The News

Intercom is a conversational relationship platform that helps businesses build better customer relationships through personalized, messenger based experiences. It’s the only platform that delivers conversational experiences across the customer journey, with solutions for conversational marketing, conversational customer engagement and conversational support. Intercom is bringing a messenger-first experience to all business-to-customer communication, powering 500 million conversations per month and connecting 4 billion unique end users worldwide across its more than 30,000 paying customers, including Atlassian, Sotheby’s and New Relic. On August 12, 2020 Intercom launched a new product release with more than twenty functional enhancement that are intended to ‘supercharge customer support’. The company says that it is its biggest ever launch and targeted at ensuring that its customers “can provide prompt, personal support without sacrificing power or efficiency”. The twenty plus new features in this release cover enhancements across Messenger, Inbox and reporting. As part of this, there are new bots that can be used in the messenger, new data attributes that conversations can be tagged with and new rules and assignment features that enable deeper automation of conversations. This gets augmented by additional view capabilities in the support inbox that help getting better overview and by offering more task bots to the messenger that help providing a high customer experience while automating jobs. Additionally, Intercom improved its reporting capabilities by adding three dashboards covering conversations, effectiveness, and team resolutions with twelve new metrics plus improved filtering capabilities. Lastly, Intercom changed the side-bar navigation by providing clearer icons, changing their sequence and renaming one of the core sections, Platform, to Contacts. This has the goal of improving usability. According to Jane Honey, Senior Product Director at Intercom, the company is especially strong at proactive and messenger based support, while the part of Intercom’s software that supported human customer service, was not so powerful. This has been addressed with the recent product launch with the goal of combining the best of both worlds, the Intercom strength of enabling great customer experience via conversational and bot based support with the strength on the back-end side of ticketing systems.

The Bigger Picture

In general, there are two varieties of customer service tools. Historically, there have been ticketing systems that support the ‘traditional’ support channels like phone or e-mail. These have evolved well over years and work well for companies that implemented them. They offer powerful routing mechanisms to find the right person to answer an inquiry and, of course, have strong reporting mechanisms as well, to track and improve the performance of a service centre. Then there are a newer breed of systems that base upon conversations and bots. Many of these started not as customer service systems but have their origins in marketing. Being newer, they often are not as sophisticated in routing incidents or in managing the performance of a service centre as the more traditional tools. On the other hand, a conversation is more natural to humans and, if combined with a bot, many simpler inquiries can be answered without being worked upon by a human service agent. Still, there can be any number of conversation requests at any given time. As a result of this, the former variety of support systems tend to be better for the service team while the latter tend to provide a better customer service experience for the customer while imposing additional stress on the service agents. Customers want reliable support, and they want it fast. They do not want to wait for prolonged times until their requests are answered. Service centres want efficiency and scalability, service agents want and need a good user experience, too. As a consequence of this, both types of help desk platforms are increasingly utilizing AI based mechanisms to identify the customers’ issues, so that they can be resolved via chatbots before the need for them to be addressed by a human arises. If they need to be escalated to a human then they need to be routed to the right agent, with the complete history of the current incident and enough of a history about earlier ones for the agent to hit the road running.

My Analysis and PoV

Intercom is essentially a messenger first platform in contrast to Zendesk and as such appears to be rather competing companies like helpshift or Kustomer. Still, according to G2, Intercom performs well in the upper right corner of the help desk software category. The company finds itself in a tight group with Zoho Desk, Freshdesk and, of course, Zendesk. As the helpdesk software vendors are embracing conversations, bots, and AI to help their customers to resolve incidents fast and efficiently, the competition has become fierce. Additionally, this category is a playground for lots of nimble and flexible specialists that can easily compete with powerhouses like Salesforce. Lastly, especially the bigger players are getting increasingly similar from a functional angle. Their capabilities can be offered via all relevant channels. The conversation is more and more becoming main the vehicle as opposed to tickets, and the internal efficiencies of the service centres are supported more and more – by all of them. The difference is that traditional ticket based vendors are moving from a ticket metaphor towards AI and conversations, while the newer breed of vendors that base customer service on conversations are moving towards internal efficiencies. With its “conversational support funnel”, Intercom has a clear strategy on how to enable customer service, starting from proactive support that offers immediate, AI assisted support using chatbots, via improved automation and self-service towards direct support via dedicated service staff. Coming out of the messenger area, Intercom has not been so strong in the latter areas, hence, according to Jane Honey, this was the focus of the current release, and the reason for Intercom pitching with the slogan of ‘taking on Zendesk’. What she also said is that businesses haven’t yet fully caught up to the shift towards a messenger first support model. This is an important release for Intercom, as efficiency and a great user experience are as important as customer experience, especially in times where service centres get more and more decentralized and support operations out of necessity get more online. Combining the conversational experience with the power of tickets and enabling service agents to be equipped with the customers conversation history and an AI that suggests good solutions to provide to the customer is the immediate way ahead. The ability to pro-actively provide support will help even more. The time of the search box that is not connected to a service bot is clearly over. Let’s see how these enhancements pan out for Intercom.

The News

On September 7, 2017 Helpshift announced the appointment of a new CEO in a blog post. Salesforce veteran Linda Crawford, with a tenure in the CRM arena that stretches back to 1996, took over the role, being the successor of co-founder Abinash Tripathy. Before, she held various positions at Salesforce, Rivermine, and Siebel. Most recently she held the position of a Chief Customer Officer at Optimizely. This track record certainly qualifies her to have a go at growing an interesting company to the next level. Helpshift itself is the company that created the mobile in app support market back in 2011 and, so far, has a keen focus on this area under the leadership of Abinash. Investors include Cisco, Intel, Microsoft, and Salesforce. Both, Microsoft and Salesforce, have been lead investors of the Series B financing round in June 2016.

