Zendesk just launched their Trends Report for 2024 giving insight into where Customer Experience and support is moving towards this year.
Last year's report had 5 big trends:
- AI experiences are becoming more evolved and seamless
- Conversational experiences are empowering consumers
- Customers are eager for deeper personalisation
- Consumer well-being and sentiment are reshaping CX
- CX teams are breaking down silos as they become more integrated
Looking back at those trends today, you can clearly see that Zendesk made big shifts and improvements in their platform to accommodate these trends. The launch of Zendesk AI, major expansions in their Messaging platform and bots, and improvements in Agent Workspace with regards to Teams integrations, Slack bots, and custom objects, made sure that Zendesk could offer a technical solution for each of the trends.
This year's report doubles the amount of trends with a whopping ten insights that will allow you to Unlock the power of intelligent CX.
The trends are split across three Innovation Areas: AI & intelligent Experiences, Data & Trustworthy Experiences, and Next Gen & Immersive Experiences.
Conveniently, these three areas map nicely into two new add-ons Zendesk launched in 2023: Zendesk Advanced AI and Advanced Customer Data and Privacy Protection, with the third area partially covered with the acquisition of Tymeshift last year.
Let's dive in.
AI & Intelligent Experiences
Trend 1: Generative AI will accelerate the delivery of a more humanised journey that feels personable & interactive.
But can generative AI help businesses provide the kind of warm, human service that feels like interactions with local, family-owned businesses? More than two-thirds of CX organisations think that it will help their business provide that warmth and familiarity, even if they serve millions of customers.
One of the main benefits I find in these new Generative AI-driven bots like those from Zendesk or Ultimate is the fact that they work completely differently than the decision tree-based bots from before.
It used to be that a bot was the chat equivalent of a phone IVR. Do you want sales or support? Support. What kind of question do you have? "something broke", "product support", "upgrade questions" and so on, until you hopefully either make the right decisions and get an answer, get completely frustrated and restart the conversation until you find the right incantation, or start typing human, human, human, until you reach an agent.
Now with Generative AI bots instead of me searching for an answer, I can just ask a question "How do I update my iPhone?" and the bot will reply with the answer without the need of navigating a decision tree of predefined options.
That combined with a specific tone of voice, context from my previous questions and, hopefully, the data the company has on me, will result in better answers and quicker resolutions.
Is Zendesk there yet to fulfil the needs of this trend? Partially. Asking a question and getting data from your Help Center works like a charm, but there's no actual conversation happening here, and the data available to the bot is limited to the Help Center, the Zendesk AI bot cannot execute API calls or collect information from external or internal resources to give a richer answer with more context.
Trend 2: Chatbots are rapidly transforming into digital agents that have the capacity to do more.
Chatbots are doing a better job at surfacing information quickly, and they're also improving their ability to tailor responses to better fit where customers are in their journey.
A good customer care approach has always and will always revolve around self-service and knowledge management. Even before chatbots, having a knowledge base with your top 10 support inquiries and a system in place that can automatically reply with those topics will dramatically reduce the workload of your customer care team.
Chatbots and AI offer a more powerful version of this since it improves on two aspects:
- It can turn a customer inquiry into an intent and have a better grasp of what the customer needs instead of doing a basic keyword search in the knowledge base.
- It can turn the available knowledge sources into a custom answer tailored to the needs of the customer, instead of replying with a link to the article or the exact article contents.
It comes as no surprise that these two improvements are what make agents better at handling support than regular search, and now that bots have these same capabilities, even more customer questions can be handled by the bot without any human interactions.
What's missing is the more complex workflows. Autonomously handling refunds, booking changes, reservations... Some of these are already possible by using the API endpoints for the Zendesk Bot, and I'm sure we'll be seeing more of these items become natively available in the Zendesk Bot once elements like Conversational Commerce become more widely available within Zendesk.
