Separate accuracy from consistency
Accuracy asks how close a count is to what actually crossed the boundary. Consistency asks whether another round would use the same rule. You need both, but consistency is the part you can improve most directly in fieldwork.
Draw one observable counting line
Choose a doorway, pavement seam, sign, or other visible line. Count a person only when they cross it. A vague area makes it easier to count someone twice when they pause, turn around, or move near the edge of view.
Keep categories mutually exclusive
If you classify direction, audience fit, or venue behavior, assign each person to one button. Overlapping categories can make the total look precise while quietly duplicating people.
Run a short two-person validation check
For a busy location, have two people independently count the same boundary for five minutes, then compare totals and disagreements. This does not establish a universal accuracy percentage, but it can expose an unclear boundary, confusing categories, or a flow that one observer cannot reliably cover.
Record conditions that increase error
Crowding, groups crossing together, blocked sightlines, multiple entrances, and interruptions all raise the chance of missed taps. Note these conditions and avoid presenting a difficult round as equally reliable as a clear one.
Repeat comparable rounds instead of over-extrapolating
A 15-minute count is a sample of that time and condition, not proof of a full day. Repeat the same method across the hours and days relevant to the decision before estimating a broader pattern.