The BOT(tom) line – Chatbots are one of the biggest innovations in the marketing industry

Chatbots are one of the biggest innovations in the marketing industry but in order to use them effectively, brands must carefully consider the customer journey and how they implement this technology. Mike Betzer outlines the dos and don’ts so your brand can make the most of bots.

Off the top of your head, can you name the one greatest advantage that every single marketer has in their arsenal? It’s that every marketer is also a customer, so they know exactly what it’s like to be marketed to, and can appreciate very easily when it’s done well – and when it’s not.

But how often do marketers forget that knowledge when it comes to embracing the exciting and innovative developments in the marketing space? Marketers are an innovative group, often embracing the latest and greatest technology to make sure every advantage is grasped. Sometimes though, when it comes to tech, that old adage holds true: just because we can doesn’t always mean that we should.

Consider the chatbot.

If you know anything at all about the history of this innovative marketing tool, you may well be aware it’s not that new. ELIZA, considered the mother of all chatbots, was developed by MIT engineers over 50 years ago. Of course, the technology has evolved profoundly since then. Developments over the last few years have enabled bots to appear more human than ever before and, equally importantly, to be more adept at dealing with human conversations.

Brands now have opportunities to create interfaces that feel genuinely human and interact with customers and prospects in a manner that feels more personal and responsive.

But just because you can…

Before you jump straight to implementing chatbots in your digital marketing strategy, think back to the times when you’ve come across them yourself as a customer, and at what point in your journey were they used. What was your first impression? Research has found that 65 percent of customers believe chatbots are slower and less accurate than human agents – and perception is everything. Brands that rely on bots too much and use them at inappropriate times will see an ensuing effect on customer satisfaction (CSAT), leading to negative impacts on revenue, retention and reputation. However, even though two-thirds of your potential customers think less of chatbots than of human agents, more than half say they don’t really care if they interact with a human or a machine – the only important thing is getting the answer they’re after.

BOT(tle) it

Used wisely, a bot can ensure your customers get more help, more accurately and more quickly. So how do you know when and where is the best time to use this particular tool?

To decide which is appropriate – a human agent, plain button bot or fully conversational bot – ask yourself the three following simple questions:

  • Is this a conversation that can be automated with buttons?
  • Will customers react favourably to it? (i.e. would you react favourably as a customer?)
  • Will it be straightforward to maintain?

Dos and don’ts

Once you’ve determined that a bot is not only an appropriate tool for your brand or organisation, but also suitable for a particular stage of the customer journey, the next step is to think very carefully about how you implement it. The following list of dos and don’ts will point you in the right direction:

Do

Welcome users and set expectations – just as the salesperson at a shop counter or receptionist in an office sets the tone for customer expectations, the chatbot may well be the first point of contact a user has with you and your brand. Make sure they are greeted accordingly, but also let them know what they can expect – what kinds of queries or issues the bot can help them with and, importantly, how much time this will take.

Make the most of rich messaging channels – WhatsApp, Apple Business Chat, Facebook Messenger and their ilk allow you to deliver images, links, videos, catalogues and more – and these are all ways to make the chatbot’s conversation so much more than a simple text only affair and, consequently, more engaging. Pair your chatbot with a scalable conversation management platform that can support agents as well as bots in any channel that customers find convenient.

Mix up the buttons and natural language – buttons are a natural fit for quick and easy decisions, common interactions or yes/no queries – but they can also become tedious. Blending them with judiciously placed natural language results in a much better customer experience.

Make switching to a human agent easy – nothing infuriates a customer more than getting caught in a vicious circle when the AI cannot answer their query, but sends them back to the beginning of the process time and time again.

Ensure that there are buttons that will reach human agents and carefully identify any common phrases customers may use when they want to talk to a person. Then ensure that transfers or handoffs between agent and bot are seamless and intuitive.

Don’ts

Bite off more than you can chew – always keep your unique business model in mind and create flows that support the most often reiterated interactions you have with your customers. This also means considering what your team can handle and throttling your systems up and down as necessary to avoid overwhelming capacity and delivering a poor customer experience.

Try to pretend your chatbot is a human – customers hate being taken for fools. Even the 29 percent of people who say they sometimes can’t tell whether they’re interacting with a person or a bot will respond much more positively if their expectations and understanding of the bot’s limitations are made clear from the start.

