In my view you need to make some serious improvements in WA for it to be a meaningful SEO tool.
By simply adding your top 10 results and then dividing this total by 10 you often get a meaningless average and recommendation.
Here is an example of a page title of 9 words which fits optimisation requirements.
Keyword Density: the keyword density in title for the keyword "packaging manufacturer" is 22.2%.
This keyword density is too big. Please use your keyword "packaging manufacturer" less often to decrease keyword density close to 1.7%. (This is simply nonsense!)
Keyword Prominence: the keyword prominence in the title for the keyword "packaging manufacturer" is 77%.
This means that your keyword "packaging manufacturer" stands too close to the beginning of the title. Please move your keyword "packaging manufacturer" closer to the end. Prominence must be decreased close to 4%. ?????
I see from your answers that you are aware of this and your answer is to keep our eyes open or not go for 100%.
It would be better if WA did not publish these recommendations if they are rubbish.
It would be better to eliminate the highest and the lowest score.
Recommendations:
1. Eliminating the extremes would help. List the exceptions but exclude them from the average.
2. Another solution would be to incorporate some meaningful standards in your algorithms. E.g Good titles should have a wordcount between 3 and 10 words. So exclude "bad titles" from your percentages and recommendations.
Meaningless averages and recommendations appear to devalue this software and that is a pity.