The Bigger Picture

The customer service center market is extremely crowded and contested. Helpshift itself has a strong product and a good customer base, originating from the gaming industry but also running the customer service technology behind Microsoft Outlook Mobile. The company is targeting bigger accounts and is already amongst the ranks of Salesforce partners, having a deep integration. However, the overall market is turning into a platform play, which will be dominated by two or three business platforms for bigger companies, and maybe a handful more that cover SMBs. The platform companies are also providing strong business applications, including customer service, even in-app service. And then there are companies that build multi channel customer service solutions on these platforms, too. Are they as good as Helpshift? Likely not. But they cover more use cases. Helpshift, as a niche player, needs to find the right balance between the platforms and to extend the niche to make it more defensible.

MyPoV and Advice

A new CEO effectively could have come only from Microsoft or Salesforce. Both companies are heavily invested and could need an improved offering. With the very strong focus on mobile in app support Helpshift is in my eyes betting on the right horse. The company is well placed to play a major role in what I see as the upcoming battleground of customer service: Ambient computing. The challenge facing the company is in my eyes two-fold:

• There is only a minimal amount of AI and machine learning in the system. This prevents the system from scaling to the next level. Companies like AI have an advantage here, although they seem to have far less traction.
• The strong focus on mobile in-app support for time being serves as a disadvantage. Although more and more people are turning to their mobile devices and mobile apps the web is not dead – far from it. This means that many customers need two customer service solutions. And this makes companies like Salesforce, Zendesk, Freshworks, and others, a threat.

Especially the first one needs to get tackled along with lighthouse solutions in the world of ambient computing, starting e.g. by showcasing service that is driven by voice instead of typing. Cisco might have interesting use cases here. Secondly, Helpshift could visually ‘hide’ the FAQ behind a chat interface. That would make the service solution easily attractive for web sites as well, reducing the threat by other companies. With this the company would stay a ‘mobile native’ but safeguard an open flank. Thirdly, with the ties to Microsoft that are already there, it should be helpful to integrate into Microsoft Dynamics 365. This would increase their reach drastically. The world is moving more and more into a platform play. Helpshift, as I see it right now, will not supply the platform, so the company needs to play with and on the strong contenders’ platforms – at least two of them. Additionally, the company was not too strong on the marketing frontier. This improved since the beginning of the year and I’d expect this now getting a real push. Having said all that: Abinash and Helpshift have done a great job so far. This is definitely a company to look at when it comes to building a customer service element into an app. It will be very interesting to continue observing Linda, Abinash, and their team.

A few weeks ago I wrote an article about customer service in a world of ambient computing. This article looked at customer service from a customer’s point of view. In it I described how I see customer service getting humanized again by leveraging the advances in AI technologies like Natural Language Processing, speech-to-text- and text-to-speech generation along with intent determination. Leveraging these technologies customer service will turn into a conversation and it won’t matter anymore whether service is delivered by a bot or by a human. For the customer it will all appear to be the same. Instead of FAQs or web searches, bots will be the first line of support and escalate a problem to humans if they cannot solve it on their own. The obvious question is whether there will be an impact on the customer service center? And it probably does. Call centers, and with it the service agents as well as their managers, already now are under intense pressure to deliver, and to deliver more efficiently. With the increasing use of call deflection technologies like FAQs and communities there is a trend for the incidents facing the agents becoming more challenging. For example Helpshift states that already with its technology it is able to deflect about 90% of all incidents, which are solved via the native in-app FAQ that is delivered by the them. This statement basically says that the support staff is basically relieved of dealing with simple matters but has the chance to take up the more challenging ones. Still, in a world of ambient computing any given app can have hundreds of millions of users. Let’s say that any given day just one per cent of 100 million users have an issue. The well working FAQ deflects 99% of these. That leaves the service center with 100,000 calls. In one day. And they are the harder ones. Still, let’s be optimistic and say that an agent can solve 10 issues an hour, giving him 80 in an 8 hour shift. This would mean an overall call center size of 1,250 agents is needed to cope with this demand. Each of them under a tremendous stress level. With the systems behind bots becoming more and more intelligent the difficulty of raised issues will increase, even if FAQs and web searches are essentially hidden behind a bot interface that essentially makes the human agent the second point of contact again as opposed to the third, which likely means that the customer’s level of annoyance is slightly less elevated than in a third level scenario. At the same time it seems that call center agents are not prepared for handling this stress level. The employee turnover rate remains high and is probably even rising.

What Does This Mean For The Service Center?

Call centers are therefore facing a double challenge

1. Contain cost. This is achieved by more automation, which in turn puts more strain on the employees
2. Employ and retain a highly skilled set of service agents, which additionally have matching character traits, which drives cost. Skilled people tend to be more expensive than unskilled ones, and moving a call center into a low-salary country helps only so much – if at all. Training comes at an expense as well. This will be somewhat augmented by reduced hiring cost

