One of the things that keep startup entrepreneurs like me, awake at night is the ultimate question – “Will my customer come back and do another transaction with me?” Every month at Stylofie (online marketplace for beauty and wellness services), we put together a monthly sales forecast – breaking the nos. into new customer sales as well as repeat sales.
Both metrics are closely monitored and tracked, month on month. Any business, whether it’s online or offline needs a base of loyal customers to become sustainable. It is also easier and less expensive to up-sell/cross-sell to existing customers – that’s Marketing 101.
Repeat purchases have always intrigued me. Today, customers have a huge number of choices online across categories, whether it’s ordering grocery, buying a phone or cleaning their laundry. The ubiquitous smartphone has created many a new habit-forming behaviour for the urban consumer and unless the consumer finds value & a frictionless experience, he/she will seldom come back. Whether its ecommerce, mcommerce, or brick-commerce, great experiences drive the customer to come back to the purchase platform.
At Stylofie, we observed a recent trend in our data – our repeat orders were rocketing up but our customer churn rate started to go up as well. This got us worried & thinking. What has changed that we didn’t see it coming?
Why are customers not coming back after their first transaction with us? At the same time, we also saw some customers doing as many as 3-4 transactions a month with us. Given our category i.e. beauty services, it was high (how many times can you cut your hair or get a pedicure in a month).
We worked the phones, analysed the transaction data and looked at monitoring tools. We wanted to deeply understand this entire phenomenon driving repeat transactions.
Why Repeat Customers Mean So Much For Business
|One time Customer||Repeat Customer|
|Average order value||702||787|
|Margin per transaction||2.2%||14.9%|
The initial cost of customer acquisition (CAC) for new/repeat customer is the same – you recover the cost through retention & repeat transactions. As we peeled the onion on the average order value for one time customer vis-a-vis a repeat customer, we found that not only is the order value higher by INR 85 (~12%) , there is a phenomenal difference in the average gross margins (2.2% vs 14.9%) per transaction.
Our repeat customers not only contribute to the top line but also significantly more to the bottom line. Every customer who does only a single transaction with us actually drags down our bottom line. Also a repeat customer is more likely to recommend Stylofie to a new customer, thus reducing our CAC as well.
So, for us, the problem statement was really clear – What can we do to convert our one-time customers into repeat ones? However, before we try to even solve for this, it was important to understand – what exactly makes a customer on Stylofie – a repeat customer? What is that magic sauce?
What Turns A One-Time Customer Into A Repeat Customer
As I was pondering over this one evening, I got a call from an old friend of mine at Genpact (my ex-employer). He mentioned that he was launching his analytics/big data startup and asked me if he could help me with some of my startup business problems. It was like a hungry person getting invited to a 5 star buffet and I immediately discussed the problem at hand – Can you help me predict if a customer will do a repeat transaction on Stylofie, on the basis of their first purchase with us?
My friends from Catalytics.in started working on the problem statement & data with great earnestness. Through several calls late in the night, the team asked several questions about the business and about the patterns that they saw in the data. They finally built a statistical model that we could use to predict if a consumer would come back and do a repeat purchase on our platform.
The Catalytics team applied a segment and predict approach, where they first segmented the Stylofie clientele based on their characteristics and then applied Predictive Modelling to identify who would be a repeat customer.
The model uses various characteristics of the customer at their first visit like –
- If the customer is referred by a friend.
- The booking platform by the customer at first booking – iOS/ Android/website.
- Whether the customer used Promo Code at the first booking.
- The choice of salons at the first booking, like Salon 1 vs. Salon 2 etc.
- Feedback rating post-service, etc.
The entire data discovery and predicting modelling procedure was carried out using R, a powerful open-source statistical software.
As D-day arrived, I got the much-awaited predictive model, in an excel file; I had an early morning call with the Catalytics team who explained the finer nuances of the model and its workings. I was super excited to see the model – it was almost like test driving a new car – filled with thrill and caution.
Predictive Model Assistance
It was now time to put the rubber to the road and see how the model behaves. I first ran past a few data points of customers who have never come back, to see if the model predicts them accurately. Then it was time to work on the next set of data – repeat customers who have made several bookings with us; the Excel-based model gave a probability score and the results stunned me. The model was near perfect in predicting the repeat usage based on the first transaction details. It was magical!
I was truly zapped, to see the power of data at work. Here’s an excel file based formula, that was predicting the future of our business at the most fundamental unit of measurement – for each customer order.
The model had demonstrated an amazing accuracy in predicting the probability score of a repeat purchase and as we begin to deploy the model in our operations, more ideas and suggestions have started pouring in to optimise it and drive decisions based on its output.
We are also exploring with the Catalytics team on extending the model to predict when will the customer return and how often.
Now I can partly sleep easy – with the confidence that I can predict whether a first time customer will come back for a repeat transaction on our platform. For a startup entrepreneur, that’s manna from heaven! Of course, sleep doesn’t come very easy for entrepreneur these days, because while we have a model for predicting customer behaviour, we are yet to figure out a model for predicting VC investment behaviour! Perhaps, time for another call with the Catalytics team!