Trend 3: Disconnect grows between CX leaders and agents on everything related to AI; strategy, tools, and role impact.
The problem is that many agents aren't so sure. They look at the new tools with a mix of wariness and deflation, fearing what AI will mean for their job security.
The big promise of AI is that it can automate a lot more processes and lower the workload for agents. It generates insights by tagging each conversation and automates offering answers to customer inquiries.
This gain in efficiency can have two results which are on the opposite side of the spectrum. On one hand, you can start from the status quo. You're kind of happy with your CSAT and First Reply Time metrics and see the reduction in tickets as a way to reduce headcount. Less work means fewer people needed to do that work while not really impacting your customer experience in a negative way.
On the flip side is the idea that you can use that extra time to improve your customer support. You can give agents actual time to dive into the complex issues or you can use that time to look into processes and documentation. The former allows you to have deeper, more personal conversations with your customers and really solve their issue, the latter improves self-service and offers more knowledge to your bot to deflect even more tickets.
However you look at it, if you use the impact AI can have on your support interactions as a trigger to improve, you create room for agents to grow and become experts in handling cases or documentation.
So it's important that, as a company, you lay out your vision and choose to go for quality or quantity and be transparent on why you implement AI in your company and what you will do with the efficiency gain. You can choose the short-term solution of reducing headcount and freeing up costs. Or you can go for the long-term approach of creating a cycle of continuously improving your customer experience by shifting agents towards more complex roles.
Trend 4: AI transparency and decision-making are now the rule, not the exception
The need for trust through transparency becomes especially important when dealing with sensitive information that reveals a person's identity, their health history, and their financial status. No customer wants to feel like AI is snooping on them or making their data available to bad actors. As 58 percent of consumers told Zendesk, knowing how their data is collected, stored, and used plays an outsized role in whether they'll purchase a product or service from a company.
The answer to the question "who owns or runs your AI model" will become a hot topic this year.
Where 2023 was the year of "everyone uses OpenAI" as a way to hack AI into your company, I think 2024 will become the year where the question "who owns our model" is the big question to ask when enabling AI in your company. You want to be sure answers are based on actual facts, and you want to be sure that it's only you who can use your own data, and not a competitor.
Zendesk has its own AI models trained on Zendesk Data, and made available to their customers. The answers generated to reply to tickets are based solely on the data in your Help Center or your own ticket data in the case of suggested macros. It only uses OpenAI to generate a reply, but the data used to generate the reply is 100% controlled by Zendesk, and in the case of intent or sentiment mapping, OpenAI is not even involved.
As a company, this Mark Zuckerberg vision is a sure thing within Zendesk. Your customers can only get answers based on data you made available to Zendesk.
Yeah, so our view is that there’s actually going to be a lot of these that people talk to you for different things. [...] let’s say you’re a small business and you want to have an AI that can help you interface with customers to do sales and support. You want to be pretty confident that
your AI isn’t going to be promoting your competitor’s products, right?"
As a customer, the question about trust is twofold. On one side it's the question "can I trust the answer I get from the bot", and on the other side it's the question "can I trust this bot or company to handle my data correctly".
One is about making sure that the answer "can I melt an egg" is replied to correctly. The other is about "if I enter my birthday here, how securely do they store it". I don't think we'll get to a point where customers will see the Zendesk Logo on a chatbot as a way to know "hey this is safe". But I do think that a quality bot with good answers will automatically infer a feeling of trust that will give customers assurance that they can trust the bot with their personal data.
Data & Trustworthy Experiences
Trend 5: Businesses are heavily focused on being able to instantly modify user experiences, putting increased pressure on leveraging data in real time.
That preference means companies must focus on boosting their bot capabilities via Al, specifically using its power to capture and analyse sentiment and intent. Doing so will help businesses predict customer needs and resolve issues quickly and efficiently (including knowing when a bot needs to hand over an interaction to a human agent).
