An element entering the viewport is not the same as a survey impression
An optional question below a useful tool can help distinguish different visitor intents. But its denominator needs care. A page view does not prove that someone reached the question, and an unanswered question does not reveal their preference. I maintain RetroPrompt, where a page now asks whether visitors want a still portrait, a video template, or are simply exploring. This article was prepared…
An extra element appearing on a webpage does not equate to a single survey response. However, placing a question after useful content requires caution. Just because someone scrolls to that question does not mean they chose an answer. For the RetroPrompt demonstration, visitors are asked whether they prefer a vertical image, a video template, or simply exploring.
The article outlines how to measure question visibility, not visitor preferences. Before reporting an answer rate, clarify exactly what exposure means. This question sits after the page's main content and controls, so some users may never see it. For this setup, exposure is defined as at least half of the panel appearing in the viewport.
This is an operational definition, not a measure of attention or understanding. Even with a threshold set to 0.5, IntersectionObserver does not guarantee every entry represents a half-visible impression. Developers must verify both the isIntersecting property and the intersectionRatio. The code snippet shows how to record exposure when the entry meets both criteria.
The panel's visibility depends on its size relative to the viewport. A tall panel may need a different visibility target. The script also distinguishes between unanswered sessions, answered sessions, and first-choice counts. It's crucial to report these metrics separately. An unanswered session is not a vote for the default option.
An "exploring" answer is distinct from no answer at all. Visitors can modify their selection, but the report should reflect your intended measurement. Make this rule clear in your report rather than adjusting it after analysis. Test various scenarios, such as a panel with a narrow view and multiple observer callbacks. Exclude your own testing sessions from the audience report to avoid bias.
Be cautious, as self-selected survey answers do not equate to demand or willingness to pay. Compare these responses to actual useful actions and report the associated uncertainty, especially when dealing with a small sample size. The RetroPrompt OSM guide provides an example interface.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
