importyeti light logo
Menu

How We Clean Company Data

At ImportYeti, we process millions of Bills of Lading (BOLs). Company names often appear in multiple variations, with typos, codes, or formatting differences. We use machine learning to unify them into a single clean identity.

Overview

From Dirty Names to One Clean Company

Raw shipping records contain messy, inconsistent company names — we treat them as a black box of variations. Our machine learning models learn patterns, group related names, and output one standardized company page.

Dirty names (input)
017 Dt Ikea Supply
100 Com 10 Ikea Supply Ag
460 Ikea Wholesale
5100 Com 10 Ikea 5100 Com 15 Ikeasupply Ag
5100 Com 18 Ikea Supply
Ikea Suppl
Ikea Suppl Ag
Ikea Supplu
+ dozens more variations…
Machine learning

Name joining, standardization, noise removal, and deduplication — trained on millions of BOL records.

Clean name (output)
IKEA Supply
importyeti.com/company/ikea-supply
Machine Learning

What Our Models Do

Name joining — A single company like IKEA can appear under dozens of different names. We group all variations into one unified company page instead of fragmented data.

Standardization — We remove noise such as internal codes, suffixes (Inc, LLC, Ltd), and formatting inconsistencies. Example: IKEA Supply AG, IKEA Supply Inc., and Ikea Supply all become IKEA Supply.

Personal data removal — Models also detect and remove personal information that may appear in shipping records (names or sensitive identifiers that should not be publicly displayed), keeping the dataset compliant and clean.

Why It Matters

Easy to Visualize — Dirty vs. Clean

Without cleaning, the same company is scattered across many pages with different names — hard to grasp at a glance. After machine learning, you get one visual story: one page, one name, one view of global shipping activity.

It is much easier to visually show a single IKEA page than to explain 100 separate pages for IKEA with slightly different names. That clarity makes search, analysis, and lead generation significantly more powerful.

Without cleaning
~100+ pages

Same company, many names — fragmented, confusing, impossible to scan in one view.

With machine learning
1 page

IKEA Supply — one clean name, one accurate view of all related shipments.