Reading Instagram UGC performance signals before you feature it on-site

Why raw engagement is a weak signal on its own

A post's like count mostly measures how scroll-stopping the image was in a feed, which correlates loosely, if at all, with whether the viewer was considering a purchase. A striking photo can rack up likes from people who have no intention of buying anything; a plainer photo with a genuinely curious comment thread underneath can indicate real purchase consideration despite modest reach.

The signals worth reading

Comment content over comment count. Read the actual comments, not just the number. Questions like "where's this from" or "does this run true to size" signal a viewer moving toward a purchase decision. Generic compliments ("gorgeous!") do not.

Saves over likes. A save is a private action a user takes when they intend to return to the content, often while comparison shopping or waiting for payday. It's a closer proxy for purchase intent than a public like, and worth weighting more heavily when a UGC platform's analytics expose it.

Product visibility, not just aesthetic quality. A beautifully shot photo where the product is barely identifiable underperforms a plainer photo where the product is the clear subject. This matters more once the content moves to a PDP, where the visitor already has purchase intent and wants to see the actual product, not an art-directed scene.

Settling it with a real test

Engagement signals are a filter, not a verdict. The only way to actually know whether a candidate piece of UGC will lift conversion is to run it, ideally as a holdout test against the current PDP layout, and measure revenue per visitor directly. A/B testing UGC placement covers the framework for that test once you've picked candidates worth spending rights-request effort on.