In this article, we discuss the customer retention challenges facing D2C companies as they scale, how to get to the root cause of customer churn so you can fight it, and why customer support tickets are a key solution to tackling churn. A granular, data-driven tagging taxonomy is vital here, so we address that, too.
Meal kits are having a moment, or more like a few years, in the spotlight.
With Nestle acquiring two meal kit companies in two weeks (Freshly and Mindful Chef), and Gousto reaching a $1 billion valuation, global lockdowns had thrown this sector into our homes. Even independent restaurants are adding to the competition, with Dishoom and Padella now offering their unique dishes for anyone to cook.
With this rise comes the inevitable pain of scaling fast to accommodate for demand, while preserving and improving customer retention levels. Even with growth from new customers, meal kit businesses are going to need solid retention numbers to stay viable. But there is often a missing link when trying to increase customer retention; understanding exactly why the churn is happening.
This is where customer support can play a crucial role. Support teams go through 100s and 1000s of tickets each day; they know exactly why customers are churning. However, this intrinsic knowledge of the customer often doesn’t translate well in support ticket reporting. This is mostly due to manual, generic tags that are difficult to action.
When support teams have a sophisticated taxonomy, backed up by AI (automated, granular tagging of conversations), the whole business can start to work on the specific issues faced by meal kit businesses that come in every day.
Even with this recent boost to sales and new customers, lockdown conditions will not always be here (although it can definitely feel like that!). With high customer acquisition costs and enticing introductory offers, customer retention has always been the most important metric for meal kit brands.
There's a lot of money to recoup in recurring subscriptions before you can just break even and the analysis on churn (shown in the below image) for this sector does not make pretty reading.
This means providing a seamless and delightful experience is essential to a meal kit business thriving but it is obviously difficult to obtain.
Imagine you have just finished work. It's a Monday and you're just happy it's the evening (congrats). You've paid for the convenience of having your ingredients delivered and are looking forward to cooking dinner with your newly arrived meal kit... but you open the box to find the spice mix is missing, or the chicken is a few days past its best, or maybe even the box didn't arrive at all. You were expecting a heartwarming meal and now you need to figure something else out quickly, maybe go to the shop and buy replacements or a whole different meal.
Enter the hangry customer. This is not the same as when a pair of shoes you ordered are a couple of days late; this is a one-time experience that has been damaged and is very difficult to make right. Who do we think will churn faster? A hangry customer or just an angry one?
Standard delivery of a meal kit includes several ingredients per recipe and numerous recipes sent in one box, each week, oh and those ingredients can be highly perishable and must be packaged so that they don't become damaged or go off. And this is just for one customer's order!
Meal kit businesses have a logistics nightmare to solve and there are several points along the customer journey that can lead to customer churn. It's easy to see why top complaints on Trustpilot for businesses like Hello Fresh are missing ingredients.
So, we've seen that meal kit businesses are facing a challenging trifecta: their operations are difficult to get right, when they go wrong the reaction is an emotional one, and to make things worse, customer retention could not be more important.
What's the answer to this perfect storm they find themselves in?
The first step to fighting churn is to make sure you have a clear understanding about what’s driving it.
Understanding churn means understanding the experience customers are having with your brand, their expectations and main pain points along the journey. This is usually done through decoding answers given on NPS or post-churn surveys where the sample size can be small, and based on what is top of mind for the customer at that exact moment.
"The people making the decisions need to understand that customer service is the only pure route to what's actually going on with your customer."—Alice Godfrey, On Hold Podcast
However, it's more effective to analyse why current customers are raising support tickets. Customer service tickets are the purest form of customer insights; they contain a wealth of information about the problems customers face at each touchpoint in their journey with you. This is where a taxonomy (the way you tag, categorise and organise your support tickets) can help to report on what is driving customer retention.
We continuously see that there are challenges with manual approach to support ticket tagging that most teams opt for, and that these challenges stop organisations from acting against customer churn.
Actionable customer insights are needed or your CX improvements are likely to be slow or killed off by uncertainty. But, there are challenges to this.
Due to generic and sometimes inaccurate tagging and reporting, confidence across teams when looking at support-driven customer insights is often low. Action is especially limited in a scaling support team, where sharing insights becomes ad-hoc and difficult.
Typically, companies manually assign ‘reason for contact’ tags to support tickets. But, they’re often too generic to use effectively in root cause analysis.
Generic tags usually come from the fact that it's unreasonable to expect agents to pick the most appropriate tag from a list of 200 options. Instead, support teams are compromising the quality of insight to fit in with manual processes. For teams that really want to understand the issues so they can get to a fix, reading through conversations ends up being the solution, which is always extensively time-consuming.
One common problem we have found from our analysis of meal kit reviews is the issue of missing ingredients. Knowing that missing ingredients is a big problem and that it is rising in volume, is important to know. But it’s impractical and can often lead to teams having more questions. Such as, "which ingredients are missing?", "which recipes are the biggest culprits?"
