While we did not have this data readily available, we promised to have a response back to her within several days. I began to look at what I had to run the analysis: My numerator – Products purchased on-site (no-brainer, easy check). Now for my denominator: interactions (nope, too broad), samples distributed (why not, let’s give it a go). So with the two pieces of my sales puzzle, I began to run my conversion.
About half-way through, a thought dawned on me…”what if people had more than one sample?” “What defines a sample?”. Yikes! The logic was flawed! I couldn’t show a conversion rate to a client unless I was sure both of the pieces to my sales puzzle were clean, and using consumers sampled as my base was NOT clean!
So what do you do? Ideally, you don’t have this problem. In the case of sampling, you collect a “Consumes Sampled” metric, so the nice lady who wants to try all three samples before she makes her purchase can without inflating your results.
If you don’t have this data, start collecting it as quick as you can! Providing accurate data that hasn’t been deflated by the little boy who dropped his sample (twice) will make all the difference in the world, and your client will thank you for it.