Titles on web documents (the bit in the <title> tag, not the headline) were originally specified for administrative neatness, to ensure every document had a descriptive title. They quickly became mainly important for SEO purposes, and have remained so ever since. However, as we know, it’s far from a given nowadays that the search engines will use the title provided in their results, with Google in particular considering that it knows better.
This suggests to me that we might return to an old technique of ‘scraping’ the titles shown in the search engine results to find out what Google really prefers.
Supposing we wanted to optimise our title for a specific search. I think it’s fair to say that what Google is showing for the first 10, 20 or 50 results on that search is what it likes the most. These may not be the titles specified by the website owners, but (in some cases at least), titles that Google thinks are better.
So let’s see what it likes!
The easiest way to do this is just to copy the titles out of the Google results and paste them into a document so they can be examined in a list. We can do this one at a time, or we can just [select-all] the results page and paste everything into a plain text editor to get rid of the graphics and links. The titles can then be copied and pasted into a separate document more easily. Or if you’ve a real data extraction bent, view the source of the results pages and pull out the <h3> lines.
Once we’ve got a list of the first 10, 20 or 50 search result titles isolated in their own document, we can start to look for commonalities. Does Google seem to want to show the manufacturer name? Does it like titles with two or three segments separated by dashes or pipes? Then we can experiment with rewriting our own titles to something more like the format Google seems to like displaying.