# Could one build a model on how to optimize e-mail marketing.. and forecast its value?

- March 9, 2018
- Uncategorized

What if we could forecast our e-mail marketing efforts? What if you know how much it would increase your revenue if you changed your e-mail frequency, the amount of new permissions you get per month or the CTR of said e-mails.

Surely there must be some truths that we can rely on (in almost all cases).

Even though no model is perfect, it might improve our approach, our thinking, our tactics and the overall strategy of how to approach e-mail marketing optimization.

## Try the calculator

I have build a calculator that you can use for free (but when sharing it, please give credit – that is all i ask).

Let me explain how it works:

The calculator obviously uses the mathematical model in its entirety. The way you use it is by editing the column “Current” to your estimated current metrics, and then you only input changes to the status quo setup in the 4 columns for hypotheses. Then the calculator will constantly re-calculate the present value for the current and each of your hypotheses and show you the “trend” of profits in the bottom graph.

See this .gif where I first edit the “status quo” column and then experiment with more frequency, which adds extra costs and also makes open rates potentially drop. Play around with it and find a sweet spot!

Then to use it, follow * this link* and then click “File > Make a copy” and save a version of the Sheet to your own Google drive. Otherwise you will be competing with others in the same Google Sheet ;o)

## The mathematical model behind

I love this part, but most of you do not want to read this.

So feel free to skip this part about the underlying math.

The **open rate** is a measure of how many percent (in decimals) that open the average promotional e-mail. **Click through rate** is a measure of how many percent (in decimals) that click their way from the email to the website or webshop. **Conversion rate** is the rate that you convert people from newsletters and emails to customers once they reach your website. **Average order value** is, as you guessed, the average order value, however, only from the traffic you get form your newsletters or email.

**Existing permissions** is the amount of permissions you have today. **New permissions** is the amount of new permissions you get per month (or whichever t-period you want to use). **Permission growth** is the percent (in decimals) that your amount of new permissions grow with month over month. The **unsubscribe rate** is how many percent of your total permissions you typically lose per email you send to your list.

**Frequency** is the amount of emails per month, t. The **die rate** is how quickly people change emails, and therefore by default bounce from your list. Most of the experts I have talked to on estimate around 15% per year, which is why you should probably just keep the 1,17% per month.

**Net present value** is how much money you will make in todays worth. So if you use your next best marketing channel as the discount rate, then you will typically find the NPV to be significant lower than if you use the rate for borrowing money, as the discount rate.

**Initial cost of investments** represents the amount of money you will invest today to either keep status quo or inflict a change. **Budget** represents the monthly cost of the department sending emails, software fees and so forth. Inflation is a metric for how much prices increase year over year.

**Discount rate**, might be the most difficult to explain. Lets image you put $100 in a portfolio of stocks today – then you might have $161 in 5 years (10% growth per year). If this is your “alternate” use of your money then you want to know if you should, instead, spend your money on email marketing.

## Unknowns

The math is correct, however, nobody knows how much the open rate, click rate or conversion rate will drop if you double your frequency. So, we do not know how they all the variables are internally correlated, but we can make qualified guesses and that is exactly the point of economics models.

## Model assumptions

You can either create a quite complex model that includes pretty much very variable, like this on, and then you don’t have to rely as much on assumption. On the other hand, you can create a simpler model, but then it would require more assumptions.

Say you wanted unsubscribe rate to be a function of frequency. It might be an exponential function, a linear or whatever, but the higher the frequency the higher the unsubscribe rate. However, in this model you can actually decide more variables for yourself – that is cool.

Coolness comes at a price, because you need to think harder. If you increase frequency from 2 to 4 then what might happen? Your budget might increase, unsubscribe rate might increase, open rates might drop, click rates might drop and conversion rate might drop? We don’t know for sure but you need to consider the other variables as you test your hypotheses.

## The biggest weakness

Not every email you send it promotional/commercial. You might have an anti-churn or re-engage automation flow that is not at all about selling products. Where do you input those emails? Well you don’t, because they don’t effect your cash flow directly.

However, if you believe that investing 100K in an re-engage flow will increase open rates of commercial emails by 20% then that is a feasible hypothesis to match up against other initiatives. Simply add 100K to initial investment and increase open rate by 20% of the 2 commercial emails per month you have.

Good marketers will not rely on email newsletters but have smarter and more segmented flows. However, this still offers a way to test and validate ideas quickly.

**CHRISTIAN HØJBO MØLLER**

Christian er CMO i sin start-up Candidlab som i dag er ekspanderet til 11 lande. Han har tidligere som Lead- og Senior Konsulent i verdens største mediebureau gruppe, GroupM, arbejdet med kunder som HBO, Ford, Just-Eat og Toyota.

**Connect på LinkedIn, følg på Twitter eller send en besked.**