For me one of the main benefits of bots powered by Generative AI is the aforementioned idea that they allow customers to ask questions instead of customers navigating a flow to find an answer.
- A generated response based on knowledge base data (or other sources)
- An API powered flow that pulls in context (e.g. order status) to reply to the customer
- An escalation to the right agent based on intent, sentiment and conversation content
Most customer interactions possible today (In Zendesk) are very closely linked to indexed knowledge base content as a basis for an automated self service approach. But recent updates in e.g the Zendesk Bot for more dynamic conversation experiences make the bot capable of pulling in external data to offer more complex answer flows.
I really hope Zendesk will keep improving that part of their bot in the next year, making it easier to pull in data from CRM, webshop or other systems and combine that data with Generative AI to have more complex and data-driven conversations with customers.
So now that the Knowledge Based bots are a solved matter from a technical standpoint, the focus this year will shift towards conversations powered by other types of data, further reducing the need for agents to do these repetitive lookups, and once again giving them more time to dive into the complex and unique questions.
Trend 6: CX leaders are the new drivers of data privacy as AI & personalisation take on a greater role.
They know that it's not good enough to simply have Al tools for personalisation; those solutions must keep customer data secure.
Looking back at Relate 2023 it's almost surprising this trend wasn't Trend 1 in this report. At the keynote they showed off this awesome slide which kinda sums up the entire trend.
There's a couple of benefits of powering your customer interactions with AI. One benefit is that you can better capture intent, which also means that instead of asking a lot of data in a form, you can ask for only the data needed to solve the inquiry, since you can closely match question and intent.
Within Zendesk we've now got automatic ticket deletion, build-in redaction tools in the Agent Workspace and more secure integrations to make sure data is shared correctly across platforms.
Similarly, if we can have API integrations in the bot that pull in the data directly from the source, we need to pass less data to agents, or store less data in tickets, since we can resolve inquiries before they even reach the ticket stage.
It's all small changes, but they can amount to a big reduction in customer data stored, while not loosing the ability to correctly and swiftly resolve their inquiries.
Trend 7: Security is no longer an add-on but is seamlessly incorporated throughout the customer journey.
Thankfully, CX leaders have options for seamlessly integrating security measures into customer experiences, most of which aren't new: multi-factor authentication, encryption of service interactions, and being transparent with customers about security and data privacy practices.
Security and Privacy go hand in hand, and this is a trend one where Zendesk is has made big efforts in the last 12 months, starting with a renewed focus on Privacy and Security at Relate 2023 and the release of more complex role and permission settings, messaging and end-user authentication, an entire new security add-on and more.
If you're interesting in securing your Zendesk instance, take a look at my Security checklist published earlier this month.
Next Gen & Immersive Experiences
Trend 8: Live and immersive experiences are now heavily influencing the future of online shopping.
Now, 80 percent of consumers expect chat agents and support representatives to assist them with everything they need. The line between support and sales has begun to blur.
Remember Zendesk Labs announced at Relate 2023? We've still not seen any actual feature releases or EAPs for their idea of conversation commerce, even though this concept is closely related to this eight trend.
As mentioned in one of the earlier trends, shifting conversations towards bots gives agents the room to dive into the unique issues. Similarly, the more complex scenarios bots can handle, the more is expected of the bot by consumers.
In a classic scenario I might get an article about the return policy, or in a AI powered scenario I might get a custom reply based on my question.
For example, if I booked a flight but want to move my return journey one day forward. My first step would be to open the airline's app and view my booking and try to change the booking there. If that doesn't work, I'd either open the in-app chat and ask for "how to change my return flight," or even better, I could get a proactive bot message asking me about my booking changes.
But even if the answer I get is "yes, this is possible" or "no, it's not possible, you need to book another flight," to actually resolve my question, the chatbot or agent would need context (who am I, what flight, which ticket type), would need power (update, cancel, reschedule bookings), and would need to do it right then and there in the chat.
