- A unique title for each page, of roughly 50 to 60 characters, naming the trade and the town before the brand.
- The meta description does not make a page rank, but it makes people want to click: write it for a person.
- A single H1 stating the subject, subheadings in order, images described: the page reads better, for everyone.
What Google reads first
A web page is made of two parts. The body, which visitors see: text, images, buttons. And the head of the code, the head section, invisible on screen, where a few lines are addressed to browsers and search engines. Google reads both, but some lines weigh more than others in understanding what the page is about and in composing the result it will display.
These lines are often left at the default settings of whatever tool was used to build the site. You then find pages all titled ‘Home’, or bearing only the brand name, descriptions left empty or copied from one page to the next, images with no description. None of this stops the site from displaying; all of it stops it from being understood. For a local business, the difference lies in these details: the search engine must know what your trade is, where you practise it, and which page answers which question.
The examples that follow concern a fictitious business, the Maison Lumière bakery in Arras, invented for the purpose. The name does not refer to any real company; it serves only to show, on a concrete case, what each rule becomes once applied. Replace the trade and the town with your own, and the reasoning stays the same.
The title, the page's shop sign
The title tag is one of the clearest signals of what a page is about. It also usually serves as the clickable title in search results, and as the name of the tab in the browser. It is therefore both a pointer for the search engine and a shop sign for the passer-by. In Japan, the kanban, the wooden sign hung above shops, said in a word what was sold within; a page title does well to do the same.
Space is limited, and it is measured in width rather than in letters: beyond roughly 600 pixels, the displayed title is truncated and ends with an ellipsis. In practice that comes to around 50 to 60 characters, depending on the width of the letters used. So put the essentials at the start, where the eye lands and nothing will be cut: the trade, then the town, then the brand, after a separator. A longer title can still be read by the search engine, but its end may never be seen.
An example, before and after
For the home page of our fictitious bakery, ‘Home’ says nothing to anyone, and ‘Maison Lumière’ says nothing to someone who does not yet know it. ‘Artisan bakery in Arras | Maison Lumière’ states the trade, the town and the brand, in that order, in fewer than 50 characters. The page for cakes to order would have its own, different one: ‘Cakes to order in Arras | Maison Lumière’. Each page thus carries a unique title, which says what it contains and sets it apart from its neighbours.
The meta description, an invitation
The meta description is a short text, also placed in the head of the code, that summarises the page. Google says so plainly: it is not a ranking factor. It matters all the same, because it often serves as the snippet beneath the title in the results, and that snippet weighs on the decision to click. It is an invitation more than an argument: two sentences telling the person they have come to the right place.
Google may replace it with a passage from the page that better answers the search typed; this happens often, and it is no failure. Write it all the same, one per page, different for every page. Beyond roughly 150 to 160 characters it is often truncated, and the limit varies by device: put the useful information in the first half. Avoid lists of keywords, capital letters and promises the page does not keep, which disappoint the moment someone clicks.
For the Maison Lumière home page, that might give: ‘Sourdough bread, pastries and cakes made every morning in Arras. Order online and collect in store, Tuesday to Sunday.’ Every word informs: what is sold, where, how to order, when to come. Nothing in it is vague, nothing superfluous.
The H1 and the subheadings, the page's plan
The H1 is the main heading visible on the page, the one visitors read when they arrive. Each page needs one, and it states the subject in clear terms. It need not be identical to the title tag: the title is written for the search result and must fit on one line, while the H1 speaks to someone who has already arrived and can breathe a little more. But the two must speak of the same thing, or the search engine and the reader receive two answers to a single question.
Beneath the H1, H2 subheadings, then H3 within a section, sketch out the plan of the page. They help hurried readers, who scan the page before settling. They help people who browse with a screen reader, and who jump from one subheading to the next as one leafs through a table of contents. And they help search engines understand how the content is organised. A subheading is chosen for what it announces, never for its font size: larger text is a matter of style, not of hierarchy.
The plan of a home page
- One H1: ‘Sourdough bread, baked every morning in Arras’. It states the subject of the page in one sentence.
