Get best of the week

What is the result of email marketing: how to make a forecast

Projects for the implementation and maintenance of email marketing can cost from 120,000 to 300,000 rubles per month. Therefore, the business owner wants to understand what this investment will ultimately give him.

The forecast of the result of email marketing allows you to understand:

  1. How will the customer base grow?
  2. How many letters will be sent?
  3. How will the distribution rates change?
  4. Will the average bill for mailing orders change?

There are also non-standard forecast options when mailings do not directly generate income, but are an additional attraction of customers to various stages of the sales funnel. However, the process of constructing such forecasts differs only in the final calculations, when it is already required to get answers to the questions posed.

Let's understand all the intricacies of the forecasting process for newsletters.

Draw up a mailing list


Before making a forecast, you need to immerse yourself in the client’s business, understand his desires and draw up a plan that would bring maximum benefit. For this part of the work, we are responsible for sales managers and marketers.

The result is a well-developed communication scheme, where the mailings that already work with the client are highlighted in red:

Well-developed communication scheme with newsletters that already work with the client

And if the client did not have any mailings configured? Anyway, we analyze the business and draw up a diagram from scratch.

We collect the source data


We build our forecasts for the year ahead with a monthly breakdown. This separation helps in the process of further work to keep track of whether we are moving in the right direction. Therefore, we collect initial data for each month of the past year.

For this we use:

  • data from analytics services;
  • Data from mailing platforms
  • if necessary, data from the client (for example, if the client is going to increase traffic, then the expected percentage of growth is needed).

Let us consider in more detail what indicators are required to be collected and why they are needed.

Sessions, Users, and New Visitors


We receive data from Google Analytics or Yandex.Metrics. You can calculate the database growth by the number of users or sessions - it depends on the task. We advise you to collect everything at once. Make sure that the service returns results using 100% of your data, that is, without sampling.

Initial mailing data for the year



Promotion data


We collect this and all other data from any platform: Mindbox / MailChimp / ExpertSender or others.

We need to know:

  • how many letters were sent;
  • how much is delivered;
  • number of discoveries (unique value);
  • number of clicks (unique value);
  • number of transactions;
  • income.

Promotion data

Delivery Rate, OR, CTOR, transaction rate and average receipt are calculated using formulas.

If possible, use unique discoveries and clicks - this will allow you to see data in the context of users.
If the platform does not collect or erroneously read transaction data, download the data from an analytics service, for example, Google Analytics, provided that the e-commerce data transfer is configured correctly. Select the desired source / channel and month, check for sampling. Do not forget that the transaction rate in GA is calculated from sessions (in case you will use not sessions, but users in the forecast).

Data from Google Analytics



If there were no newsletters, then you will have to focus on the average indicators on the topic + expert opinion. Also consider whether the products are premium or budget, because the indicators for the mailings they have are different.

We have quite a variety of customers, so there are statistics on all the main metrics for various fields. For example, the average values ​​of Open Rate and Click To Open Rate in promotional mailings for:

  • cosmetics in one price category: OR - 7.6%, CTOR - 10.5%;
  • developers (budget segment): OR - 37.3%, CTOR - 15.3%;
  • clothes: OR - 10.8%, CTOR - 26.4%.

Trigger Mailing Data


Before you start collecting data, you need to determine how detailed the result you want to get.

There are three calculation options:

  1. . , . Metric chain for forecasting trigger mailings

    , , , .
  2. . , / . , (welcome, / ). . Metric Chain for Forecast Forecasts
  3. . , . , ,


Base dynamics

If this is a non-standard forecast for mailings - for example, as we said at the beginning, you need to predict not revenue, but growth of leads in the sales funnel - then you can request data from the client.

Analyzing data


When all indicators are collected in one file, you can begin to analyze them.

Evaluate the following indicators.

Change in traffic over the previous 12 months


This will help determine the seasonality, identify possible anomalies.

Try to understand the causes of the abnormal values. For example, a sharp increase in traffic in November in most cases is normal: companies spend Black Friday. But, if some unique campaign was held, which caught up with a lot of traffic this month, next year you should not expect the same traffic in this period.

Customer base dynamics


Determine how many people unsubscribed and signed up each month. Check if there is a connection between the growth of the base and traffic in each month.

