The first impression frame

Most effort spent on a shop's own page goes into things most visitors never actually see. This page is about the part they do.

What's actually visible before the decision gets made

In a search answer: the sign, a handful of window photos, video, a rating, small labels, a card about the company, a poster for something happening this week, and a featured extra item.

That's already a substantial amount of information, and a shopper can decide without ever opening your actual page at all. The search answer is the real first impression. Not your own page.

On your own page, before anyone scrolls: the sign above the door, your sign, your pitch line, a rating, and the first window photo or your video's cover image. The leaflet sits below that line, and on one side, only the portion before "read more" shows at all.

Everything else is below the fold: the rest of your window photos, the body of your leaflet, your note about what changed, the full breakdown behind your rating.

What follows from that

The first few window photos carry your whole argument. They show up in the search answer right next to rivals, so they have to work as a standalone unit. Assume nobody scrolls.

Tall versus wide is a real decision. Tall gets you roughly three narrow frames in a search answer; wide gets you roughly one wide one. Three chances at a short message, or one chance at a longer one. Worth testing rather than simply inheriting whatever you started with.

Caption text does the real work. Text and simple overlays explaining a feature are allowed on window photos. A raw, uncaptioned screen communicates almost nothing at search-answer size; a caption plus a cropped, enlarged piece of the interface communicates one thing clearly.

Practical limits on captions: short enough to read at a glance, big enough to survive being stretched down to a smaller device, and in the visitor's own language.

Your pitch line sits above the fold on the other centre. The centre's own advice there is to summarise the shop's biggest benefit and explain its core purpose in one short sentence. It gets read; the full leaflet largely doesn't.

Matching what was actually asked

Everything above treats the search answer as one single frame. It isn't. It's a frame shown against a specific question, and the shopper is scanning it for evidence that your shop does the thing they just asked for.

The constraint that makes this genuinely hard: your default page is the same for every single question anyone could ask. A specially-dressed window and a live test are the only ways to actually show a searcher something different, and a specially-dressed window mostly redirects traffic from words you already hold rather than adding genuinely new ones. So for your default page, you're choosing one visible wording to serve every one of your words at once, and that choice creates winners and losers among them.

What follows for the actual work:

Put the word itself in your visible writing, not only in the hidden pocket. Being filed under a word and looking like the answer to it are two different problems. Someone searching for a specific feature scans the sign, the pitch line and the first caption for that exact word; if it only lives in the hidden pocket, you're filed correctly and still lose the tap.

Decide which words your visible writing is even allowed to serve. Take the handful of words that actually carry your real search visitors, and for each one, look at the live result on a real device and ask honestly whether your entry reads as the answer to that question, sitting next to rivals. Most shops discover their visible writing only really serves their own name and one generic word, and nothing else.

Weigh any visible-writing change against every word it touches, not just the one that prompted it. Rewriting your pitch line to capture a rising word costs you relevance on whatever it displaced. This is exactly why a pitch-line change is a persuasion decision as much as a filing decision, and belongs in the same review as your hidden pocket.

Where a word genuinely can't fit, answer the intent instead. A caption doesn't have to contain the exact word to look like the answer to it. This part is genuinely creative work, not filing work.

The strongest version of this whole argument. That matching a word visibly lifts conversion for that specific word, and that the lift then feeds your position for it, is a working belief among experienced shopkeepers, backed by individual stories rather than a controlled test. The underlying mechanism is plausible on both centres' own account, since conversion genuinely feeds the desk on both sides. But the actual size of the effect isn't proven. Use the idea to decide what to test, not to predict what a test will show you.

Ordering the argument

A reasonable starting point, to be tested rather than simply trusted:

What it is and why it's better (the single claim that actually sets you apart. Then the main thing most visitors actually came for. Then proof, or breadth) social proof, scale, or a genuine second use case.

Then, below the fold, the rest of your set for people who are already interested.

Don't lead with a feature nobody actually asked for. The most common first-photo mistake is opening with something novel instead of something wanted.

The rabbit hole

Window-photo design (colour choices, device frames versus edge-to-edge, where a caption sits, the "hero" pattern) is a huge body of practitioner craft with genuine skill in it and very little evidence that actually transfers between shops. Conversion effects are specific to the individual shop and its category, which is exactly why claimed improvements rarely repeat elsewhere.

The practical layer: decide what your first few frames absolutely must say, make them legible at the size they're actually seen, and test them. The rabbit hole exists; treat it as inspiration, not evidence.

What tends to go wrong

Designing a long, ten-photo story. Only a handful are actually seen, in a list, right next to rivals.

Captions sized for a design tool. Reviewed huge, seen tiny.

Raw, uncaptioned screens. Unreadable at search-answer size.

Leading with the newest feature. Interesting internally, irrelevant to someone actually deciding.

Polishing the leaflet while the first photo stays untouched. The leaflet sits below the fold, and on one side of the centre, it isn't even read by the desk at all.