Four Providence Bakeries, 437 Images, and Not One Price a Computer Can Read. Here Is the 20-Minute Fix.

A bakery is the most honest business on the street.

There is no upsell. There is no consultation. There is a case, and the case is full or it is not, and by two in the afternoon the answer is visible from the sidewalk. The bread was good or it was not. You do not need a review platform to tell you. You need eyes.

That honesty is exactly why a bakery is so badly served by the way most bakery websites are built. The case is the product. So when it comes time to put the shop on the internet, the instinct is to photograph the case. Photograph the loaves. Photograph the cannoli tray on a Saturday when it looks the way it deserves to look. And that instinct is correct about what customers want and completely wrong about what a computer can see.

This morning I fetched twelve pages across four independent bakeries in Providence and East Providence. Every page that looked like it might hold products: the menus, the bread pages, the pastry pages, the specialty cake pages, the catering pages, the daily specials. All twelve returned normally. All twelve are real, working, cared-about websites.

Between them those twelve pages carry 437 images.

And zero prices that a machine can read.

One point of honesty on that first number, because I would want it from anyone quoting a figure at me: 437 is a count of <img> elements, not a count of photographs of bread. Some of those are logos, icons and social badges. When I filtered for what looked like genuine content photography I got 249, but that filter misses images loaded by script, so I do not trust it enough to lead with it. The count I can defend is 437 image elements. The count that matters is the second one, and it is zero either way.

I want to be careful about what that does and does not mean, because the easy version of this article is a sneer and these are not sneerable businesses. These are shops with decades on them, the kind of place people drive across a city for on a Sunday morning. Their websites are not neglected. Three of the four have Google Analytics installed, which means somebody is paying attention. All four are on HTTPS with valid certificates. All four have a working mobile layout. Somebody made deliberate choices here.

The choice they made was to show the food. It is the right instinct. It is just incomplete, and the part that is missing is the part that gets found.

What "on the internet" actually means now#

Here is the distinction that matters, and almost nobody explains it to a shop owner before selling them a website.

Your website being online means a person who already knows your name can find your hours. That is a brochure. It is useful and it is not nothing.

Your products being online means something different: that the individual things you sell exist as separate, named, priced, machine-readable facts. Not a photograph of a price list. Not a PDF. Not a Canva graphic of the menu posted to Instagram. Text, with a number next to it, marked up so that a search engine, a voice assistant, or an AI answering someone's question at 9pm on a Tuesday can tell that "Portuguese sweet bread, $6.50, in stock" is a thing that is true about your shop.

Right now, for these four bakeries, that fact does not exist anywhere on the public internet. Somebody searching "portuguese sweet bread near me providence" is being answered by whoever did write it down. Which is usually a delivery platform taking a cut, or an aggregator that got the price wrong in 2023 and never checked.

While structured data isn't required to appear in Google Search results, it can help Google understand your page better and display it as a rich result.

What I measured this morning, anonymised#

Four bakeries, run through Google's own PageSpeed Insights API on 28 August 2026, mobile. I am not naming them. They did not ask to be graded in public and the point of this article is the pattern, not the shop.

Two things I have to say out loud before anyone panics at that Lab LCP column.

First, where Google has enough real visitors to report actual field data, the real numbers are good. Site A's real-visitor LCP is 1.1 seconds. Site D's is 2.3 seconds. Both are inside Google's 2.5-second "good" threshold, which is Google's own published boundary and not a number I invented. The laboratory figures beside them are a simulated slow phone on a throttled connection, which is a stress test, not a customer. Sites B and C do not have enough traffic for Google to report field data at all, so for those two I genuinely do not know what real visitors experience. Not measured. I am not going to pretend otherwise.

Second, the total page weights are real and they are heavy: Lighthouse reports 928 KiB, 18,472 KiB, 3,051 KiB and 8,551 KiB respectively. That second figure is roughly 18 megabytes for one homepage, almost entirely photographs of food. That is a lot of bandwidth to spend saying nothing a search engine can quote.

Three of the four also carry no LocalBusiness markup of any kind, which means Google is inferring their address and hours rather than being told. The fourth does declare itself, correctly, as a Restaurant with opening hours and a postal address. Somebody there did the work. They just stopped one step short of the products.

What you need#

• The list of what you actually sell, and what it costs. You have this. It is on a board behind the counter or in your head.
• Access to edit your own website, or the phone number of whoever has it.
• Twenty minutes.
• No money.

That is the whole kit. There is no platform to sign up for in this walkthrough. If somebody tells you that you need Shopify to do this, they are selling you a cart when what you need is a menu.

Step 1 (5 minutes): Write the list as text#

Open a plain text file or a blank document. Write one line per item. Name, then a short honest sentence, then the price.

Portuguese Sweet Bread. Round loaf, baked daily, sold whole. $6.50
Malasadas. Fried to order Saturday mornings only, half dozen. $9.00
Almond Macaroon. Flourless, made with our own almond paste. $2.25

Forty items is a good target. It is roughly a bakery's real range once you stop trying to be exhaustive, and it is genuinely about five minutes of typing if you already know the prices.

Three rules while you type.

Use the name a customer would say out loud, not the name on your invoice. If people walk in and ask for "the sweet bread," the item is called sweet bread, and you can put the formal name after it.

Write a real sentence, not keywords. This is the part people get wrong when they have read half an SEO article. Describing what a thing is beats stuffing the page with the word "bakery" nine times.

