The meta description meta tag is considered to be a tag useful to use on web sites, whether they are large or small. For a small web site (<50 pages), it’s entirely feasible to write the meta descriptions by hand. What happens, though, when we’re talking of a web site with hundreds or even thousands of pages (online stores, large publishing web sites)?
A first solution is to automatically take the first sentence in the description/text of the page. The problem with this approach is that the sentence an article starts with is likely to have much more or much less than the recommended size of 110/130 (minimum) to 155/165 (maximum) characters with symbols. In this case, my advice is to avoid using the meta description at all and just let Google determine itself what’s the best text to take from the page.
I want to present, in this article, a second solution for creating meta descriptions for large web sites.
The formula for my solution looks like this:
[title of the article/page/product] [brand name] [small motivational text 1] [small motivational text 2] [small motivational text 3]
For example, a meta description could look like this:
Google Nexus 6P – Amazon.com. Trusted by millions of shoppers worldwide. Shipping within two working days. Free delivery within the US.
I have used the following replacements:
[title of the article/page/product] = Google Nexus 6P
[brand name] = – Amazon.com
[small motivational text 1] = Trusted by millions of shoppers worldwide.
[small motivational text 2] = Shipping within two working days.
[small motivational text 3] = Free delivery within the US.
How to generate the [small motivational text #] text portions, and still avoid having duplicate meta descriptions on the web site? My suggestion is to create some buckets of texts for each of the 3 blocks of texts.
[small motivational text 1] = Trusted by millions of shoppers worldwide. / 30-days return guarantee. / One of the best prices on the market. / Big discounts compared to brick-and-mortar stores. / etc.
You should create the buckets of texts with a different number of texts in each one (I’ll explain why below) – for example, the first bucket 10 texts, the second one 12, the third one 14. Also, you should avoid having duplicate texts from one bucket to another. So, if in the first bucket you talk about discounts and prices, avoid doing so in the second and third bucket. How should you attribute texts from the buckets to each of your pages in your web site? My suggestion is to do the following (if you show the following algorithm to a programmer, that person should be able to implement it):
- Convert each letter and symbol in the URL of each page to an ASCII code.
- Make the sum of each ASCII code above and get a large number.
- Get the module of dividing the number above to the number of each item in the bucket. Since the buckets have a different number of items, you will get, each time, a different description for each URL.
I suggested, above, to use the ASCII code of the URL because that’s the least likely to change when doing changes to a web site. Also, if you’re going to do a site migration in the near future, you could use just the second part of the URL, so not this: www.site.com/folder/page.html, but this: /folder/page.html.
Note: Also see the Yahoo! Group on which I present similar issues: IMRo. To join, email firstname.lastname@example.org and reply to the confirmation email.