Please look at the attached and let me know where the difference is?

The attached show 2 reports.
- Yr to date
o Shows the month to month comparable predictive
- July (benchmark graph)
o Shows the comparable for July

Problem I have is that surely July in the yr to date should reflect the same figure as the July benchmark Graph when looking at the [censored] Totals?

There is a perfectly legitimate reason why they are different.
The reason is that the 'comparable' for the year up to date is only taking that into account that is comparable from the beginning of the year. In other words, it will not take something in account that came online in the first 6 months of last year.
The 'comparable' for the benchmark is taking into account all that is comparable for July only. This means if it came online in the first six months of last year, if will be taken into account. That is why this one is higher than the other one.
The reason why the algorithm is different in this respect, is because with the year to date, the user wants to identify a trend over the year, in other words, he is actively comparing the last year graph bars to each other. If we do not discount meters that came online after the start of last year, we will show him a graph that increases, and he would 'compare' that to the previous months and thinks he is using more energy, when in fact the existing ones perhaps did not.
Anyway, hope that clears it up!