Number and indicators of promotional and trigger mailings


Determine the approximate number of monthly promotions. And trigger mailings - if you use a generic version. Please note that the number of trigger mailings is calculated as a percentage of the base in a particular month. Does seasonality, stocks and other external factors influence OR, CTOR, conversion rate?

Determining the approximate number of promotions



Indicators for trigger mailings with a generalized forecast



Income


Look at the change in income and average bill. Are the changes justified? If yes (for example, permanent stocks), do not forget to take this into account when forecasting.

Trigger data for a detailed calculation option


Trigger data for a detailed calculation option

Determine what percentage of the total trigger mailings amounted to a particular chain or letter, given the launch date. For example, if a chain of letters “Abandoned basket” was set up only three months ago, then it is necessary to compare the volume of shipments on this chain with the total volume of shipments of trigger mailings for this period. Otherwise, you will receive incorrect indicators.

Evaluate key indicators and their relationship with the type of trigger, stocks, and so on.

Getting to the forecast


We remind you that we build our forecasts monthly for the year ahead. You can, of course, build on a longer term, but we do not advise. The veracity of the results is decreasing from year to year.

Forecasts are made in the following categories.

Traffic


We do not affect traffic in any way, unlike advertising. Therefore, it is worth clarifying whether the company plans to use new methods of advertising (or vice versa, reducing existing ones). If changes are planned, we find out the expected monthly percentage of traffic growth.

Now, given the data from the client and a retrospective analysis of traffic for the previous year, we can build a forecast.

Traffic forecast

Base size


At the company, we are sensitive to the subscriber base, so before working on mailings, if necessary, we take a number of steps towards obtaining the current base, which must be taken into account when building the forecast.

What should be considered when calculating the client base?

Base cleaning


Determine the percentage of invalid addresses. Put the decrease in the size of the base for this amount in the first month of settlements.

DOI Size


Double Opt-In (two-stage subscription) mechanics allows you to filter out invalid addresses and maintain the current customer base. The amount of DOI depends on the type of business. If possible, rate on similar customers. In the forecast, consider this indicator monthly.

Base gain


Analyzing the initial data, we can notice that the monthly percentage of the base gain from traffic is in a small range, so for calculation you can take the average indicator. Do not forget about DOI.

The average indicator for calculating the forecast for the size of the base

Unsubscribing


You should already have data on the percentage of unsubscriptions in each month after analyzing the source data. The average indicator can be used in calculating the predicted results.

Or consider unsubscribing in the next steps of forecasting - I’ll talk about this later.

Now you can calculate the size of the base at the end of each month.

Base size calculation

Promos


Based on the mailing list compiled by managers, we know the approximate number of promotions per month. If you do not have such information, rely on baseline data and plans.

We got all other indicators from the analysis of the source data, making the necessary adjustments.

Indicators for forecasting promotions

It remains only to substitute everything in the desired form and calculate.

Forecast Calculation

Your forecast will be even more accurate if the values ​​of OR, CTOR, transaction ratio and average check take into account seasonality, stocks and other external influences. This condition also applies to the forecast for trigger mailings.

In this part of the forecast, the number of unsubscriptions from promotions is calculated. More unsubscribes from trigger mailings will be calculated. Their amount can also be used when calculating the unsubscribe from the previous paragraph.

Trigger Mailings


General forecast


As in the promotions, we have identified and adjusted the main indicators for the calculations.

Trigger forecast metrics

And framed in the forecast.

Calculation of forecast for trigger mailings

Detailed forecast


For each new trigger message, specify OR, CTOR, percentage of unsubscribes, and transaction rate. We take into account in what chain and in what place is the letter. As a rule, the later departures, the lower the rates.


If possible, when determining forecast indicators (including the average bill and the volume of shipments), rely on similar conversations. For example, if you have an abandoned basket, you can use rough metrics for the abandoned view. But you cannot use OR with transactional letters for a welcome chain.

We make a calculation for each letter.

Calculation for each forecast letter

Mixed forecast


We most often use this type in forecasting trigger mailings. The method combines all of the above steps.

We collect the results of the forecast


Forecast Results

Do not forget that the more you study the source data, the business itself and the current situation, the more accurate the forecast will be.

More articles:


All Articles