The solution to it will be multi-faceted and increase a trend that is already visible. Implement intelligent systems that more than offset the higher salaries demanded – and deserved – by the fewer call center agents, through an increased solution rate and through them being of more help to the service agents. These systems will significantly rely on machine learning out of a variety of sources Employ communities by incentivizing to other users to help other users. These communities will be managed by community managers, with increasing support by AI-driven bots. Highly data driven prioritization and intelligent grouping and routing of incidents to the best matching agent, bot or human. This will involve sophisticated Natural Language Processing capabilities but will help in solving multiple calls regarding the same problem in one process Improved collaboration, bot – bot, bot – human, human – bot, human – human, to further increase the service center’s efficiency. Bot to bot collaboration and bot to human collaboration are for smooth handovers, as for the foreseeable future bots will stay focused on narrowly defined scopes. Human to human collaboration is again a smooth handover to the right expert, but is also about educating the colleague by helping out with own specialized experience. Finally, human to bot collaboration is about the human training the system on the go. Last, but not least, by hiring the right people. An early 2017 study by Harvard Business Review on Kick-Ass Customer Service revealed that call center managers are hiring the wrong people. In scenarios that increasingly deflect calls it needs more highly trained controllers and rocks with a mindset for collaboration, rather than empathizers. While empathy is important what matters most when dealing with a customer in an aggravated mood is a fast and efficient resolution. With this, the role of the manager will change, too, into the direction of being a servant to the team and taking care of roadblocks for the agents and fostering collaboration. This collaboration mandatorily extends into the product department. The best issue is the one that doesn’t even occur. Data from the call center, and from the app itself, gives unique insight into possible problem patterns. And the best problem to have is the one that doesn’t even occur. DevOps gives an idea on how this can get achieved. Ah yes, don’t script too much. Scripts are a good guidance for someone unknowledgeable, which the future call center agent is not. The future of the call center lies in a high degree of automation, powered by highly skilled and motivated agents. That brings the human back to customers and agents alike.

A few weeks ago I wrote an article about customer service in a world of ambient computing. This article looked at customer service from a customer’s point of view. In it I described how I see customer service getting humanized again by leveraging the advances in AI technologies like Natural Language Processing, speech-to-text- and text-to-speech generation along with intent determination. Leveraging these technologies customer service will turn into a conversation and it won’t matter anymore whether service is delivered by a bot or by a human. For the customer it will all appear to be the same. Instead of FAQs or web searches, bots will be the first line of support and escalate a problem to humans if they cannot solve it on their own. The obvious question is whether there will be an impact on the customer service center? And it probably does. Call centers, and with it the service agents as well as their managers, already now are under intense pressure to deliver, and to deliver more efficiently. With the increasing use of call deflection technologies like FAQs and communities there is a trend for the incidents facing the agents becoming more challenging. For example Helpshift states that already with its technology it is able to deflect about 90% of all incidents, which are solved via the native in-app FAQ that is delivered by the them. This statement basically says that the support staff is basically relieved of dealing with simple matters but has the chance to take up the more challenging ones. Still, in a world of ambient computing any given app can have hundreds of millions of users. Let’s say that any given day just one per cent of 100 million users have an issue. The well working FAQ deflects 99% of these. That leaves the service center with 100,000 calls. In one day. And they are the harder ones. Still, let’s be optimistic and say that an agent can solve 10 issues an hour, giving him 80 in an 8 hour shift. This would mean an overall call center size of 1,250 agents is needed to cope with this demand. Each of them under a tremendous stress level. With the systems behind bots becoming more and more intelligent the difficulty of raised issues will increase, even if FAQs and web searches are essentially hidden behind a bot interface that essentially makes the human agent the second point of contact again as opposed to the third, which likely means that the customer’s level of annoyance is slightly less elevated than in a third level scenario. At the same time it seems that call center agents are not prepared for handling this stress level. The employee turnover rate remains high and is probably even rising.

What Does This Mean For The Service Center?

Call centers are therefore facing a double challenge

1. Contain cost. This is achieved by more automation, which in turn puts more strain on the employees
2. Employ and retain a highly skilled set of service agents, which additionally have matching character traits, which drives cost. Skilled people tend to be more expensive than unskilled ones, and moving a call center into a low-salary country helps only so much – if at all. Training comes at an expense as well. This will be somewhat augmented by reduced hiring cost

The solution to it will be multi-faceted and increase a trend that is already visible. Implement intelligent systems that more than offset the higher salaries demanded – and deserved – by the fewer call center agents, through an increased solution rate and through them being of more help to the service agents. These systems will significantly rely on machine learning out of a variety of sources Employ communities by incentivizing to other users to help other users. These communities will be managed by community managers, with increasing support by AI-driven bots. Highly data driven prioritization and intelligent grouping and routing of incidents to the best matching agent, bot or human. This will involve sophisticated Natural Language Processing capabilities but will help in solving multiple calls regarding the same problem in one process Improved collaboration, bot – bot, bot – human, human – bot, human – human, to further increase the service center’s efficiency. Bot to bot collaboration and bot to human collaboration are for smooth handovers, as for the foreseeable future bots will stay focused on narrowly defined scopes. Human to human collaboration is again a smooth handover to the right expert, but is also about educating the colleague by helping out with own specialized experience. Finally, human to bot collaboration is about the human training the system on the go. Last, but not least, by hiring the right people. An early 2017 study by Harvard Business Review on Kick-Ass Customer Service revealed that call center managers are hiring the wrong people. In scenarios that increasingly deflect calls it needs more highly trained controllers and rocks with a mindset for collaboration, rather than empathizers. While empathy is important what matters most when dealing with a customer in an aggravated mood is a fast and efficient resolution. With this, the role of the manager will change, too, into the direction of being a servant to the team and taking care of roadblocks for the agents and fostering collaboration. This collaboration mandatorily extends into the product department. The best issue is the one that doesn’t even occur. Data from the call center, and from the app itself, gives unique insight into possible problem patterns. And the best problem to have is the one that doesn’t even occur. DevOps gives an idea on how this can get achieved. Ah yes, don’t script too much. Scripts are a good guidance for someone unknowledgeable, which the future call center agent is not. The future of the call center lies in a high degree of automation, powered by highly skilled and motivated agents. That brings the human back to customers and agents alike.