A few things impact accuracy. First, the subjective nature of manual tags can mean agents interpret them differently and assign tags differently (this can be somewhat overcome by a time-consuming QA process).
Agents usually only have the option of assigning one tag to a conversation—leaving a lot of issues under-reported. For example, the review below shows us there are often multiple problems affecting one order. If this person logged a ticket, there would likely only be coverage of one, making your reporting inaccurate.
If you’re not sure all issues have been covered, and you are left with more questions such as, "but which ingredients are causing this?", then discussing and prioritising fixes becomes difficult. It's much easier to push that issue aside until "we have more data".
When support teams scale to meet a growing customer base, sharing issues with the relevant team often slows down and flagging with the relevant team can slow down.
While before communication could be as easy as a daily stand up, now, with much bigger teams, there is a need for faster time to insight and a more streamlined issue sharing process.
However, generic ‘reason for contact’ tags make this difficult. For example, a spike in mentions for a certain recipe will probably not be covered by support analytics and your chance to resolve the root cause of the problem is missed.
To increase customer retention, we need to know the specific drivers of customer churn so that we have the confidence to act upon them. Support teams need a data-driven tag taxonomy (a standardised way of naming and classifying) to apply to their customer conversations if they're going to achieve this.
As we have seen above, doing this manually with a limited number of topics leads to generalising issues and inaction. Instead, support teams need a taxonomy fit for purpose; one with specific tags, that allows you to drill down into customer pain points and is easy to understand by the rest of the business.
This level taxonomy quality, if done manually, would be a huge effort, and would require training and ongoing QA processes to ensure it was used successfully. This is where a custom AI system is of value. When specifically built to detect issues with meal kit businesses, your AI model can understand every support conversation and automatically tag them with granular and multiple tags that cover all issues with great accuracy.
Examples of insights possible thanks to a data-driven taxonomy include:
An AI meal kit taxonomy can be applied to any channel, such as Zendesk or NPS and spans the entire customer journey (from discovery and first order to delivery and post-experience) so that support teams can quickly analyse and flag issues with relevant teams.
With easy to understand tags, anyone in the business can dig into an issue to find out why it's happening. If we take an increase in missing ingredients as an example, you can see which exact ingredients have been mentioned more in the last week, and those which have been mentioned less. Maybe there has been a 20% increase in people mentioning old ingredients, with a 50% increase in mentions for old basil, this could be the culprit of this particular rise in support tickets. Spotting trends in the data is made really simple by this type of taxonomy, and it takes out the need to manually read through conversations and do some back of the envelope maths to figure out which issues are bigger.
When you’re confident all issues are being exposed, you can start to prioritise issues and push for change so teams see a reduction in ticket volume and churn. One simple way to prioritise fixes is by just seeing which are the biggest contributors to ticket volume. This is a great way to start, especially because support teams can easily tie the cost of the issues to ticket resolution time. However, it doesn't account for how much an issue actually impacts a customer.
Imagine for a moment that the tag "recipe selection not working" resulted in 200 tickets in the past week, while "late delivery" resulted in 150 tickets. We might say that recipe selection was the biggest issue if we just look at volume. However, when we take into account NPS analysis with a granular taxonomy, we can do a driver analysis to understand which issues are most important to customers (and lower ratings the most). From this analysis, we could see that late delivery results in much lower ratings and negative experiences for customers than recipe selection issues. Taking into account which issues most impact customer retention, it may turn out best to prioritise solving the causes of late deliveries.
With automatic tags that fit meal kit data well, there is little need for manual tags. Teams across the company can even start to get automated daily email updates on support ticket insights e.g. the biggest issues and biggest changes in the data.
Automatic updates can help surface when a topic like "damaged packaging" has become one of the top contributors to support tickets in the last couple of days, or that the recent issue of "discount codes not working" has been resolved leading to a decrease in mentions. With this level of granular tagging, support teams can even start to look for anomalies in the data and set up alerts for when there are unexpected changes in the data— flagging serious issues in a timely way.
Often we see there are a lot of insights in support data, but these often stay siloed in support due to poor tags and manual reporting. With tags that know meal kit issues at a granular level, support leaders can now confidently pass on issues to the teams that can, in turn, confidently act on them.
When tags are made easy to understand, different teams such as product or operations, can dive into the insights themselves and find what's relevant to them quickly and easily. So operations can get updates on the performance of wide-ranging delivery issues and product can look into that specific issue with subscriptions. Teams who can readily access this data then feel empowered to take action themselves and reduce ticket volumes while also increasing customer retention.
At SentiSum, we build each customer a customised AI-driven taxonomy so they can unlock support ticket analytics and start tackling the drivers of churn. Book a demo with us to learn how we collect, analyse and share actionable insight across your company.