It matters more on cash-on-delivery orders; we covered that in the cash on delivery guide.
- It doesn't automatically message the customer to "confirm your address". If you need to ask the recipient, you do it.
- It doesn't change your checkout. Making province and district a dropdown and phone mandatory is a setting in the platform you use. It's the cheapest countermeasure, and it's still yours.
So Shipink solves the mechanical part of the address, not the whole: province, district and spelling. The street and door side is still between you and the recipient.
Why Now?
11.11 is two weeks away, Efsane Cuma (Black Friday) a month. In campaign week, reading the address queue by hand is a small load per order and a large one in total. In the peak season plan we recommended not changing infrastructure after the thirty-day mark. Address checking isn't in that category: adding a check that fails safe is not changing your carrier or your label flow.
As volume grows, "glancing at it" stops working. At 10 orders a day, reading addresses by eye is reasonable. At 150, nobody has time to read every address, and errors pile up in the ones that go unread.
An address error in a campaign comes back as a return or complaint at year end. A parcel stuck in redelivery in November lands in December, when both customer patience and your support capacity are at their lowest.
A Test You Can Run This Week
- Look at the failed or returned shipments from the last 30 days. How many were address-related? If you have no reason field, create one.
- Check your checkout form. Are province and district dropdowns or free text? Is phone required?
- Count the orders in your list with a missing address. If it isn't zero, you have a job you fix by hand every week.
- Write down your own address error rate. Divide the number of shipments with an address problem by total shipments. Look at the same number before every campaign from now on.
Let the Error Show Before the Label
An address error is one that gets more expensive after it enters the carrier's system. It can be caught at every point from checkout to label; what matters is that when it isn't caught, a flag calls you.
If you want province and district matched automatically and missing ones collected on one screen, you can try Shipink for free, look at the automation features, or write to us to talk through your setup.