Set and forget – we already know how far bot technology has come since ELIZA debuted in 1966, and as with any piece of technology, bots continue to evolve. More importantly, so do your business and the desires of your customers. Stay up-to-date not just on the latest technological developments, but also on customer sentiment, to make sure you’re meeting their needs (without them having to ask you over and over again).

Offer too much choice or blanket text walls – these can be surefire ways to frustrate users. Keep your bot’s responses swift, relevant and easy to navigate.

Overdo the length of a button-based flow – this may be tempting, but customer requirements can often fall outside of preconceived options and a limited set of choices is a pathway to irritation.

Rely solely on a bot – if more than half of your customers are happy to interact with a bot as long as they get the assistance they require, that still means nearly half prefer not to use them at all. Conversations shouldn’t always be automated, and it’s vital that you plan for the unexpected and incorporate fall-backs and detection methods for any possible dissatisfaction.

___________________________________________

This article is written by Mike Betzer and originally published here

First-of-its-kind chatbot developed to support genetic counselling

Scientists at CSIRO, Australia’s national science agency, in partnership with Melbourne Genomics Health Alliance, have developed an Australian-first digital conversation agent (a ‘chatbot’) that could support patients in making informed decisions about genomic testing for future health risks.

Dubbed ‘Edna’ (which stands for electronic DNA), the chatbot is the first of its kind globally developed specifically to support genetic counselling for adults being tested to ascertain future risk of preventable or treatable conditions (known as ‘additional findings’).

These conditions include treatable genetic disorders such as hereditary breast cancer or cardiomyopathy.

Derived from real-world patient interactions, Edna is designed to answer the most generic and simple questions asked by patients, which then creates more time for genetic counsellors to focus their highly specialised skills on deeper and more specific issues relevant to patients.

Professor Clara Gaff is Executive Director of Melbourne Genomics, a ten-member alliance that includes CSIRO.

“If the healthcare system were to provide this kind of testing – which is beyond immediate medical need – one of the challenges is the genetic counselling time required to support patients’ informed consent,” Professor Gaff said.

“This prototype chatbot shows how we might employ technology to meet this need.”

CSIRO researcher Dr Dana Bradford, who led the development of Edna, said chatbots simulates human conversation through artificial intelligence.

“For chatbots to accurately recognise content in human speech – and provide a meaningful response – they need a large body of data to draw on, called a chatbot ‘brain’,” Dr Bradford said.

“Many chatbot brains are developed from open source data but this is inadequate for highly specialised fields like patient decision-making.”

“We developed Edna’s brain by systematically analysing transcripts of actual genetic counselling sessions for additional findings.

“This expert basis for Edna makes all the difference in applying this new technology.”

Edna is a downloadable smartphone app which can collect a patient’s family history and analyse human responses for signals that interaction with a genetic counsellor may be needed.

Edna’s development was part of a larger proof-of-concept study led by Melbourne Genomics to better understand the implications of offering additional findings testing to patients in Victoria.

“The Edna chatbot represents a significant movement toward feasible, real-world-informed digital health processes that can support patients’ informed decision-making about testing for future disease risk,” Professor Gaff said.

Edna is currently undergoing a feasibility trial with patients, genetic counsellors and genetics students, and is slated to undergo a larger-scale patient trial in the near future.

“When built in partnership with healthcare experts and patients, chatbot technology has enormous potential to provide and collect basic information in complex fields like genetics,” Dr Bradford said.

“Not only is the service on-demand, so people can access it whenever they wish, but it could free up highly-skilled expert time to build more effective care.”

The Edna chatbot was recently published in the peer-reviewed journal Patient Education and Counselling .

For more on the Additional Findings proof-of-concept study, read the protocol paper A novel approach to offering additional genomic findings—A protocol to test a two‐step approach in the healthcare system.

___________________________________________

This article is written by Ofa Fitzgibbons and originally published here

Three Steps To Overcoming AI Marketing Fears

AI creates unknowns, and they loom large over marketers. The 2020 State of Branding Report found that 56% of surveyed marketers think AI could negatively affect their brands by diminishing creativity, reducing jobs, or impacting differentiation.

A recent Brookings Institute study shows that higher-skilled jobs like marketing specialists are most likely to be affected by AI. But read further, and it becomes clear that opportunity is much greater than risk.

