Financial Model of the loyalty program — how we were calculating it
Exploring the possibilities of increasing sales with the help of loyalty programs. Our take on Customer Analysis & Data Segmentation.
How do loyalty programs make money? In this article, we’ll answer this question by showing a project we’ve recently worked on. This case study demonstrates how customer analysis, proper data segmentation, and an effective loyalty program helped a brand make money and foster customer loyalty.
Our team of business analysts knows how to gather data on businesses. They analyze the number of purchases, the amounts spent, clients, client loyalty, campaigns, and budgets. Then, they put this all into a united report. This process is necessary to understand what is going on in marketing and sales in the language of numbers.
This job allows us to look at businesses from the point of view of real data. This includes calculating the loyalty program financial model that will be supporting business objectives.
How We Chose a Loyalty Program Type
The brand has rather specific products, so the clients don’t make purchases often and usually they don’t spend much money — about $30. In this case the most successful loyalty program type is the discount growth depending on the gained purchase amount.
For this type we need to calculate:
- what amounts will make the discount grow;
- what discounts there will be;
- the starting amount for a client to enter the loyalty program;
- the amount of profit this loyalty program can bring to business.
To do these calculations, first of all, we needed to gather all of the data on transactions and clients, as well as their interaction with the brand.
We’ve Analyzed the Data on Clients and Purchases
Customer analysis plays a crucial role in building a loyalty program, especially if you’re approaching this process professionally. If you don’t know whether your customers are loyal or not it’s difficult to see the full picture.
We’ve downloaded the data on the purchases from the online store and offline cash registers. This data segmentation allowed us to find the loyal to disloyal customers ratio:
And then, for more effective customer analysis, we organized them into a more detailed graph — we examined the time period between the purchases to see how often the clients make purchases:
Then we’ve analyzed the loyal customers by the amount of the purchases and the time period between the purchases:
Individual filters allowed to learn the amount of the loyal clients with the certain amounts of purchases:
We use other tools for dividing the clients by loyalty, like RFM analysis.
Building a Sales Forecast Based on The Data Segmentation
We had the following data:
- the amount of clients in the last period for every month;
- the sales tendency;
- the clients to purchases ratio (%).
Based on this data segmentation we built a forecast for the period from October 2019 to September 2020, which is when the new loyalty program was to be implemented.
It was a “bare” forecast — what the sales would be if the loyalty program wouldn’t be implemented.
We’ve Estimated the Possible Financial Models
The next step is a hypothesis about the sums of the purchases and the discount percentages.
Based on this hypothesis later we will calculate how much of the additional revenue the loyalty program will bring to the business.
We’ve thought the hypothesis through together with the client. It is based on the positioning and the products’ price. If the brand is premium, the amount needed to get the loyalty program card and steps to go to the new status will be higher than if the brand is inexpensive.
Together with the client, we’ve picked three calculation options:
Still, we need to take into consideration the amount of the new clients to come. We don’t know the exact growth, but we can suppose its forecast data. For our predictions to be more accurate we’ve suggested three possibilities for the increase of new clients: pessimistic, conservative, and optimistic.
However, it may depend on the discounts too. Maybe, the bigger the discount, the more clients will come. This is why, depending on the hypothesis model, the increase can be different:
I.e., as a result, we’ve been calculating every model in three possibilities: pessimistic, conservative, and optimistic. Also, we did this independently for online and offline, as the sales dynamics differed in both cases. This is where data segmentation came in handy again.
We’ve Calculated the Number of Clients in the Loyalty Program Potential Status
The next step — gathering the existing data on the clients, purchases, and purchase amounts by the status of the potential loyalty programs — is crucial in such cases. Now we understood how many people (%) were in either status.
We’ve Highlighted the Groups of Those Who Need To Buy Less To Go to the Next Status
There are those who are far away from the new status and those who are close to getting a discount in the highlighted groups. For example, the status begins with $150 and the buyer spent $60 — this buyer is far from getting a status. Those who’ve spent $130 are way closer to the discount. They are easier to convert into a new status. So we’ve highlighted the groups of clients by the sign of being close to the status:
The more the client bought, the more loyal they are. Those who’ve bought for $200 are more loyal than those who’ve bought for $65. It is easier to finish selling to them and therefore the higher the status, the larger the steps towards it.
We’ve Forecasted the Additional Revenue From the Loyalty Program Implementation
The final part is to calculate uplift — what additional revenue can the company get from implementing the loyalty program.
We’ve gathered data into a table to do this:
- the percentage of the clients on different steps towards statuses;
- the total number of the clients on different steps towards statuses;
- the presupposed average upsell check;
- the presupposed average upsell check except for the value of the gift, discount, and delivery;
- the presupposed conversion — how much the clients will want to buy additionally to go to a status.
We did these calculations by months for every possibility of the loyalty program in three forecasts (pessimistic, conservative, optimistic) and independently for offline and online. As a result we got an uplift by months for the future period for every option.
We’ve also been taking the purchases up to $65 and calculated the uplift by them for a year at once. Though $65 doesn’t give a status, these clients ought to get a gift, so they should have also been taken into consideration.
We’ve Chosen the Most Profitable Loyalty Program Financial Model
As a result we got a forecast about what uplift we can expect from all the presupposed options of the loyalty program. We’ve compared the trade and sales forecast with the loyalty programs with the forecast without them and managed to choose the most profitable option:
We understand the expenses for gifts, discounts, delivery, and uplift ratio as “profitable”.
As the calculations are only a forecast, our work doesn’t stop here. After implementation, we need to watch how the loyalty program works for the client and to what extent it correlates with the forecast and make changes, if needed.
Result
$1,372,330
such an uplift the calculated loyalty program will bring in the pessimistic scenario
400 Excel tables
were used for the calculation