Earlier in June I had the opportunity to talk to Barry Coleman, CTO of Agent.ai, an about 2-year-old company at the time of writing this. The company spun off of manage.com, a very different business that enable the delivery of in-app advertisements. In order to support this mission more and more, first internal, then external support capabilities were needed. At first they built chat functionality for internal and for support purposes. Then there was the question of how to efficiently provide 24/7 support. This resulted in giving birth to a bot structure that can help customer service agents in an assisting mode, called co-pilot mode, and an autonomous mode, called autopilot. And it gave birth to Agent.ai. Agent.ai’s mission is to enable “exceptional customer service for all”. While this mission is not particularly unique, their approach is. First, Agent.ai has built its customer service software around a machine-learning platform. Second, the company provides their solution without asking their clients for a huge upfront investment or the need to have of AI-proficient developers in house. Third, they wanted to avoid the pitfall of inflated expectations. With AI and machine learning being very hyped topics at the moment, this is a very valid concern. Going backwards through the objectives, Agent.ai opted for offering very specialized bots first. As there is no general AI yet, this is pretty straightforward. Specific, tightly framed topics are far easier to support with AI and exposed by bots than broader bodies of knowledge. For example, specializations include the handling of order inquiries or of support call closure surveys. The second objective was achieved by doing all the heavy lifting, including the customer specific training of the AI in their own system, by providing specialized bots, and by offering APIs for their customers to implement own specialized bots. One interesting aspect is that Agent.ai’s software fabric allows the individual bots to collaborate with each other and communicate internally with agents and externally with customers. This collaboration is necessary due to the strong specialization of the bots and is mainly controlled by a ‘central’ AI-based bot that resides in the Agent.ai infrastructure, called ‘AVA’, which is an abbreviation for Automated Virtual Agent. AVA is the brains of the system. The job of the AI bot is to understand speech and to identify a user’s intent using NLP, neural networks, and deep learning. This intent could be a request for information or a call to support an incident. With this done the AI bot dispatches the incoming request to the corresponding specialized ‘intent’ bot that can take up the transaction and hand it over to another bot, or escalate to a human agent in case they get stuck. The system is trained from a variety of sources, such as FAQ, existing documentation, and e-mail trails. Chat transcripts prove to be especially valuable as they allow for identification of both, problem and a solution. These transcripts also offer an excellent means for continuously training the bots while being in co-pilot mode, the mode in which they suggest answers, along with a confidence level in the answer, to human service agents. The usage of chat protocols along with the service agents choosing to use bot recommendations or not, allows for constant recalibration of suggestions’ confidence levels. Which leads to the topic of trust; user trust as well as agent trust – and to the question when a specific bot can be put into the wild and work autonomously. The answer to this is surprisingly simple although there is no explicit measurement: If suggestions consistently exceed a defined high confidence level then the bot is good to go unsupervised and escalates issues it cannot answer itself to a human service agent. Another possibility of identifying trust levels is the change of customer sentiment in the course of a transaction. Working in co-pilot mode, with the ability to have bots work unsupervised, human agents free up the time to work on novel problems. Typically, these can be the issues that bots haven’t been trained for, and maybe cannot be trained for. Barry emphasizes that “human-machine cooperation is really important”.

My Take

Agent.ai has an interesting story to tell. The idea of offering an affordable infrastructure to provide 24/7 mobile in-app customer service using bots that are driven by machine learning and AI is probably not new but consequently implemented. Bots can considerably speed up the support transaction by continuously listening to specific queues. With well-trained bots this can lead to positive support experiences by showing that a customer’s time is valuable. This also applies to the co-pilot mode, when the bots can already prepare suggestions along with confidence levels that help the service agent prepare herself for an issue. In addition to providing a toolkit for mobile in-app support, Agent.ai is supporting nearly all major messaging platforms, which allows for richer customer profiles as well as for a wider reach for both, Agant.ai’s customers, and Agent.ai itself. Agent.ai’s customers can offer their customers availability on the channels they prefer without being in the need to look for additional vendors to cover different messaging channels. Agent.ai’s bot-driven mobile first approach puts the company into an interesting position. Mobile in-app specialists normally do not support messaging services with the correct argument that the service engagement can be made far more personalized. This is due to more information being available to the service agent via the SDK. It simply can provide more information than the messaging service will ever do. On the other hand there will be many users who simply do not want to install vendor apps. Integrations with Zendesk and Salesforce give exposure to the world of the ‘big guys’. Zendesk does believe that bots are not yet far enough to be really useful in customer facing service interactions. Meanwhile Salesforce does not have any bot capabilities either, as far as I know. Both companies offer integration into major messaging apps, with Zendesk also offering an app SDK, though. Still, this leaves an opportunity for innovative vendors. I believe that the strategy of covering the breadth of mobile along with the ability to cover small to big customers is pretty strong. It puts Agent.ai dead into a spot that no major vendor covers, while at the moment having a technological advantage. However, there are also some concerns. I suspect that the approach of taking away the ‘heavy lifting’ from customers may lead to consulting services, which do not scale well. In addition Agent.ai is a young company, which always raises the fear of viability. Agent.ai says it has more than 1,000 customers from small to large in different industries. These customers are using SDK and web client most, followed by the Facebook Messenger. While this sounds like a big number there is no information on actual users. Still, this is a company to watch.

