Buyers of news data may be rewarding the wrong providers. According to Opoint, the headline figures used to compare news APIs, such as total sources and article volumes, can be pushed upwards by sloppy crawling as easily as by genuine reach.
A provider’s source list is only a snapshot of its configuration. It can include sites that stopped publishing years ago, single publishers split into dozens of subdomains, feeds counted separately from their parent sites, and outlets that produce nothing worth indexing. Opoint notes that this is not always dishonest, and is often just poor housekeeping. For the buyer, the result is the same.
The core problem is that the easiest way to improve traditional benchmarks is also the wrong way. A crawler that collects duplicate URLs, holds on to dead sources and indexes menus and sidebars as article text will perform well on most conventional measures. It produces more sources, more articles and more search matches, while delivering worse output.
This creates a trap. When a provider fixes its duplication problem and stops treating the print, mobile and AMP versions of one story as three items, its article count falls. On paper, coverage looks weaker. In practice, the corpus becomes cleaner, deduplication becomes cheaper and searches return fewer false positives.
Opoint identifies seven common sources of inflation. These include one site counted as many, dormant “zombie” sources, duplicate URLs, archive content misdated as new, non-editorial pages indexed as news, auto-generated content farms and page furniture parsed as body text.
As an alternative, the company proposes diversity metrics, which ask how much content is genuinely different rather than how much exists. Within-domain diversity measures distinct domain and headline combinations against total articles. Headline diversity compares distinct headlines with total output. Language and country diversity counts what actually appears in the feed, rather than what sits in a configuration file.
Opoint’s own October 2026 figures show 20.4m articles in a week, with 12.2m distinct headlines, giving headline diversity of 59.6%. Within-domain diversity stood at 97.1%. The company stresses that headline diversity naturally falls over longer windows as stories are syndicated, and that providers stripping wire copy will score higher without crawling any better.
Opoint also argues for live output reporting over inventory claims. Its September 2026 data shows around 97,000 sources across 120 languages and 193 countries delivering content daily, rising to 145,000 sources weekly. Sources only count if they published at least one article in the period.
For buyers, Opoint suggests three questions: is it alive, is it different, and is it really there? Providers with clean extraction and deduplication should find them easy to answer.
Copyright © 2026 RegTech Analyst
Copyright © 2026 RegTech Analyst