Is this a sales scenario? Yes, since I need to potentially buy a new ticket or pay a surplus for the rescheduled journey. Or is this a support scenario? Yes, because I don't know if it's possible and how to reschedule my ticket.
Trend 9: Voice is carving out a more advanced role focused on handling complex and escalated issues.
When companies create a seamless transition from digital to voice channels for handling complex issues, consumer confidence in the former rises. That in turn becomes a virtuous cycle in which customers feel increasingly satisfied with reaching out via digital channels, thus lessening demand for phone options.
People who know me, know I'm not the biggest fan of voice as a customer care channel. Voice requires both the customer and the agent on the other side to make time now to handle and resolve an issue, whereas more often than not an asynchronous approach of asking a question and getting an informed response or solution offered an hour later is often way more rewarding.
This idea, sadly, starts from the negative idea that most customer care teams are not able to resolve issues now and then, and that the person picking up the phone is often not enabled to make the decisions or changes needed to resolve an issue.
However, like the trend report mentions, if a lot of customer interactions shift towards the automated chat channels I prefer, this actually frees up time for agents to handle only the complex issues that require an actual human interaction. And, just like with Siri or Alexa, asking via voice to add "coffee to the shopping list' is way faster than typing it out, and a phone call or conversation does allow for a more nuanced discussion often needed to handle these complex issues.
So if companies follow the idea I mentioned earlier of using the efficiency gain made possible via AI to give agents the room to dive into the complex issues, then voice might become the logical channel to have those longer and deeper conversations that actually solve my issue.an easy way to turn a chat conversation into a call via the web widget, and the new Generative AI for Voice turns long conversations into short summaries, allowing for easier escalation to the right person to handle the ticket.
Trend 10: Predictive agent management tools are finally eclipsing traditional methods.
However, leaders now have agent management tools at their disposal that can take a lot of the guesswork out of running a support operation. These tools can offer both operational and strategic foresight, helping managers make better staffing and training choices.
It's no coincidence that, in the last twelve months, Zendesk has bought both Tymeshift, an agent scheduling and forecast tool, and Klaus, an AI-powered Quality Management platform. They both fill gaps in the platform that enable bigger customer care teams to get insight in what agents are doing, and make sure the right type and amount of agents are available to handle customer inquiries.
Both purchases are fairly recent (or in the case of Klaus not even fully completed) so there's not much to say about how the tools currently integrate deeply into the Zendesk platform, but given the wealth of historical data available in your own Zendesk, the insights they can pull based on all the thousands of other Zendesk customers, it's only logical that they leverage that data to give insight into agent availability natively, instead of handing that data off to an external party.
And if AI Bots shift agents towards more complex interactions, you want to make sure that, if a customer needs a human, the right person is available to chat, call, or reply to you.
I have to admit, this was a tough article to write. Where my overview of the Customer Service Trends by Ultimate last year went fairly easy due to the trends being practical and more technically focused, Zendesk's trends are a bit more fluffy and less grounded in technical reality.
That being said, they do force me to take a bird's eye view of the Zendesk Platform and how it plays in the broader ecosystem of customer care and companies, instead of writing weekly about the nitty-gritty feature releases launched by them.
If I have to pick one favourite Trend, it's gotta be Trend 4 (AI transparency and decisioning are now the rule, not the exception).
I'm a big proponent of "control your own data". It's the reason I write on my own website and not just post on LinkedIn. It's the reason why I use Apple products, since they have a very strict stance on privacy, and it's the reason why I like to work with Zendesk or Ultimate.
They both allow companies to build solutions based on their own data and tweak it to their own needs, while integrating with basically everything. But regardless if a customer contacts you via Facebook or email, searches your FAQ, or chats with your bot, you have to be sure you give answers based on the data you provide, and that you can report on all the interactions you have. And your customers have to be sure that the answers they get can be trusted.