How do you tell a visitor where to hold their phone?

You cannot know their phone

The antenna position varies by model and there is no way to know what a visitor is carrying. Any instruction naming a part of the phone will be wrong for someone.

They also do not know where their own antenna is. Most people have never needed to.

So an instruction that assumes knowledge on their side fails twice, and the visitor concludes the tag is broken.

Instruct the tag instead

Print a target on the tag: a symbol, a circle, a contactless mark. People aim at a marked spot and wave vaguely at a blank surface.

This is the highest return change available in a public deployment, and it costs nothing at print time.

Make the target the size of the tag rather than a small icon in a corner, since the whole antenna area is the useful region.

And place the tag where a phone naturally arrives: at hand height, on a surface someone can approach squarely, not on an edge or above a counter.

The words that work

Short and about the action. Hold your phone here. Touch and hold for a second.

Mention holding still, because a brush is the commonest failed gesture and one word fixes it.

Say what will happen, since a person who knows a page will open is far more likely to tap the notification an iPhone shows, or the one Android shows for a web link from Android 17.

Avoid the word scan, which suggests a camera and sends people looking for a code. Tap and hold describe the gesture accurately.

Watch real people

Ask someone who has never seen the tag to use it, and say nothing. What they do is the requirement.

Watch where they present the phone, how long they hold it, and whether they wait for anything to happen.

Then watch a second person, because one person's habits are not a sample.

Almost every fix that comes out of this exercise is free: a bigger target, four different words, a tag moved a hand's width. That is why it is worth doing before anything is printed in quantity.