A data analyst runs a hypothesis test to determine whether a new checkout page increases conversion rates. The test uses a very small sample and fails to reject the null hypothesis, concluding there is no effect. Later, a much larger follow-up study with the same design clearly shows the new page does increase conversions. What type of error most likely occurred in the original analysis?