Show HN: Human Chatbot

6 points by txt-me 3 years ago | 8 comments

We are a small team bootstrapping a human chatbot. It is like a chatbot, but real people answer the chats. You install our widget on a website, we connect our team and immediately start answering, 24/7.

How does it work? It is cloud customer service. The operators are shared between the websites, hence lowering the price.

What happens if the operator doesn't know how to answer? We collect the visitor's contacts and create a ticket. You then reply to the ticket, and the system sends an email to the customer.

Would love feedback from HN!

jaclaz 3 years ago |

I am clearly missing something.

How would the agent(s) be capable of answering customers' questions?

Besides the "ordinary" and "very basic" ones like (say):

"Is the shop open next Sunday?"

"I am sorry, no, the shop is open monday to friday from 9:00 to 18:00."

And which kind of access should/would these operators have?

I mean (one of your examples):

"I am trying to pay, but my card is getting declined"

"I am sorry, let me check it with the bank"

What does (or can) the operator do, like:

1) actually check with the bank

2) open a ticket (or send a mail) about the issue

Only to show how old I am, once upon a time customer support personnel (telephone) was "in-house" (employees that knew about the company operations and could actually solve problems in a timely manner).

Then it was externalized to "call-centers", where operators were taught a small subset of the needed knowledge[1] and given very little power to resolve anything.

Then it was (often) moved abroad and further restricted in what they could do, but still there was the need to teach lots of things[2] to the operators.

If your operators assist for multiple companies, they will need to know quite a bit about each company, wouldn't they?

How do you plan to transfer the needed knowledge to them?

[1] AFAIK by means of (usually) quickly written "procedure manuals"

[2] I believe by means of FAQ and FGA collections

txt-me 3 years ago | |

That's an excellent point!

The operators are assigned to the clients, so even though they are not dedicated, they spend time replying to the same questions about the same businesses for weeks, months, and years. They learn.

From our practice working with SaaS, the tickets to chats ratio is from 10% to 20%. We handle about 80-90% of requests without bothering the main customer service team.

You still need in-house support, but without night and holiday shifts and only to handle the tickets, not conversations.

jaclaz 3 years ago | | |

>The operators are assigned to the clients, so even though they are not dedicated, they spend time replying to the same questions about the same businesses for weeks, months, and years. They learn.

I understand that, though the weeks/months/years to learn sound like a long time, but assumed that (through lessons, slides, procedure manuals, FAQ/FGA, whatever) the learning curve is shortened, my question was a bit different, it was more like about how much (globally) an operator should learn to be able to assist on many (how many?) different supported products/firms.

>From our practice working with SaaS, the tickets to chats ratio is from 10% to 20%. We handle about 80-90% of requests without bothering the main customer service team.

Good for you, to me it just doesn't sound right, those percentages seem more like those of an "info-line" than those of a support channel, probably it depends on the specific nature of your customers ( either the Saas is very much not documented or - for some reasons - the users of the Saas are particularly clueless and ask for assistance for a large number of trivial issues).

The availability 24/7, even if it is "nominal" (i.e. even in the case of a much larger percentage of chats that end in tickets and thus are resolved by the in-house support in two-three days time) seems to me like a nice feature to offer to customers.

astonfred 3 years ago |

Cool idea. $1 per hour makes it look very affordable (which it is, in a way) while generating a healthy $720 of potential MRR. Smartly priced. Can you share some numbers about the size of the current user base?

txt-me 3 years ago | |

Thank you for your nice words! Frankly, it is the first time we have made this feature public. We've been running it on some of our friendly projects and just reached the point when we have confidence in taking new clients.

IceMetalPunk 3 years ago |

So... by "human chatbot" what you mean is "outsourced call center without phones"?

txt-me 3 years ago | |

You may phrase it that way, yes! Although it's pretty strange to say that chat support is "calls without phones"

IceMetalPunk 3 years ago | | |

Well, yes, but a "call center" is a thing while I've never heard of a dedicated "chat center" :D