No business is happy losing a customer — we’d all prefer to resolve whatever prompted them to leave and try to rebuild the relationship. But in many cases, that’s not an option. By the time CX teams realize churn has happened, the customer relationship is beyond saving — or they’ve ceased contact altogether.
Sometimes a customer will get in touch to discuss their concerns, and the retention team can hopefully work their magic. But when that doesn’t happen — which is most of the time — how else can CX teams predict which customers are at risk of churn?
The key is understanding that churn isn’t usually an event — it’s a process. Frustrations or issues build up over time, and eventually, that results in a lost customer.
The goal for CX teams is to recognize the early warning signs — the hidden churn indicators — so they can respond proactively when that process starts, not wait to react when it ends.
In this article, we’ll look at the challenges involved in achieving that goal — and offer practical advice on how to overcome them.
01 — The role of data in detecting unknown churn signals
Hiding in plain sight — the churn indicators all around you.
Customer interactions generate enormous volumes of data. Support tickets, web chats, emails, social media posts, feedback surveys, online reviews, voice calls — and the list goes on.
Hidden within that mountain of data are subtle indicators that some customers might not be satisfied with your service — but finding them is another story.
Sometimes the signs are obvious — an angry WhatsApp message, a complaint submitted to the main company email address, a tweet calling out your service levels. Those are usually dealt with reactively. But if you want to be proactive, you’re going to have to search beyond the loudest voices and find subtler changes in customer behavior that indicate a problem is developing.
When manual processes and traditional analytics fall short
The main problem CX teams face is the “needle in a haystack” issue. Sure, you know that the churn indicators you’re looking for are hidden among the vast amounts of customer data, but if the only way to identify them is to pick through the data piece by piece, you need a near-infinite amount of time or a near-infinite number of agents.
Even traditional analytics software can’t do much in the face of so much unstructured data. There’s no way for simple search tools to interpret the many different ways customers might begin to show they’re unhappy, especially when it’s split across multiple channels, stored in multiple formats, expressed in subtle terms, and has very little metadata to link it all together.
We’ll look at solutions to this resourcing issue in Section 3. First, let’s look at why churn is such a major issue — not just for CX teams, but company-wide.
02 — Churn impact across the organization
Why does churn matter so much right now?
In uncertain economic times, many companies are looking to save costs wherever possible. CX teams are at particular risk of cutbacks, as they are commonly viewed as a cost center, rather than a revenue generator. Reducing churn is one way that CX teams can demonstrate their value to the organization, by contributing directly to company revenues.
But whatever the economic climate is — it’s a basic fact that each day that passes without properly addressing the root causes of customer churn is a day that your company is losing money unnecessarily. Acting early on churn indicators in order to retain more customers creates immediate savings by offsetting acquisition costs. And implementing a continuous monitoring process to deal with new indicators as they arise helps prolong the customer lifecycle — which raises CLTV, increasing the total value of predictable recurring revenue.
All hands on deck — why reducing churn is a company-wide objective
As we said above, customer churn has a measurable impact on the bottom line — so it’s obvious that it affects the whole organization. But it also has specific effects on different departments within the business.
For the product team, churn might be an indicator of poor UX design, or lack of a key feature. For marketing, it might indicate that re-engagement campaigns aren’t performing as intended. For CX, it may be a sign that customer expectations aren’t being met, or response times are too slow.
It’s vital that CX teams take the lead on ensuring that churn metrics are reported across the organization and that strategies to reduce churn are agreed upon with all relevant departments
In Section 5 we’ll look at the specific actions that different departments can take to combat churn — but for them to do that effectively, they need to be alerted as early as possible. So how can the CX team get ahead of the issue?
03 — Leveraging AI and Natural Language Processing (NLP) to reveal churn indicators hidden in your data.
We’ve already talked about the “needle in the haystack” issue of finding hidden churn indicators among the vast amounts of customer data that CX teams have to deal with. And we know picking those “needles” out by hand is close to impossible. So what we need is … a magnet?
Separating the signal from the noise — with a little help from tech
AI and Natural Language Processing technologies can effectively bridge the resource gap when it comes to interpreting and categorizing unstructured data, and translating it into a suitable format for further analysis.
