So you distributed over 9,000 samples during the week. That’s great!
Now, more importantly, how many people sampled your product?
This important distinction is commonly lost in reporting on data collection. Double and triple sampling is surprisingly common, particularly because you want to be sampling more than one version of your product, and it makes sense to let people try all the versions you have. So adding this into statistics you collect is easy enough, but does it offer us anything more than that you weren’t getting from samples distributed?
Well the obvious benefit is that it’s more accurate, at the cost of likely being a smaller number than samples distributed.
There are other benefits as well. With this new information in hand we can compare to interactions to determine how many people aren’t sampling despite talking to a brand ambassador. So let’s say we have 100 consumers sampled and 150 interactions. That leaves us with 50 consumers that are interacting with a BA but not trying the food.
Why?
Are we in a market where the product doesn’t appeal to 33% of the customer base? Are we not making samples quickly enough to give one to everyone who walks by? Are we in a location that doesn’t lend itself to sampling? With this information we can refine both the program, as well as activation locations to optimize our sampling.
Finally, and most valuably if you are recording products purchased on site, you have insight into instant purchase intent. If you sample 500 consumers, and see 50 products move off the shelf, you have a solid 10% on site purchase intent.
This type of information is not only valuable for evaluating individual markets, when combined with survey data allows us to do an event by event immediate value analysis (if 10% of those were current customers who were going to be purchasing anyway, then your event generated 45 new customers).
So what can we gather from this? The value of granularity is what appears first and foremost to me. Following that, the more diverse your data collection (within reason) the more insight it can offer you. Initially, it may seem as though you are collecting redundant data, but in fact that slight redundancy lets us better understand both pieces of data.
Photo Source: http://farm8.staticflickr.com/7156/6820671197_ab8c1f1052.jpg