The industry already knows there’s a problem with IP-to-postal address accuracy. And now that we’ve got the data to show just how big a problem it is, and where the causes lie, we can now do something about it.
Our recent report sought to answer two questions: How accurate are IPs linked to postal addresses, and how accurate are IPs linked to email addresses? Here’s what we found:
- There are 3x too many IPs per postal address. In the data sets we reviewed, there were far more IPs tied to a single postal address than there should be. That means a lot of extra, unnecessary, and inaccurate links.
- Providers only agree about 6% of the time. These vendors barely align with each other when matching IP to household. So depending on who you buy from, you could be getting a totally different picture.
- IPV6 is drastically underrepresented. 54% of traffic to Google’s servers comes over IPV6. But only 15% of the data sets available from providers were IPV6. That’s a huge gap, and it tells you the market isn’t keeping up.
- Timestamp definitions vary by provider. Timestamp quality and confidence levels vary based on use case. So we need more transparency—and more control—to tune data to the outcome we care about.
What Do We Do With This?
Obviously we can’t just stop using IP as an identifier. But we as an industry can (and should) take steps to resolve these issues. They include:
- Ask your data partners what their approach to IP is and ensure you’re aligned.
- Push for segmentation by organizing data into tranches of high to low confidence tiers.
- Use Truthset ratings to sort your IP data by accuracy and quality to know what’s safe for targeting vs. measurement.
… And maybe most importantly, start talking about this openly across the industry so we can develop a more robust solution in collaboration with each other.
Ultimately, a study is just a list of numbers. What we’re really trying to solve for is getting value back to the brands. And value starts with accurate data.