The Wall Street Journal notes of the Brookings study, “It is possible artificial intelligence will allow some workers to dispense with time-consuming tasks such as data analysis, and focus on potentially more profitable activities, such as meeting clients. Those workers could become more productive and command higher wages.”

AI fails like hackers using a mask to unlock iPhone facial recognition are public reminders that AI carries risks. Driving cars also carry risks, yet millions of people drive every day. The difference is that we’re used to driving, and we’re more comfortable assessing the risks.

There are three important steps marketers can take to better assess AI risks and move forward with valuable projects with confidence so that AI won’t diminish creativity, reduce jobs, or impact differentiation.

  • Use Data Patterns To Increase Creativity

    While marketers fret over major AI mishaps, thousands of extremely valuable AI projects are kicking off every day. According to Adobe, top-performing companies are more than twice as likely to be using AI in marketing than their less profitable peers. And IDC reports that retailers alone invested nearly $6B in customer-focused AI in 2019.

    And the most profitable marketers are using advanced AI-driven tactics like predictive personalization. AI actually has the power to increase creativity because it can analyze a lot of data and provide a wider variety of insights and recommendations.

    Rather than think of using AI like trusting a self-driving car, think of it as small AI-assisted elements like parking assist and lane indicators. These smaller assists help the driver without replacing them.

    The first step marketers can take to embrace AI without fear is to think in terms of moving from simple data points to data patterns in order to increase productivity and creativity. Marketers trying to segment an offer might divide customers into high and low-value buckets and perform an A/B test.

    This would lead to two different marketing messages. But, the segmentation can be further improved with AI that can recognize patterns over time and develop a much wider range of segments and offer combinations. AI might determine that some spenders are driven by seasonal signals or signals like the location.

    Complex insights are difficult for a person to evaluate, but with AI, they become clear very easily. Marketers suddenly have a lot of knowledge that can drive very differentiated, personalized marketing that can be more creative and more effective.

  • Let AI Do the Worst Jobs First

    AI frees the marketing team to move on to more interesting projects that can be layered on top of AI. James Manyika, Chairman, and Director at McKinsey Global Institute said that AI will change marketing jobs, and will actually create more jobs, rather than take them away. Let AI categorize millions of data points into patterns. Let AI sort and recombine text and image elements for a test.

    When that AI identifies twenty messages that could work on a whole bunch of new customer segments, that requires more creative power, not less. When AI tests show that personalized newsletter content drives more conversions, it requires more content, which needs to be written and edited. Even AI’s written content needs a human touch!

    As AI takes away some of the manual tedium of marketing, new positions are being created such as specialists in marketing intelligence, marketing optimization, and experiential marketing, which all harness the power of AI. For marketers just kicking off AI-driven projects, it’s important to identify what the AI will do, but also what the people are freed to do once it starts working.

  • Let AI Be the Stage for Brand and Customer Interaction

    Think of an image for AI and I bet some kind of robot comes to mind. Marketers are by nature creative, uniquely minded individuals, and so the idea of adding a level of robotics to a marketing strategy feels very unsettling. But AI is not a replacement for a brand, or for customer experience, but rather the stage that supports the interactions between the two.

    Banks are some of the most bullish AI investors, and they actually lead the way in implementing AI into their customer experience. The main reason is that banking customers prioritize efficiency, accuracy, and security, three things where AI can help tremendously. As the nature of daily banking changes and the mundane transactions can be automated, banks are actually free to differentiate themselves more, not less.

    For example, Bank of American uses an AI-driven chatbot named Erica to interact with customers on their app, interacting with more than seven million users to date for over 50 million transactions. Retailer StitchFix uses AI to create personalized monthly shipments of new clothes, ensuring that customers get a highly differentiated product, not just another little black dress. The AI gets better over time as it incorporates feedback from individual customers.

    Adding AI in places that could use added insight, efficiency, and scale increases performance. But, what about those small risks that still exist? What if an AI-driven test sends an ad for a winter coat to someone who’s actually in Florida? These risks can be mitigated with a strong commitment to data quality being fed into technology and by employing a personal review of any outgoing and incoming interaction.

    AI might recommend content that doesn’t make sense or might create a segment that doesn’t actually respond as predicted. These small fails only affect an individual campaign or small group of customers, which is actually a less risky fail than a national campaign flop. And what’s better, the learnings can be fed right back into the system for better results next time.

    ___________________________________________

    This article is written by TechFunnel (Team Writer) and originally published here