Put the price. I know. Prices change, and there is a real fear that publishing one commits you to it. It does not, any more than the board behind the counter does. A price you can update in thirty seconds is infinitely better than no price, because no price is the single most common reason someone closes the tab and orders from the place that told them.

Step 2 (5 minutes): Put the list on one page, as text#

One page. Call it Menu, or Our Breads, or What We Bake. Paste the list in as actual typed text, with headings for your sections: Breads, Pastries, Cakes, Holiday.

Keep the photographs. Put them beside the text, not instead of it. The photographs are why someone chooses you; the text is how they find you. You need both, and right now every bakery I measured has one.

While you are in there, two thirty-second wins: make sure your phone number is a real tel: link so it dials when tapped, and make sure the page says your hours in text. One of the four sites I measured has a phone number printed on the page that is not a clickable link, which on mobile costs calls for no reason at all.

Step 3 (5 minutes): Mark it up so machines can read it#

This is the step that separates "a page with a menu on it" from "products on the internet," and it is a copy-paste job.

Paste this into your page's HTML, anywhere. It does not display. It is a note to machines.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Bakery",
  "name": "Your Bakery",
  "url": "https://yourbakery.com/",
  "telephone": "+1-401-555-0100",
  "priceRange": "$",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "1 Atwells Ave",
    "addressLocality": "Providence",
    "addressRegion": "RI",
    "postalCode": "02903",
    "addressCountry": "US"
  },
  "openingHoursSpecification": [{
    "@type": "OpeningHoursSpecification",
    "dayOfWeek": ["Tuesday","Wednesday","Thursday","Friday","Saturday"],
    "opens": "06:30",
    "closes": "17:00"
  }],
  "hasMenu": {
    "@type": "Menu",
    "name": "Daily Case",
    "hasMenuSection": [{
      "@type": "MenuSection",
      "name": "Breads",
      "hasMenuItem": [{
        "@type": "MenuItem",
        "name": "Portuguese Sweet Bread",
        "description": "Round loaf, baked daily, sold whole.",
        "offers": {
          "@type": "Offer",
          "price": "6.50",
          "priceCurrency": "USD",
          "availability": "https://schema.org/InStock"
        }
      }]
    }]
  }
}
</script>

Then repeat the MenuItem block for each product, changing the name, description and price. That is it. It is tedious and it is not hard.

I did not want to hand you markup I had not checked, so before publishing this I validated it against schema.org's live vocabulary file: 3,219 term definitions, downloaded this morning. Every one of the 7 types and 22 properties in that block is a real, currently-defined schema.org term. Zero unknown terms. You are not copying something I invented.

I also built the full 40-item version to see how heavy it gets, because that is the objection I expected. Minified, forty products with names, descriptions and prices comes to 10.0 KB. The heaviest site in my sample is currently serving 18,472 KiB. The entire machine-readable catalogue of that bakery would cost it roughly one twentieth of one percent of what it already spends on photographs of the same food.

Step 4 (3 minutes): Check that it worked#

Go to Google's Rich Results Test at search.google.com/test/rich-results, paste your page URL, and run it. It is free and it requires no account.

You are looking for your items to be detected, and for warnings rather than errors. Warnings usually mean an optional field is missing and are fine to ignore at first. Errors mean a typo, and the tool will point at the line.

The most common mistake by a distance is a missing comma or a stray quote mark, because JSON is fussy in a way that human writing is not. If it fails, paste the block into a JSON validator, fix the character it complains about, and re-run.

Step 5 (2 minutes): Tell Google it exists#

Open Google Search Console, put your new menu page URL into the inspection bar at the top, and click Request Indexing. If you have never set up Search Console, that is a separate twenty minutes and it is the most valuable free tool a small shop is not using.

Then, separately, put your five best products into your Google Business Profile as Products, with prices. That profile is what fills the map panel, and it is a different system from your website. Doing one does not do the other. This is the single most common thing shop owners assume is automatic.

Where the twenty minutes goes#

Twenty minutes. If you do only Step 1 and Step 2 and never touch the markup, you will still have done more for your findability than most of your competitors managed this year. That is the free fix, and it is the one I would do first.

What I could not measure, and will not pretend otherwise#

The Google Places API returned a 403 for me this morning, so for all four bakeries I have no figures on Google Business Profile completeness, star ratings, or review counts. Those are usually the most interesting numbers in an audit like this. I do not have them today, so they are not in the table.

I also did not verify Facebook or Instagram presence beyond finding links on the sites. Facebook returns an error for real and fake pages alike when checked from a server, and Instagram returns a success for handles that do not exist, so anything I told you about those would be a guess dressed up as data.

And I have not tasted any of the products. On that subject my opinions are strongly held and entirely unmeasured.

When to call someone#

Do Steps 1 and 2 yourself. Genuinely. Nobody knows your case better than you, and outsourcing the writing of your own product list is how you end up with a website that describes a bakery in general rather than yours in particular.

Call someone if you get to Step 3 and cannot find where to paste the markup, if the Rich Results Test throws errors you cannot read, or if your site is built on a platform that will not let you edit HTML at all. That last one is worth knowing about early, because it is a constraint on everything else you will ever want to do, and it is better to discover it on a Tuesday in August than in the week before Christmas orders open.

If you are a Providence bakery and you want the twenty minutes done for you, or you want to know what your four nearest competitors are actually doing, that is a phone call.

774.559.8992 · Joshua.Amado@AvalonPartner.com