With mobile phones taking over our lives and conversational interfaces becoming ubiquitous there is certainly a new level of demand arriving at customer service centers. Customers do not accept a mediocre service experience anymore. With their smartphones they have the means to get to customer service with nearly no delay and they are certainly willing to use it. And they do it. In this situation customers are often already feeling some frustration or disappointment because they couldn’t achieve what they wanted to achieve in the first instance. They already had their taste of a suboptimal customer experience. Frustration, disappointment – customers’ negative emotions towards a brand have a corresponding negative impact on the business. Customers just might go buy somewhere else. After all, in times of smartphones this has become simpler than ever. The support center now has best chances to add the feeling of being disrespected and outright anger into the mix. Or it can create a feeling of relief, of being respected, valued, even some satisfaction; this in spite of having come into the need of asking for support. Here the service agents have the opportunity to create a positive customer experience out of a poor one – one that will overlay the negative one.

Use Customer Service To Create Positive Emotions

Source: Sitecore/Avanade Which one is better for the company – and the company’s bottom line? The answer to this question is pretty obvious. Inmoment Research recently released a study that clearly established links between positive experiences and positive outcomes for a company. And this was not the first study finding that investing into positive customer experiences results in positive outcomes for the company. A 2016 study by Sitecore and Avanade showed a number of tangible business benefits that can be related to this investment. The open secret is that there is an easy way to turn a customer who is on the edge into a satisfied and happy one. This way works around recognizing what the customer values:

• Their time
• Their channel
• Getting their issue resolved

This way clearly involves the smartphone. After all this is the switchboard of their lives and in all likelihood also the device where the issue occurred.

Go Social? No Way!

You may say: “Go social! Twitter, Facebook! People go there all the time! And they are getting fast and efficient help there!” Do they? Most customers aren’t getting fast and efficient help on social networks. Only few do. Mainly those with loud voices, because of having large follower numbers. And why are people resorting to social media? They use them as a last resort after the traditional ways didn’t work. In other words, because they are disappointed, frustrated and likely even angry. So they act as derailers of the brand, instead of being brought on the way of becoming promoters or even ambassadors.

Resolve Issues Fast, The Customers’ Way

This is where conversational interfaces, messaging and mobile in-app support come into the picture. This combination allows a customer to directly cut through to efficient service instead of repeatedly iterating through information they have already given before. Via the app the customer and many of the interactions (s)he did before the incident are already known. An agent can get to work directly, maybe even supported by a bot as a copilot, or even an autonomous one. The emotional advantage starts right in the beginning. In most of the cases the customer is already known and can be addressed by name instead of being in the need of introducing him-/herself. And due to the information that is already available the issue has been routed to the right service agents who in turn can get right at work when the ticket arrives on their desk. And this may very well be before they even reply, thus giving themselves the chance to be equipped with relevant knowledge and perhaps already a solution suggestion. As a result the customer feels treated respectfully and with appreciation of his/her concerns and time. Now, it is an illusion that every call center has enough staff to be able to immediately react to every incident and inquiry customers may have – ideal but an illusion. There is nothing worse than the electronic equivalent of holding music and the repeated statement that “your call is important to us”! Its mere existence actually proves the contrary.

Use AI to Improve the Experience

This is where AI and machine learning, exposed by chatbots take center stage. In a supportive role the AI can already supply the agent with possible solutions to the incident at hand before the agent takes up the issue. These solutions can get derived from the FAQ, internal documentation, a community, or from previous incidents that dealt with a similar topic. The ranked listing of suggestions can then be presented to the agent by a chatbot as part of the conversational interface that is used in between customer, service agents, and the technology. This already has the potential of significantly speeding up the resolution process, thus addressing what the customer values. It is a better engagement model that likely creates a positive experience – maybe a little wow moment. When enough trust has been established to the AI, the chatbot can get into an autonomous role and engage with the customer before an agent takes over. The human agent then becomes the escalation point for issues that the system cannot resolve without human interaction. To get into this escalation the bot could offer and schedule a callback by a person or simply do a handover to the next service level after informing the customer about the necessity for this. In this model even an explicit FAQ can be made obsolete as the chatbot can serve as a front end to it.

The results?

Conversational interfaces, alone or supported by chatbots have a profoundly positive impact on the business. The main benefits are:

• Customer service is addressed the customer way
• Faster incident resolution
• More satisfied customers who have a better customer experience

Satisfied customers have more positive emotions about a company. A positive base is more likely to perceive an engagement as positive, which means it turns into a positive customer experience. And positive customer experiences create positive business outcomes. It is that simple.

