So the URL's are fixed. What can we do next?
One of the things we did with the previous site was to rename all the images to contain 'real life' words. So instead of being called something cryptic and automatically generated by the server (eg. 1545687464_d.jpg) we made the filename contain the product name. In this image 
you can see a screenshot containing the webpage for a Jeld Wen Canberra Hardwood Folding Sliding Door Set. I have also shown a picture of the image file itself. This image is called _Jeld_Wen_Canberra_Hardwood_Unfinished_large_20080625141651.jpg_. The first part (Jeld_Wen_Canberra_Hardwood_Unfinished) comes directly from the product description. The next part (large) is the scale of the image (because we also have thumbs and extra large detailed views) and finally the numbers are the date & timestamp when we uploaded the file.
The other thing to ensure we did was to have a very descriptive 'Mouse Over' text. This also helps in browsers with images turned off. This image shows the pop-up from another set of folding doors called Wellington.
So what is the benefit of this? Firstly the hope is to get the desired relevant keywords on the page more, but in a context sensitive manner (ie. it is not a good idea to just spam the page with these keywords as this would be marked down by Google). Secondly, but less obviously as it does not work too well just yet, is the hope that Google Images will pick up on these files.
So now we have at least twice the amount of keyword use on each of the product pages.
The next thing we needed to change was the CSS to ensure the correct use of <h> header tags.
