Your homepage can look finished while its shared link tells a different story. Before a reader opens your site, they may see only a title, a short description and an image. That small card deserves its own check.
For this Storefront Experiments check, we inspected our public homepage on 30 October 2026. The useful finding was straightforward: the page already declared a branded sharing image, and the referenced image could be retrieved. No new homepage correction was justified by that check.
Here is how to inspect your own link, record what you can verify and decide whether a change is actually needed.
What our homepage declared
We fetched the public homepage at storefrontexperiments.com and inspected its social metadata: the fields a page supplies to services that build link previews. The page returned HTTP 200. Its declared sharing image also returned HTTP 200 as a JPEG.
| Field | Observed value |
|---|---|
| Open Graph title | Storefront Experiments |
| Open Graph description | Building an online store with AI — real experiments, results and lessons. |
| Declared image dimensions | 1200 × 630 pixels |
| Image alternative text | A description of the brand, storefront and experiment notebook |
| X card declaration | summary_large_image |
The downloaded JPEG was also 1200 × 630 pixels. The title and description identify the project. An individual article needs its own specific promise rather than simply repeating the site name.

1. Start with the exact URL you will share
Copy the public address from the published page. A homepage, an article and a private preview are different destinations. Record the URL, check date and intended sharing service so that you can repeat the same test later.
Open the address without relying on an editor session. Can a visitor reach the intended content? A valid image does not help if the link itself leads to an unexpected destination.
2. Read the card as a new visitor
Check the title, description and image together. Can someone identify the topic and understand why they would open it? Compare the preview with the actual page: a useful title should lead to the content it promises.
For this article, “Your Website Looks Good. Check Its Link Preview.” names the task. The cover says “Check before you share.” They support the same idea without promising traffic or sales results.
On WordPress.com, start with the post’s featured-image setting. The official featured-image guide explains where these images can appear, including social previews. Check the resulting page rather than assuming the selected image is what every service displays.
3. Check the image at a small size
Look at the image as a small card, not only as a full-size design. Is the main subject recognisable? Does the headline remain readable? Is essential information close to an edge that a different crop could remove?
Our homepage declares a 1200 × 630 image. That is a record of this page’s setting, not a guarantee of the same crop everywhere. Inspect the destination service you intend to use. Keep image alternative text descriptive; it should explain the image rather than repeat a list of search keywords.
4. Separate the page check from the platform check
A metadata check establishes what the page declares. Retrieving the image establishes that our request could access the asset. Neither proves how Facebook, LinkedIn, X or a messaging app rendered the link at that moment.
For the next step, paste the exact public URL into the intended service’s composer and inspect the preview before posting. If that service provides a preview debugger or refresh control, use its current instructions. Record the visible result and the time.
If the card differs from the page’s current declarations, investigate the fetched URL, image access and the service’s stored preview. Avoid repeatedly changing a correct image simply because one display has not yet caught up.
5. Change one thing, then repeat the same check
If the image is wrong, replace that image and inspect the same URL again. If the headline is vague, improve it and repeat the title check. Keep the original observation so that a later comparison has a clear starting point.
For this homepage inspection, our record is: branded image declared; referenced asset accessible; no new edit made. Platform rendering and click performance remain untested. A check that finds no new problem is still a useful result.
Use AI to review the evidence you provide
Review this page title, description, sharing image and preview screenshot: [paste inputs]. Check whether they communicate the same topic and match the destination page. Identify unreadable text, ambiguous wording and possible cropping risks. Separate visible observations from assumptions. Suggest one change only if the evidence supports it. Do not claim you tested links, platform rendering or click performance unless test results are supplied.
Before sharing your next article, keep one short record: exact URL, declared fields, accessible image, actual platform preview and any correction. Mark an unfinished step as untested.
For the wider customer journey, read Draft, Connect, Check. For the writing behind a page, see what to check in AI product descriptions.
Get the next practical experiment
Subscribe for new Storefront Experiments posts and unlock the free Ecommerce Prompt Playbook: 12 prompts for store-building decisions, copy and checks.
Check scope: public homepage HTML and retrieval of its referenced Open Graph image, 30 October 2026. No social-platform preview screenshot, before-and-after correction or click-rate experiment was completed. The cover is an AI-generated editorial illustration, not test evidence.

Leave a Reply