Auto-tagging customer communications based on content, sentiment, and intent allows CX teams to isolate and identify specific communications, link similar communications from the same customer across different channels and relate these to other factors such as customer lifecycle stage or demographic data.
Identifying the advance indicators which predict churn
The specific insights uncovered by an AI or NLP solution can then be reviewed at a higher level to spot trends or patterns and identify the initial indicators which preceded the change in behavior or sentiment.
For example, you might reveal that following a subscription price increase, the incidence of negative mentions on social media increased by a certain percentage, and over the following month this could be linked directly to a proportional drop-off in subscription renewals — showing the direct effect of price increases on customer retention.
Or an increase in your email newsletter unsubscribe rate among a specific demographic may have been followed by reduced activity on a proportion of those user accounts — establishing that disengaging from marketing communications is an indicator of future disengagement with your service.
When you’ve identified these indicators, you should report on them cross-departmentally in the same way we discussed for the standard churn metrics in Section 2 — allowing other teams to get involved early and avoid customer loss.
We’ll explain how that works in Section 5, but first, let’s look at some real-life examples of the type of indicators that can predict churn ahead of time.
04 — Case studies: Real-life examples of hidden churn signals prompting action.
Example A:
A fintech company starts receiving more customer requests for account statements. Nobody seems unhappy or complains, but the volume of this type of request is higher than usual.
Why is this a hidden churn indicator?
Sudden statement requests can indicate that customers want to review their position ahead of closing their account or migrating it to another provider.
While they might not seem particularly unhappy, if they are left to their own devices at this stage they could disappear permanently.
Example B:
A SaaS company notices an increase in billing and PO-related issues. More customers than usual are getting in touch to query line items on their invoices, or questioning whether they approved additional costs.
Why is this a hidden churn indicator?
Increased queries about billing specifics can indicate that customers are either struggling to pay or planning to consolidate their spending by cutting back on additional features.
Example C:
A food delivery company received a higher rate of gift card orders just after the holidays — most of them from new customers who had never registered with the website before.
Why is this a hidden churn indicator?
A rush of first-time customers using gift cards suggests they’re spending free cash they received as a present, but they’re not that likely to stick around.
In this case, it’s not so much a matter of customers churning because they’re unhappy — but more that they might “bounce” before the company can convert them to repeat customers.
05 — Acting on churn indicators across your business
Now that you understand the key indicators that predict customer churn, it’s the responsibility of the CX team to liaise with other departments in order to take a unified approach to deal with the issue. Here are some examples of specific proactive actions different teams can take to boost retention.
1-to-1 engagements — CX team
CX teams should implement a process to respond to specific churn risk signals with proactive customer outreach. For example, once a customer account has been inactive for 28 days, a CX agent is automatically prompted to carry out a customer care call, checking in to establish if there’s any reason why the customer hasn’t logged in for a month.
1-to-many messaging — Marketing team
The CX team can also liaise with the marketing department to respond to less critical churn indicators. For example, if customer spending has dropped off in a specific region over the past week, the marketing team can execute a re-engagement campaign targeting all customers in the area with a time-limited offer or discount.
Campaign results should be carefully analyzed to assess the success of the campaign, but also to cross-reference with CX data to establish which customer didn’t respond and may need further engagement.
Addressing root causes — Product team
When indicators related to software updates, usability problems or other product-specific issues are triggered, CX leaders should collaborate with the product team to proactively address root causes by prioritizing fixes or developments that will have the greatest impact on retention.
06 — Speed and proactivity matter.
The final point on this topic — what we called the “golden rule” at the top of the piece — is that speed and proactivity are the two most important factors when dealing with churn.
Once you’ve put solutions in place to identify the early warning signals that you’ve linked to potential churn, you can’t sit back and let the AI handle everything for you.
Customer expectations are always evolving. Your processes, your product, and your service levels are always evolving. So your churn strategy has to evolve in parallel — or you’re going to miss new indicators as they emerge.
You need to proactively monitor your data to identify new signals, and once identified, develop processes and strategies to deal with them — as fast as possible.
If you’re a CX operator interested in implementing an AI or NLP solution to help you identify churn indicators early, and you’d like to understand more about the capabilities and limitations of the technology, feel free to book in a slot to speak to one of our CX experts. No sales pressure, just friendly and practical advice.