- An H2 ‘Our breads’, with two H3s beneath it: ‘The starter’ and ‘The flours’.
- An H2 ‘Ordering a cake’, which links to the dedicated page.
- An H2 ‘Visiting the shop’, with the address, opening hours and phone number written out in full.
Images and the page address
Alternative text, the alt attribute of an image, describes that image for anyone who cannot see it: a blind person whose screen reader reads the page aloud, a visitor whose connection has not loaded the image, a search engine that does not look at images as we do. It says what the image shows, in one simple sentence: ‘Sourdough loaf sliced on a wooden board’ rather than ‘image1’ or a string of keywords. A purely decorative image, a pattern or a rule, receives on the contrary an empty alternative text, alt="": the screen reader then passes over it in silence, which is exactly what is wanted.
The page address matters too, for the search engine as for the person who reads it in a result or receives it in a message. A good address is legible, short, made of words rather than identifiers: /order-a-cake says it all, a string of numbers and parameters says nothing. Above all it must remain stable, for an address that changes breaks the links that pointed to it and loses what it had gained, unless a redirect is put in place properly. It is one of the pitfalls of a redesign, set out in the article on redesigning a site without losing its rankings.
Structured data and the canonical address
Structured data is a block of code, invisible to visitors, that describes the page in a vocabulary shared by search engines, that of schema.org. For a business, the LocalBusiness type, or one of its more precise types such as Bakery for a bakery, lets you state the name, address, phone number and opening hours in a form that leaves a machine no room for doubt. The rule is strict: what the structured data says must match what the page displays. It helps search engines understand, with no guarantee of an enhanced display in the results.
This data extends your business profile: name, address and phone number must be identical in both, as the article on the Google business profile shows. Google offers a Rich Results Test, which checks that the block is well formed; it does not say whether it will be displayed, only that it can be read. An error in this block does not break the page, which explains why it often goes unnoticed for a long time.
The canonical tag
The same content can often be reached at several addresses: with or without ‘www’, with a tracking parameter added by a campaign, with or without a trailing slash. For the search engine, these are so many seemingly separate pages saying the same thing. The rel="canonical" tag designates the reference address, the one you want to see in the results. Google treats it as a strong hint rather than an order, and follows it all the more readily when it agrees with your internal links and your sitemap.
The top of a page, line by line
Brought together, these rules fit in a few lines, at the head of every page and then at the start of its body. The illustration below shows them in their usual order for the home page of our fictitious bakery: the title, the description, the canonical address, then the H1 and the structured data. None of these lines is hard to write; the difficulty is to write all of them, for every page, and to keep them up to date as the site changes. It is a craftsman's work more than a technician's, made of rereading.
On our own site, every page carries a unique title of 60 characters at most, with the brand at the end. An automatic check rejects a title that is too long, or a description outside 70 to 160 characters, before anything goes live: the page is not published until the rule is met. This safeguard is nothing spectacular, but it prevents a new page, written in haste one evening, from reaching the results truncated or mute. We apply the same discipline to the sites we build for our clients.
The most common mistakes
- Pages all titled ‘Home’, or with the brand name alone.
- The same title and description on every page, left as they were by the site-building tool.
- No H1, or several competing for the role of main heading, because style has been confused with structure.
- Images with no alt text, or with the file name standing in for a description.
- A carefully filled-in meta keywords tag: Google does not use it to rank pages.
A good title does not describe what you want to sell: it names what the person is looking for.
Checking what Google displays
Once the lines are written, look at the result as Google shows it: search for the name of your business, then for your trade and your town, and read your titles and snippets as a stranger would. Changes do not appear at once, since Google must first revisit the page; in Google Search Console, the URL inspection tool lets you ask it to read the page again. The performance report shows, page by page, how many times it appeared in the results and how many times people clicked. A page that is shown often but rarely chosen often has a title or a description to rework: that is where to begin. The in-depth work, content and links included, belongs to our search discipline.
What you can do this week
Tick as you go: nothing is saved, it is simply a reminder.
Founder and director of Digital Air Media, a digital creation house opened in Paris in 2017. He writes here what the house applies to its own pages before offering it to others.