Transcription

There’s a common belief that data is something pure that just tells it how it is and can always be trusted, but that’s nonsense. Like words, data can be manipulated and mis-represented in all sorts of ways, and choosing an inappropriate time time within which to present your data is a really common example of how that happens.

All too often I hear people say “Traffic from a particular source has gone up by x%, or conversions have doubled” and then when you ask over what period, you’ll instantly realise that the data doesn’t mean anything. Maybe they’re comparing January to December in a market where December is dead and January is really busy (and the senior DM’s you are reporting to will know that!), or perhaps they’re looking at data from this week compared to last week, when perhaps the website or app doesn’t generate enough traffic for that data to be reliable.

The point is just to really think about the time frame in question. Most commonly I would want to look at medium term trend data, so that might be over six to 12 months, or to be comparing it to the same month the year earlier. That is unless you are dealing with huge data volumes, in which case looking at shorter term metrics will be reliable enough to take meaningful insights from.

See you next time.