Many people assume that a search engine by it’s very nature attracts users. However, if you own a new search engine you have to market that engine to your target audience. You can gain users through advertising, word-of-mouth, PR, link exhanges, or SEO (search engine optimization).
SEO on a search engine can be done in two ways.
- You can target your homepage by aiming to meet searchers on the main engines looking for your type of search engine e.g. ‘real-estate usa search engine’
- You can redistribute your content in a useful fashion to the main engines
We used the latter method to multiply out traffic to our local shopping site. It is in Google’s interest to have quality content from a smaller search engine merged in with it’s data-set, and they even make this service possible with Google Sitemaps. However care needs to be taken that there is still some added value in the method of searching once a user comes from Google’s page onto your own, or else none of the users will return to your engine, or even worse – you will be considered a spammer purely after advertising money.
Google’s webmaster tools provides an XML format to submit up to 50,000 URLs in one file, and more can be submitted by spreading your data-set across multiple files. It can take a few months for Google to index this quantity of pages.
The principle involves selecting a number of query terms from your data-set (in our case it was product names)
http://YOUR_SEARCH_URL?q=red+hat
http://YOUR_SEARCH_URL?q=garden+fork
The title of these pages needs to be relevant to the query terms. The descriptions of the products on your results pages will then be indexed on Google, and anyone finding you will be introduced to your engine as a second tier results page.
It is best if there is a call to action to modify the search using your more specific search facility.
Note: I feel a careful decision needs to be taken before embarking on such an effort. There are many people using this sort of technique to introduce a page of adverts which people then click through on. It needs to be for a genuinely unique set of data that is actually useful for an end user.
