Slow support responses

(Scott) #124

That makes sense so seems like we’ve got a way to go before we get back to good old Monzo support then :-1:

(Hugh Wells) #125

This is the things that’s really hard to scale :disappointed: It’s all very well saying we’re hiring at 7 people per week, but there’s a week of training and it really does take a couple of months for people to get their feet under the table and feel comfortable in the tooling and knowing how to deal with things.

(Andy) #126

There’s only so many support tickets you can look at in a weeks training. Can’t cover everything and as you say for many people some stuff just doesn’t sink in until you start doing the job live.

That’s before you even get close to some of the obscure questions customers ask or some of the one off issues that happen like what happens when a switch fails but the balance transfer has happened etc

(Scott) #127

I understand that your recruiting a lot more people and of course training and general learning the ropes is in order, surely you must have an urgent number or way whereby an experienced agent can deal with the query? I’ve always loved how Monzo works and your values but like I’ve said customer support has to be key specially when your an app only based bank so us (your customers) need the reassurance that you are there for us when we need you most :innocent:

(Neil M) #128

I think working on your support model would help. From the limited experience of the project I’ve worked on at the beginning we had a hands on approach for the early adopters and then focussed on self help and documentation.
This is how you should scale, majority of the ‘easy’ issues should be solvable in App etc without having a talk with a person (tier 0). -takes production focus rather than feature development. Potentially a super user capability but obviously GDPR etc concerns
Anything that’s not should be escalated to a COp with limited experience( tier 1). Trainability is the most important feature to look for while scaling. Risk taking and thinking outside the box
Anything that can’t be solved with them should he escalated to an experienced COp(tier 2). The urgent flag should be reworded to important or a different word (everything is urgent if you leave it long enough). Tagging system (literally takes to seconds to figure out if it’s urgent i.e fraud etc or alternatively just a can’t find it)
Tier 3 - escalate to back end if serious and can effect multiple users ( outages etc)


I’ve now been waiting over a day for a response from support. It’s fine; my query’s nothing major. But if Monzo is keen on being transparent, then it should tell us how long we should expect to wait before we send a message. From a service perspective, not knowing when my request is going to be answered is… well, it’s very legacy bank.

(Nick) #130

First thing you see on the ‘Help’ tab is the question “How can we help you today?” and a search bar. Tap on the search bar, and even before you start typing, the most common queries are suggested.

I’m not sure what more Monzo can do at this point to make the ‘easy issues solvable without talking to a person.’

I’m hopeful that the urgent flag is a short-term hack, and that Monzo are still working on machine learning to identify urgent queries without intervention, as they have spoken of wanting to do in the past.

I don’t think there’s that much wrong with Monzo’s basic model at the moment, the problem is down to there being a collective “eyes taken off the ball” moment that allowed user growth to outstrip COp resourcing, and the main limiting factor is getting new COps onboarded, trained, and confident.


Isnt all this because their original plan was to one day have 1 op for like 120k customers or something…and the reality of it is, it doesn’t work (yet…)

(Ryan) #132

I still don’t understand why so many people need to contact support in the first place. Could monzo provide some insight into what the most common queries are?

@anon40728597 I get the impression that monzo and starling made the assumption that nobody would need to contact support as everything would just work but the complete opposite has happened.


Yeah I agree .

Because chat is so easy…sometimes you ask something you otherwise might not bother…


Probably going to be an unpopular idea (not sure I’m that keen myself :laughing:) , but maybe it should be more “difficult” to contact support? Using Amazon as an example, there’s usually a number of self-service steps you have to go before you can initiate a conversation/call. I know Monzo kinda do this already, but it’s still really easy to start a chat. (can’t believe I’m implying that’s a bad thing :upside_down_face:)

Again, using Amazon as an example they also offer a callback function, which is really useful. Surely that would be better than an automated message and hang-up?

(Andy) #135

There’s loads that Monzo could be doing to deflect chats - changing how easy it is to chat, improved help articles, using chat bots with AI and ML, robotics to automatically move chats through queues and processes


That would be a big no from me, I’m afraid.

I’d rather Monzo worked on the reasons why people contact support - and automate things behind the scenes rather than give up on being able to chat straight with a human. :frowning:

(Noel Edmonds Beard Sculptor ) #137

This thread and my others on here like it about how slow support is and the phones not being answered really now do help to verify the trustpilot reviews and even though we know most of them are fake it is not helping one bit.

