Bitter but Useful Truth About Bots for Business. Part 1
Defining the main fallacies most businessmen have and getting rid of them
Victor Seleykov
As soon as the first bots appeared, the business was promised a fortune through savings on client services. Today Juniper Research says bots will save $11B for banking, healthcare, and retail by 2023.
But let’s be honest, not everyone gets a chance to see the money profit, although bots indeed become better, cheaper and smarter — driven by natural language understanding and processing systems, machine learning, and other techs.
But what is that that business stumbles upon when chasing AI? How to predict investments and resources? How to be the one who saves, and even makes money on Conversational AI? Let’s get this sorted out.
First of all, we have to define the main fallacies most businessmen have and get rid of them.
We don’t know why we need it
Pretty often clients anchor too much hope for conversational AI. Quite often these inflated expectations are typical for SMBs, who have no idea of a bot, its working process or its artificial intelligence. Sometimes companies seek development with heady ambitions like “I wish I could idle the whole sales department — let that bot do the sales” or “I want that bot to bring more leads”. But bot won’t plug the strategy or marketing hole, it won’t make money or clients fall from the ceiling.
A bot is no panacea, it’s a tool that you have to create first, and then use and develop it constantly
Once you hire the whole sales department you do not forget of its existence for a year, right? Your employees come to you with ideas and problems, while you control KPIs and provide training. The same happens with a bot. The problem is that the business doesn’t know what it needs.
When you got one employee responsible for the accounting staff, delivery, marketing, and calls — it’s not easy to determine the problem a bot is supposed to solve. Initial consultation? Delivery handling? Leads generating? Well, you could experiment a little.
Enterprise-scaled companies are having clear tasks — vulnerabilities are evident because it is numerical. Big corporations are looking for a gap in their expenses in order to fill it. And, as usual, they decide to add conversational AI to their contact centers — there’s always something to optimize and the metrics are specific and familiar. But still, optimization itself is not a goal.
Before you reach out to the vendor (or before you start bot development by yourself) you got to figure out what you are trying to achieve — to increase the service rate and level, to cut expenses or to improve agents’ efficiency? There are also less obvious goals. For instance, companies rarely consider business scaling up by means of through-put capability extension. Or customer communications’ personalization by means of omnichannel and client data usage by a bot.
Imagine that, you’ve opened three new branches which caused intensified request stream — in this case, you don’t have to expand your contact center team, you can add a bot and this alliance will do the job.
Besides, business esteems bots as an instrumentality to save money, not to earn or reach new audiences. Although, a good marketing skill for a voice assistant is able to bring a clear profit! Again, the most known use case here — Nike’s sale during the NBA game — advertised $350 shoes were sold in 6 minutes when over 15000 people did their orders via Google Assistant using voice only. Well, you don’t have to be this ambitious — a simple voice game that gives a discount for your product when the user wins would also boost your sales!
We are confusing profit and efficiency
It’s not hard to count cost reduction for a contact center proceeding from OpEx (operating expense) — space and hardware lease, salaries, training of staff, and other expenses that the company spends on a day-to-day servicing. You take that and compare it to investment and working hours needed to launch a bot; you take into account the number of requests that your employees usually process, and traffic that a bot can handle, and you compare it. Now, this is more or less understood.
Efficiency – that’s the hard part. Because it is about service level and the customers’ happiness. Too many factors affect bot’s efficiencies: topics it works with, agents’ availability (whether a bot can pass on a complex request to a human), extra training (whether you update dataset constantly so that a bot could always understand your clients). Based on those factors, the service level is adjusted.
For instance, a bot is used in tech support in a bank. A client says they want to reissue their card. According to the scenario, a bot is supposed to ask the client to take their passport and visit any of the bank branches. Is that efficient work? Considering automation — yes. It understood the query, defined the topic correctly and gave the answer according to the scenario. But no client will say that was efficient. All because this question fell under the bot’s competence.
It’s quite naive to expect your profit to soar right after a bot’s implementation
You have to review all the business processes when rolling out a bot. Besides, most companies have a long sales cycle. In case a bot informs you of a new Smart TV feature, suggests you try it and activates it at the same time — its impacts on profit are doubtful.
And in case your transaction cycle lasts 3 to 4 months, you got to have a good understanding of your sales cycle, upsale process, and client support, before valuating a bot. It is vital to analyze the way you are planning to identify and track your client — for instance, the source may be recorded in a bot builder’s analytics systems.
You can’t blame a bot for a gone client when it made a good job of consulting and service enabling. A client might have met bad service, delivery or damaged goods. A bot is just a part of a team
We don’t think ahead
Once you go hard for a bot — you won’t be able to stop. Once you stop tuning your bot — six months later there will be no automation at all. Yes, machine learning algorithms work, but you can’t expect a bot would develop itself. And this tuning is necessary — any growing company releases new products and services; their names change as well as the target audience. Sales geography might change which means new topics for a bot appear, change and elaborate; and this is a process that can stop only when your business stops.
You got to shape a bot’s destiny in the context of your IT system development. For instance, today you are using one CRM system and in six months you are planning to change it or implement a billing — you got to think ahead. Because a bot should be retuned and integrated with a new platform, also, API should be created — all that takes time and resources.
Bot implementation is a full-scale IT project that got to be planned a few months ahead. That is why when looking for a bot, you have to have a good understanding of your customer’s journey, and of your own roadmap
