Have you ever heard that the longer the subject line is, the worse it performs? I know I have. But there’s very little truth to that myth and this graph will prove your suspicions wrong.
Use the Correlation Analysis to make data-driven conclusions
Navigate to Data and Analytics within your Predict account and select any dataset. Scroll down to the bottom to find a graph titled "Correlation Between Open Rate and Character Count."
That was a mouthful, wasn't it?
This graph is pretty easy to decipher, luckily. The X-axis is open rate and on the Y-axis is character count.
What do all these colorful dots mean? Each red dot represents a subject line from the dataset. Each blue diamond represents a subject line from Persado’s industry data.
Move your eyes to the bottom right of the screen to find the Correlation Coefficient (R-squared value). This will tell you if there's truly a story between open rate and character count.
Any value in between -0.2 and 0.2 means there is no correlation between character count and open rate. The closer you get to -1 or 1, the stronger the negative or positive correlation is, respectively.
An example of a data set with no linear correlation.
An example of a data set with a weak, linear correlation.
Looks like this marketer doesn't need to stress about keeping his/her character count under control.
Understand the math behind the scenes
First off, let's start off with the "ingredients" behind the formula.
- A dataset of subject lines, with varying open rates and character counts.
- A scatter plot.
- An R-squared correlation test.
It’s important to understand why we use a scatter plot. Scatter plots show the relationship between two sets of data. In this graph, we want to show the length of the subject line (Character Count) and the performance of the same subject line (Open Rate).
Also, we want to find out if there's any relationship between the two sets. How does the length of the subject line affect its performance? Does it affect performance at all? We find our answers with a regression analysis. In other words, we calculate the line of best fit.
A line of best fit is what it sounds like: it's an estimation of a slope between two variables: x and y. If the graph has a negative slope, this means there is negative correlation between the two variables (i.e. the greater the character count, the lower the open rate.) On the other hand, a positive slope shows a positive correlation (i.e. the higher the character count, the higher the open rate).
The FINAL step is to find out how well the data fits the slop. If there's a positive slope that significantly fits the data, then you can argue there's a positive relationship between two variables. If a negative slop insignificantly fits the data, than there's no relationship.
*Whew*! I hope that helps explain things!
Still have questions?
Please reach out to us at firstname.lastname@example.org. Thank you!