Just on may way back from CRM evolution 2017 it is time for a little recap. The conference, once more chaired by CRM Grandmaster Paul Greenberg, was again co-located with sister conferences Customer Service Experience and Speechtek. Why there is a separate – and smaller – conference for Customer Services co-located with a CRM conference is beyond me, as Customer Service is an integral part of CRM. But be it as it is. CRM Evolution attracted around 500 professionals, being second to Speechtek. The main topics this year seemed to be Customer Engagement, Customer Experience, and AI, nothing of this coming as a surprise. The size ratio of the conferences and the topics were also confirmed by the exhibitors in the Customer Solutions Expo. We saw an abundance of little booths with AI- and bot-vendors. The mainstays of CRM had fairly small presences, notably SugarCRM, which had a big presence last year. Both keynotes dealt with delivering to maximize customer experience and to measure the result. In the opening keynote Gerry McGovern answered the question what great customer experience is in a digital world and then how to measure it. His premise is that customers want to get something done and that it therefore is imperative to help them getting it done as fast and easy as possible. That begins with page load times, goes on with simple check-out processes like Amazon’s famous one-click or Uber’s payment process – hint there is none at the end of the ride. Things are as easy as saying good-bye to the driver and opening the door. Throughout his keynote Gerry made the point that it is the relentless pursuit of customer convenience that drives customer loyalty and retention, and ultimately revenue growth. The true measure for this is the company’s ability to shave off time that the customer needs to spend to get her job done. From here on I concentrated on the ‘Filling the Pipeline’ track for the day. Michael Fauscette from G2Crowd started the day delivered his view on the state of digital marketing and Steven Ramirez of Beyond the Arc gave a practical guide on how to implement and use predictive analytics to improve marketing. The most intriguing session of day one that I attended was sadly also one of the least attended one. Lora Kratchounova presented Account Based Funnel Management, her extension of Account Based Marketing that brings a more seamless integration of the marketing- and sales departments. Paul Greenberg on day two brought in his perspective on how to properly deliver good customer engagement. He started off by identifying the current ‘breed’ of customer, followed up with what this means for engagement and gave a framework for delivering. Today’s customer differs from the one of about 20 years ago – a theme that I brought forward in my own presentation, too – good to be confirmed by the grandfather of CRM. Today’s customer is digitally savvy, connected, impatient, and expects information and responses nearly instantly. Delay is not an option. Customer engagement according to Paul is “the ongoing interaction between the company and the customer, offered by the company, and chosen by the customer.” The resulting experience is the “customer’s perception of the company over time”. I tend to disagree with the over time part but else this definition seems to nail it. Based on this he suggested a framework for delivering that bases around the ideas of expectations, knowing the customer, relationship, resources, value and culture, backed up by two case studies. The most important insight to me seems to be that it is important to treat the customer as a partner and a subject of an experience rather than an object of a sale and merely a client. The remainder of the day, with the exception of my own presentation on helping customers to a good experience through improved engagement, was more or less about AI. We started off with a breakfast with the influencers, an informal panel discussion with thought leaders Denis Pombriant, Ian Jacobson, Josh Greenbaum, Michael Wu, Brent Leary, Sylvana Buljahn, and myself, moderated by Esteban Kolsky. Reflecting the state of the worldwide discussion we more or less immediately arrived at the topics of trusting the machine and ethics. A discussion, under the participation of all attendees, that surely needed and deserved more time than we had. Later the day Brent Leary explained how voice activated conversational interfaces will change the ways of customer engagement, which I sadly couldn’t follow through due to an appointment. It was followed by a vendor panel about AI’s role in shaping customer engagement and Esteban Kolsky delivering his insights into the new reality for automated interactions in an AI world. The panel was set with representatives of SAP, Microsoft, Salesforce, Oracle, and – the only REAL subject matter expert – Michael Wu. Both, the panel and Esteban agreed that the use of AI/machine learning for optimization purposes is a little short but that there must be a significant engagement improvement coming with it. Esteban, in his customary way of driving the point, taught the audience that AI affects the enterprise in an O P A way: Optimize, Personalize, Automate. Day three somewhat returned to the engagement and experience topics with an interesting presentation by Sylvana Buljahn on how to reach the next level of customer engagement by becoming both, a customer- and employee-centric leader. It is certainly worth while listening to Sylvana. In the course of the conference I had the opportunity of speaking to SAP and received a deep dive into Thunderhead, which for me is also the vendor of the conference. The company has a truly amazing solution to improve the omni-channel customer journey by linking the customers touchpoints through their way – from first contact to retiring the purchased solution. They do this based on the ideas that the journey belongs to the customer, not to the company, that a customer can be on any given number of journeys at the same time and that it is of highest importance to integrate systems. More on both in other posts.

My Take

CRM Evolution in my eyes is still the vendor independent go-to conference of the year. If the intention is to get insight into the state of the art of things CRM rather than getting deep into any vendor, it is the right conference. The networking events are vibrant and one can discuss a lot of interesting topics, getting hit by many differently valid points of view. The conference, however, somewhat suffers from the artificial distinction between CRM and Customer Service. Customer Service is an integral part of CRM and they cannot be treated in isolation when customers shall get a positive experience: Customer Experience is the Customer’s perception of the Company (over time). This involves the collaboration of all departments, especially marketing, sales, and service, So, I hope that this distinction gets removed again.

2016 has been predicted to be the year of conversational commerce, and I’d say that this prediction largely held true. Conversational interfaces have become more and more mainstream, and their support by AI and bots has become all the rage. While people more and more turn to their smartphones and Google to find service and get answers to their questions, companies are increasingly looking at bot support to increase the efficiency of their call centers. But where is reality? In their 2016 hype cycle on emerging technologies Gartner places

• Conversational UIs in the innovation trigger phase with a predicted time of 5 – 10 years to mainstream adoptions
• Machine Learning on the peak of inflated expectations with a period of 2 – 5 years to mainstream adoption

