Five consecutive weekly reviews had built on a number the measurement window produced. Checking it against the source data changed what the strategy should do next.

• A weekly review had reported site impressions rising 202% over five weeks while clicks stayed flat, and a new strategic rule was about to be adopted on the strength of it. The monitor pulls a 90-day rolling window, so two readings five weeks apart shared 62 of their 90 days.
• Read week by week instead, impressions roughly tripled over six weeks and clicks held at about one per week. The conclusion survived. The figure quoted for it did not, and neither did the claim that it described five weeks.
• Three other claims had ridden along unchecked across the same five reviews. One had been formally disproven and closed in the CRM seven weeks earlier. Another held that the answer-engine instruments could not report a non-zero, when both run weekly and were returning a real zero.
• The homepage description was rewritten and deployed the same day, replacing wording that had been recorded as live since July but never actually deployed.