I'm looking for a scientific but not too complicated way of showing how effective an anti-microbial is.

I test various anti-microbial formulations (not commercially available) in animal feed against Enterobacteriaceae . It usually happens when we make a change to a formulation and want to test if the new formulation works as well as the old formulation. So I would usually have microbially contaminated feed that I would sub-divide the sample:

• Untreated animal feed
• Animal Feed treated with Formulation A (Formulation currently being used and thus the Control)
• Animal Feed treated with Formulation B (Formulation being tested to replace Formulation A).

The results should show a reduction in Enterobacteriaceae levels. Once I have the results how do I represent them. My supervisors always say when looking for an effective anti-microbial they are looking for one that can cause a 1 Log reduction in Enterobacteriaceae but this doesn't make sense to me because a particular dosage of a compound (eg. 1 kg per ton) that takes my count from 100 cfu/g to 10 cfu/g i.e. 90 cfu/g reduction isn't as effective another compound that takes my count from 10000 cfu/g to 1000 cfu/g i.e. 9000 cfu/g reduction but both of these are a 1 Log reduction in numbers.

I've also converted to percentage to show the percentage decrease in cfu/g as compared to the untreated sample but that also didn't seem a good enough way to show the results.

Should I use the current formulation as the standard and just compare the new formulation to it?

Example:

Current formulation caused decrease in cfu from 10000 to 3000 cfu/g thus 7000 cfu/g reduction

New formulation caused decrease in cfu from 10000 to 4000 cfu/g thus 6000 cfu/g reduction

So I can say that the new formulation is 85.71% as effective as the current formulation.

Any insights would be appreciated.

I've done three experimental reps. The third one did not look similar to the first two so I won't post those results.

With the first trial I only ran the untreated, Current formulation and new formulation. With the second trial I included a potent positive control to test if the model I'm using actually works