When Octopus reply to emails and miss the point - is it a Bot reply?

Mum161111

I'm just wondering about this and will be interested in any of your experiences or knowledge.

When Octopus reply to an email and their reply is incomplete, or some questions have been ignored, is it possible that a human has not replied?  Is there any way of knowing?

It is disappointing when they reply: "Thank you for reaching out to us ...."   then fail to answer the questions asked and/or give an inappropriate or incomplete answer.

Also, in the context of Energy, does anyone know when it will be possible to apply to Octopus for The Warm Home Discount,  Broader Group Scotland, please?  A helpful poster here recently advised that British Gas have opened their applications for The Broader Group Scotland which is different from Core Group and the England/Wales Warm Home Discount system.   Thank you.

Spoonie_Turtle

Yes those replies could very well be bots.  It was known a few years ago they were trying AI customer assistance to help. 
My experience of it has been that it's useless and takes me more energy having to respond and explain why the initial reply didn't address my issue.  One of the bot responses told me about doing the thing in the website, when the reason I contacted them in the first place was that the website didn't have that function and told me to e-mail (which I explained in my first contact) *facepalm*

WHD applications open 'later September' https://octopus.energy/blog/warm-home-discount/#scotland

Mum161111

Thanks very much for the link Spoonie_Turtle.

I  suspected that the reply was not from a human.   It's difficult to know what to do about it - especially when the reply has not related to the questions asked.  It seems to be such a waste of time.

I don't suppose asking, from the outset, for a reply from a human would help.  I don't  suppose Octopus would acknowledge that the reply was from a Bot - AI customer assistance. 

I really appreciate your reply, thank you.

zxzxzx

The issue was they grew too quickly to employ and train customer service staff so used chat gpt as a quick fix. There is an interview with the MD on line somewhere to that effect if you want to search.

Then found it a useful ‘first pass’ and continued it.

https://sifted.eu/articles/octopus-energy-ceo-chatgpt-news

Mum161111

That's an interesting link, zxzxzx.    It would have been helpful if I had known this much earlier.  I'm sorry I missed that CEO link.  I am surprised that the AI replies were supervised by a person but 80% satisfaction is quite good. 

Although I still prefer 100% human, all the time, I suppose we have to move with the times and accept some limitations.

MattMattMattUK

Mum161111 said:

Thanks very much for the link Spoonie_Turtle.

I  suspected that the reply was not from a human.   It's difficult to know what to do about it - especially when the reply has not related to the questions asked.  It seems to be such a waste of time.

Those customer service chatbots can be pretty good when responding to a single question about something rather mundane which encompasses the vast majority of customer service interactions. They do also sometimes struggle with multiple questions in the same email, or an email that is not written in a clear and concise way, using the correct terminology etc. (not saying you were not clear and concise, just that that can be the cast).

Mum161111 said:

I don't suppose asking, from the outset, for a reply from a human would help. 

No, it would not, but my understanding is that with Octopus any subsequent replies go to a real human, the first line is AI because that takes care of 90%+ of basic customer service queries, leaving the real humans to deal with the more complicated issues. If you send a reply I think you will be replied to by a real human.

Mum161111 said:

I don't  suppose Octopus would acknowledge that the reply was from a Bot - AI customer assistance. 

They will/do, they have published blog articles about using chatbots as their first line of CS and show examples of how they are able to deal with most issues and that it subsequently frees up real humans to deal with more complicated or non-standard issues. 

Mum161111

Thanks for explaining, MattMattMattUK.  On reflection I'm sure, one of the times at least, I did ask several questions.

If it happens again it would seem reasonable to reply with exactly the same questions but in the first sentence put something like:  "I'm sorry you did not seem to answer my questions and hope you can help me with my questions : .... then copy the earlier email questions??

DullGreyGuy

Mum161111 said:

Thanks for explaining, MattMattMattUK.  On reflection I'm sure, one of the times at least, I did ask several questions.

If it happens again it would be reasonable to reply with exactly the same questions but in the first sentence put something like:  "I'm sorry you did not seem to answer my questions and hope you can help me: .... then copy the earlier email questions??

Having done a lot of customer service before the days of AI and bots with some companies there isn't much difference between a bot and a human because of how systems worked and the targets people were set. I mean to give you an example, working for BT Directory Enquiries your average call length had to be under 6 seconds... 6 seconds for the customer to say what they wanted, you to type it into a system, ask clarifying questions, then select the right option of the number (the system then takes over and tells the person the number). At all times there were calls queuing and as soon as you select the number for your current customer the next customer is talking to you (the phone does the salutation for you). 

Letters/emails weren't much better, to reply to any customer query etc in 1st party claims when dealing with insurance the operator first had to choose the most appropriate overall letter topic (fire, theft, accident new claim, update, total loss, indemnity query etc) and then select pre-written paragraphs. In some cases you had 4 lines at the bottom you could free text something into but in many cases you didnt. Pressure was less as you had about 3 minutes average handling time to achieve rather than 6 seconds but sometimes you had to go off to find a paper file which may be in the filing cabinet or may be in the post tray if it has new correspondence in it. Similarly some people ask multiple questions covered by different templates so either you 1) wrote more than one reply 2) tried to shoehorn into one using the 4 lines to answer the extra bits 3) dictate a completely bespoke letter 4) answer what you could on a template letter and hope they forget their extra questions. 

These things aren't good customer service but having vast teams of customer service people with office space to house them, managers, auditors etc etc to oversee them is very expensive and for many price is king. Before the idea of real robots taking over these roles many customer service people said they felt like bots. 

Octopus has been open about using scripts to reply, with many others the bots could easily be human or AI.

Mum161111

Thanks, DullGreyGuy.  I understand. 

I think my concern/ disappointment is because I've been with Octopus for a while snd was pleased and accustomed  to their excellent customer help.  I think it has chsnged but is probably still one of the best Energy Providers we have.

Your detailed reply is much appreciated, thank you.

Lorian

I've only recently joined. I asked them to take one specific action in email and it was carried out. An action AI would not be able to do. Did take them a while to respond however. If it's like the system I use at work AI assess if it can answer or not and if the score is too low the ticket goes straight to an agent.

Gerry1

Part of the problem may be that Octopus inherited customer service teams from others such as  Shell Energy and failed companies for whom they were appointed as the SoLR.