Forrester Research in their recent AI tech radar places virtual agents and machine learning into their growth phases of their respective life cycles, giving them 5 – 10 years to mainstream, while acknowledging a successful trajectory. So, clearly, AI and conversational systems are strategic. At the same time Abinash Tripathy, CEO of helpshift, a leading helpdesk company providing users with instant, proactive, and personalized in-app support, feels that “we are closer to IoT than to having really helpful bots”. Some bots are actually harming the customer experience. And he is right. Why? Several reasons. Essentially artificial intelligence, driven by machine learning or deep learning, is not yet intelligent enough. Too many bots are still driven by decision trees, which severely limit the possible conversations that the bot can serve. Second, bots’ ability to understand natural language is still lacking, albeit improving, and probably improving fast. At the same time a handover to a human agent, or another bot – think bot-swarm – is often poor or non-existent, leaving the customer with an unresolved question and in limbo. This leads to a poor customer experience. Additionally, the bot’s, as well as the overall help system’s, integration into a knowledge base is crucial, but often underdeveloped. A customer query needs to be translated into a meaningful query to the knowledge base and/or additional information. Think “What is the status of my recent order?” or: “I cannot receive calls but can dial out?”. This needs deep integration into a corporate knowledge base as well as transactional systems. While a human agent can cover the lack of systems integration, a bot cannot achieve this. A bot is depending on the intelligent, and continuous indexing of corporate knowledge. Again, poor customer experience. In summary, Abinash maintains that “the need most bots are filling should not require an artificially intelligent conversation. If they are, they’re probably not doing it very well (yet). Rather, the best chatbots allow users to interact with their surroundings (like the baseball game example), act as refined search engines, or provide real-time updates”. At the moment “Chatbots are best used to relay simple updates”. But, wait! On one hand AI and bots are part of the future and then they are not really useful but harm the customer experience. Isn’t that a contradiction in itself? No, it is not. Merely a question of Thinking Big while Acting Small and doing first things first. As I have written before an important part of providing good service is being available to help the customers on their preferred channels, at the time of their choosing, and at their pace. And one of the main channels is the smartphone, the second is chat. Additionally, people are starting with search before going for direct support. So, the way to help customers is an integrated, intelligent, efficient service offering that helps them getting to the information that they want with minimal effort on their side. The keywords here are: Mobile, app, search, and chat. Combining this we arrive at an offering that offers customer service directly in app, with an integrated, local knowledgebase, integrated and embedded chat, as well as the easy ability to turn the conversation into a voice (phone) conversation. Due to being embedded into the app, this offering also helps the agents by delivering contextually relevant information that shortens time to resolution. The same works for embedding this offering into a web site. Add a back end that efficiently supports the agents with automation, workflow, collaboration tools, a clean user interface, and that integrates well with community systems, knowledge bases, other OLTPs including CRM systems, and the enterprise- and web content management systems and there is a strong foundation for improving even further. Once this foundation is in place, AI and bots can be deployed in a useful manner, first starting in a learning-only mode, then more and more engaging in customer initiated as well as company initiated interactions. These bots can ensure quick reaction; they gather missing relevant information can already offer solutions for the simpler problems. Minimally they keep the customer engaged and do a seamless handover to a human agent. This does not only help the customer, but the agent, too, as she is prepared and can dig right into the problem at hand. As the chatbots continue to be supported by a learning system they ‘learn’ from ongoing conversations and therefore the problems they solve can become increasingly complex. This fact, plus their ability to support the agent via continuously suggesting good solutions based upon the conversation flow and the knowledge base further increase the efficiency of the supporting agents who can increasingly support on more challenging problems instead of being in the need to ask repeatedly for the same information and answer the same ole questions over and over.  

After the Hype

2016 has been the year of bots, AI, and automation the beginning of 2017 seems to be the time of looking at wider implications. There is a lot of discussion going on in academia, politics, and on the web, e.g. the one spurred by Denis Pombriant with a very readable article, and two follow-ups here and here, in November and December 2016. Denis, supported by Vinnie Mirchandani, took a very optimistic stance – something that is highly important in times of simplification and pessimism. There is no doubt in my mind that technologies that are driven by artificial intelligence can have a tremendous benefit for both, companies and organizations, as well as consumers. Consumer technology like Amazon’s Alexa, Google Assistant, Siri, generally intelligent home automation, self driving cars, etc., can simplify peoples’ lives tremendously by taking away routine activities or making it just easier to execute them. Organizations can create improved customer and employee experiences via automating existing processes, and they could create entirely new experiences using technology – doing things more effectively. Opportunities to do so can be found within the complete value chain. Automation also serves the aspect of doing the same, or more, at less cost, i.e. more efficiently. And in the last point lies a catch. This means that less people are needed to deliver on an amount of work. This means less employed people and, on a first view, more unemployment. This means less disposable income. Because advances in technology have the tendency to benefit only a few, which are those who deliver the automation systems and those who are able to invest into them. In a pure, shareholder value driven, capitalistic system this means that capital is more important than work. The best-known arguments against this pessimistic scenario are that

• Technology does not automate jobs but tasks, particularly dull, dangerous, and dirty ones
• History shows that technology itself creates new jobs, or a variation of this theme
• AI and bots are not good enough to handle everything – a lot of work needs to be done behind the scenes by real humans
• Growth that comes with the productivity gains creates even more jobs
• There already is, or will be a shortage of personnel, due to demography

I guess they are all true, although, as a Nota Bene to the history argument, technologists argue that backward-looking analysis is not good at predicting the future. On the other hand there are arguments like

• Productivity grains deliver growth only if businesses do not focus on efficiencies
• Technology contributes to increasing income gaps
• The tasks that can be taken over get increasingly complex (‘disruption from below’)

A recent report by Pew Research found that there is a lot of fear around the longer-term availability of jobs, but not necessarily an imminent feeling of being threatened. This fear also was mirrored by the discussions that went on in the aftermath of Denis Pombriant’s articles. On the other hand this and this research by McKinsey shows that adoption of automation technology will take some time and not all areas of work are susceptible for automation at all. On a personal note I doubt their assessment on the application of expertise …

So where is the Truth?

It is fairly clear that there is a lot of fear and uncertainty. And it is obvious that there is a combination of factors that drive this fear, including economical ones like a (right or wrong) feeling of being cut off the wealth ladder. It is also clear AI and robots will not take over the world by storm. But it is equally clear that the level of automation that especially AI offers goes far beyond the realm of physical work. This is something on a level that we haven’t seen before and that reminds me of ‘disruption from below’ – automating the simpler tasks and getting ever more into the complex ones.

How does it link to Customer Experience - The Way Ahead?

There are two main themes about this

• Computers and humans have different strengths, computers regularly are better with numbers where humans are better when it comes to connecting the dots
• There is a link between employee experience – and their perception – and customer experience.

Combining them shows the path into future. Engaged employees that are secure about their own future can concentrate on delivering superior experiences. Great software helps employees concentrating on what is really important by taking away routine work that does not create much value; on the other hand it is already now visible that customers are looking more and more to the web when it comes to inquiries and sophisticated software can avoid the need to repeatedly provide the same pieces of information – thus diminishing the customer experience. The way ahead, for the foreseeable future at least, can lie only in man-machine collaboration where the machine takes over the dirty, dull, dangerous, the routine tasks and the human concentrates on non-routine work that cannot get automated – yet. Will this lead to fewer people being needed in the workforce? I don’t know. What I do know is that work will change, that humans will concentrate more on ‘judgment work’, as opposed to knowledge work, and physical work, and that we can expect types of work to emerge that most of us don’t even dream of at the moment. Interesting times lie ahead of us!

Bots are all the rage currently. By the looks of it they are at the peak of the hype cycle. We will see their deep fall into the trough of disillusionment soon. After all the well-known examples based on the Facebook messenger are somewhat underwhelming, to formulate it carefully. There is not much artificial intelligence visible - nor needed - to provide services like these. They also come with a poor user interface. And this combination of hyped examples, mixing up chatbots and AI, has the potential to kill customer experience. They certainly kill the user experience. Unless, this is, that these machines already reached a level of intelligence that they are magic to my simple mind... Which I doubt. To be sure, there are AIs around that amaze us: IBM's Watson, Apples Siri, Microsoft’s Cortana, Google's Now, ...   even Microsoft’s infamous Tay which got a pretty bad reputation in no time, to name but a few. Recently a whole class of graduate students didn't realize that their teaching assistant Jill Watson, an AI based upon IBMs Watson, was actually an AI and not a person. And I sincerely believe that in not so far future we will see AI in many places that is indistinguishable from a human. As Salesforce's Marc Benioff recently said we will have AI do things that we cannot even imagine right now. The potential is virtually endless (pun intended). But what we see right now being built standalone or embedded into messaging apps has nothing to do with AI and it often has a poor user interface. This needs to get fixed, or else the potential of falling back to an 90's customer experience becomes very real. I get the potential benefit of connecting to businesses via one single app. And I do think that it is the right way forward! After all this is the very principle that is behind my Epikonic platform with its app front end. But the way the Facebook Messenger looks right now benefits exactly Facebook, which doubtless has a fairly elaborate AI behind the scenes.

The Way Ahead

But enough of the rant. I do not want to be destructive but show a possible way ahead. Because I think that even the simple bots that we see right now have their value, while the end game is in interacting with full-fledged AIs that we perceive as humans, very patient humans. According to Wikipedia a chatbot or chatterbot is “a computer program which conducts a conversation via auditory or textual methods. Such programs are often designed to convincingly simulate how a human would behave as a conversational partner, thereby passing the Turing test. Chatterbots are typically used in dialog systems for various practical purposes including customer service or information acquisition. Some chatterbots use sophisticated natural language processing systems, but many simpler systems scan for keywords within the input, then pull a reply with the most matching keywords, or the most similar wording pattern from a database.” This definition is generic enough that even a simple flower ordering app on the FB Messenger that leads a customer through a predetermined process can get sold as a bot. And this is part of the problem. In order to make bots more useful and therefore accepted we need to work on a few frontiers. And I am sure that this will happen as part of bots moving through the trough of disillusionment. Part of this process will also change our perception of what a bot is. We stop talking of bots (who is talking of social CRM anymore?) and/or will redefine where the value add for customers and businesses is, by answering two basic questions:

1. How does does a bot help my customers to better accomplish their “jobs to be done”? Be they ordering of flowers or organizing a family vacation next spring, or whatever else. This is the harder question to answer.
2. How does it help the companies using bots improving their business? This one is related to the first question but also touches efficiency, not only effectiveness.

Supporting this these are the main three things that I think research and industry need to work on are:

• An improved definition of what a bot is, in contrast to a simple application that can achieve the same. There is no need for a “bot” that allows me to set reminders for myself or get travel tips or to subscribe to a newsletter. There is also no need for bots that fail at basic natural language detection. Scanning for keywords as it is still very common simply doesn’t cut the mustard. And the real value lies in a meaningful interaction that successfully imitates an interaction between humans. This definition also needs to involve a distinction between a user interface and the the underlying logic. I would argue that many so called bots are actually just another user interface that is plugged on top of existing applications
• Standardization; so far it looks like the APIs for every platform are different. I do not think that all APIs will ever look the same but some basic services need to get standardized, also to make it easier for developers to deploy across different platforms.
• The ‘killer bot’; at the moment we have bots that basically do the same as apps/applications. And then these ‘old style’ apps are often richer in functionality and more convenient to use. A useful bot needs to deliver a use that is hard to deliver for a conventional app. Some (rudimentary) candidates that I see are in the area of financial services, like Digit or Penny. But I can imagine many more uses in complex areas like healthcare or day-to-day self-organization. Key are rich, useful functionality in combination with a simple user interface and integration to other used applications. Text, admittedly is an interface that we are comfortable with, but that doesn’t make it a good one. How many people are struggling with typing?

Maybe it is worthwhile ceasing to speak of bots altogether when we just mean a user interface? Just thinking …