CI-201a · Module 1

Source Reliability Scoring

3 min read

Not all intelligence is created equal, and treating every data point with the same weight is how bad decisions get made. A patent filing is a legal document with enforceable consequences for inaccuracy. A LinkedIn post is someone's opinion. A press release is marketing with a byline. A Glassdoor review is an anonymous grievance. Each of these sources has a different reliability profile, and your analysis must account for that difference. Source reliability scoring is the framework that prevents you from building strategy on gossip.

The scoring framework has three dimensions: accuracy, recency, and bias. Accuracy measures how likely the source is to contain factual, verifiable information. Legal filings score high. Anonymous forums score low. Recency measures how current the information is — a job posting from yesterday is more valuable than a news article from six months ago. Bias measures the source's inherent motivation to present information in a particular way. A company's own press release has high bias toward positive framing. An SEC filing has legally enforced neutrality. Score each dimension on a three-point scale: high, medium, low.

Do This

  • Score every source on accuracy, recency, and bias before incorporating it into analysis
  • Weight high-reliability sources more heavily when sources conflict
  • Note your confidence level when presenting conclusions — "high confidence based on SEC filing" vs. "low confidence based on social media rumor"

Avoid This

  • Treat all sources equally regardless of reliability
  • Build strategic conclusions on a single unverified source
  • Ignore bias — every source has a motivation, and understanding it is part of the analysis

Corroboration is what turns a data point into intelligence. A single job posting is interesting. That same job posting combined with a patent filing in the same technology area and a pricing page change is a pattern. The reliability scoring framework helps you identify which sources need corroboration (most of them) and which can stand on their own (almost none). The rule of thumb: no strategic conclusion should rest on fewer than two independent sources. If you cannot corroborate it, flag it as unconfirmed and keep collecting.