(Hugh Wells) #139

In our case it was not hiring enough people sooner as our data models were a little off :disappointed: We’re going to be going into more detail about this in a blog soon but we are hiring at a significant rate right now :pray:

I think is certainly the approach we’d rather take :slight_smile: We recently (a few months ago I think) rolled out some improvements to the way we surface help articles in the app. We also added a self service option for authorisation reversals (ie. when the terminal times out but we still saw the transaction succeed and you end up paying twice! :see_no_evil:) which at the time was a pretty significant amount of inbound demand. By adding the self service feature we not only reduce the number of people needing to talk to us but we also make it easier and quicker for customers :pray:

Sure, I will check for you on Friday (assuming I can log into the data dashboard by then!)

That’s a really good point - I will pass this back :pray:

(Alex Sherwood) #140

I totally get that this is really hard to manage - that’s what makes it such an interesting challenge :smile:

According to an a blog post from a few months ago, a single COp can manage about 3k users each. Which means that you’re hiring enough people to manage about 21k users per week. That would be about right, since Monzo’s growing by about 20k users per week at the moment.

But that’s assuming that:

  • Efficiency hasn’t improved - hopefully the tooling & training’s got better since then
  • You’re already able to increase response times now by routing conversations to specialists (presumably if you weren’t, that would be slowing you down at the moment)
  • The switch from Intercom to Monzo Chat isn’t impacting response times
  • The number of COps per user hadn’t already fallen behind (based on the current response times, it seems like it had)
  • The new COps reach their full potential straight away - as you said, it’s not realistic to expect that they will
  • You continue ramping up hiring as Monzo’s user growth per week continues to increase & ideally (perhaps) begin ‘overhiring’ in anticipation of that growth

And then the more interesting question for me is: whether this is having a significant impact on user’s satisfaction - it looks ‘bad’ to me that I have to wait 3 days to report a bug but since that’s not urgent, it doesn’t upset me & response times for urgent queries are still great :sweat_smile:

TL;DR - I’m looking forward to a deeper dive into what’s happened here & how Monzo feels this has gone later on :nerd_face:

(Hugh Wells) #141

I think this is the key bit :wink: We’re not out of the woods yet and it’s hard to predict where things will go, so I think I’ll be reserving judgement until then!

The tl;dr for this is no, broadly. Again - this is something that will probably be covered in a future blog post but when I chatted to @liamh in September about this the stats were very much showing that:

  • When urgent queries were being answered within 10 minutes (our SLA)
  • For normal queries when we did respond; we kept responding relatively quickly and resolved the issue swiftly
  • Normal queries weren’t waiting excessively long

CSAT (Customer Satisfaction - the set of emojis you get at the end of a conversation) remained the same/similar.

The problems came when, for example, someone contacted us and waited say a day for a response. We would send a follow up and snooze the conversation. Invariably, some customers would reply when that COp had disappeared and thus the conversation would be sent to the back of the queue and the customer would have another 1 day wait… Obviously this was not ideal (The customer’s total wait time becomes unacceptable and we end up with queries going round and round, not getting resolved). We now have a Reassign queue where conversations waiting on a customer to respond are snoozed and then automatically reassigned (like urgent conversations) to the next available COp, bypassing the backlog.
This used not to be a problem (when we were burning through the queue pretty much matching demand) but when it became one, we fixed the issue :+1:

I think maybe I can add some clarity to this! :sweat_smile: Routing to specialists is done on an escalation basis - the frontline COp assess that something needs some extra :eyes: (be that for a weird payment issue, a customer who might need extra support or just to confirm procedure) and escalates it via the neat button which dumps a link and some contextual information in an escalation Slack channel. We have a general (#cops-queries) questions channel, #cops-troublshooting (for technical issues/bugs), #cops-txn-disputes (for Mastercard chargebacks/fraud) etc. A scheduled floater will look at the escalation and follow up with the COp to get any further information, take on the conversation if appropriate and get the problem resolved :raised_hands: This frees up the COp to tackle more queries and means we have those we extra context and expert knowledge looking at the trickiest of stuff. Most things are dealt with on the frontline though as we give each COp the training, support and trust to solve issues and delight customers :smiley:

Floaters are still COps, they have just done extra training and are “accredited” via a “badge” system so they have more knowledge on a certain subject. COps can (and do!) apply for badges and you can have multiple :slight_smile: Badged COps also do other “back office tasks”; for example Disputes COps raise chargebacks and deal with 2nd presentments/arbitration chargebacks - a totally asynchronous function :+1:

(Jonathon) #142

Slightly concerning. So if someone leaves then my query is stuck until they return from leave?

(Hugh Wells) #143

Not by design - that is an unfortunate human error :disappointed_relieved: I am really sorry :pray:

These type of things do get picked up in Quality Assurance and fed back :slight_smile:

(Jonathon) #144

That’s good to hear :+1:t2:

Also she was lovely so this was nothing against the specific member of staff :slight_smile: