This tool addresses a perennial problem in Digital and Direct Marketing: deciding whether the results of a test are sufficiently reliable to indicator how the tested strategy would perform if rolled-out. The tool compares two cells, the test cell and control cell, which normally represents the existing strategy with the larger volume. Typically, Marketers decide that the test results are sufficiently likely to be repeated in a roll-out if the significance is above 95%.

The tool can also be used to design statistically rigorous tests. Enter the control's cell volume and expected response rate. Next, enter the test response rate required to enable to test strategy to achieve at least the same payback as the control. Then, adjust the test cell volume until an acceptable level of confidence is reached (typically, just above 95%).

Only enter data into the cells with blue numbers, then the red number will change in accordance with the model.

This tool is a cloud-based Excel spreadsheet that calculates the significant difference using the normal approximation for a binomial distribution.

Often in Marketing, a small proportion of customers account for a large proportion of value (take for instance the 80-20 rule). This means small that quantities can often be very important. However, because they are small, sample measurements can be greatly affected by random fluctuations. It's the same problem as occurs in evaluating political polling and this tool can be used to understand the stability of those results too.

For example, and using the initial figures in the tool, let's suppose a company has previously found that 10% of customers buy the premium product, making them 4-times more profitable than the remaining 90%. However, a survey has been conducted of 500 prospective customers, and in it only 7.5% of customers said they'd buy the premium product. On the face of it, that looks like a 25% fall in premium customers (and a 6% drop in overall profit). The tool shows that this result isn't very stable. At 95% confidence, a sample result of 7.5% is perfectly acceptable even if the overall level is still 10%. The survey result is therefore a warning sign, but not a definitive measure of a change in customer behaviour.

Only enter data into the cells with blue numbers, then the black numbers will change in accordance with the model.

This tool is a cloud-based Excel spreadsheet that calculates the confidence interval using the E B Wilson procedure with continuity correction.

Do you need help writing a Sales and Marketing Plan? Download this workbook - after you answer around 30 key questions, it will produce for you a costed and formatted plan.

This tool is written in Excel 2010 and contains macros. The costings relate to the UK market.