Monzo Staff Weekly Q&A - Bruno Vaz Moço (Product Manager, Scalers)

That’s a very good point! We recently added the section with suggested articles to the search flow (on Android only). You can try tapping the search bar on Android, and you’ll see the same article suggestions there.

This means we can direct more people to search and still have the power of the article suggestions. Another change to this suggested articles section is that we’ll soon prioritise them with a machine learning model. @Neal is the data scientist behind the new model, and offline results suggest the articles picked up by the ML model are much better than the current implementation, which is based on binary rules that we manage manually in the team.

3 Likes

Do you find that people interact less with items such as Spotlights? which are great at pointing out new features etc?

We should have probably saved these for you Q&A later :see_no_evil:

2 Likes

Yep! Feel free to post these in the Q&A thread, I’ll be answering them today/tomorrow :+1:

1 Like

I’ve moved some posts here from this thread: Helping Every Customer: Behind the scenes of the updated Help tab! as the answers would fit in nicely to this weeks Q&A :slight_smile: Hope y’all don’t mind.

3 Likes

Many of the challenges we face are definitely shared with many other companies, not just high street banks. I think the biggest difference is the constraints we add on top.

We believe customer support is one of the most important differentiators of our service. The process of scaling up the COps Team (Customer Operations Team) is not a new challenge but we’re committed to doing so without compromising the level of service we currently provide which significantly changes how we can do things. If you’ve ever chatted with our lovely COps Team in past you might have felt it was a refreshing experience, especially when compared to other high street banks. Why should this experience be any different once we have 500 million users?

There’s also challenges on the daily life of people working in the COps Team. Since we offer real-time human customer support 24/7 we need to guarantee there’s always enough people working at any hour in the day to cover the number of new conversations. It becomes more complex when you factor in the wellbeing of the squads (each Team at Monzo has multiple Squads), for instance it can be quite challenging to work night shifts. Making sure people have the flexibility to take time off whenever they want, or even the ability to tell your colleagues you’re having a bad day and need to go home. These things are easy to do in small companies, but become a challenge once we have hundreds of people working in the same team and don’t want to compromise on quality.

Regarding the challenges high street banks face that we don’t, there’s many (not just from a Scalers perspective). If you think about the tech infrastructure they’re using it is much more difficult for them to quickly iterate their products, or even know how customers are interacting with them. In the Scalers Reductivity Squad we constantly tweak the Help section of the app. On a daily basis I check how users are interacting with the new features. The amount of (passive) feedback we get is critical for us to continuously improve the app. And we have this data stream in near real time (15 min lag). High street banks often times need to operate without this feedback loop which makes things much more difficult for them. More than this, the way our product teams are structured allow us to deploy new code to production on a daily basis, this can be a new machine learning model, a new user experiment, a new app flow, or new app content. It’s a luxury high street banks might not have and has a massive impact for users.

High street banks are also structured in a very different way. Their customer support teams are not as close to their product teams as they do at Monzo. For instance all of the Scalers team sits physically in the same space as all the COps team in London. This allows us to react very fast to the issues our users are facing. Everyone in the COps team can add outage banners to the apps, or add content to the help screen that appears on every user’s app (literally just typing on their laptops and hitting publish). We also involve people from the COps team in many of the projects we do, this helps us in two ways: product teams get first hand insight of what works and doens’t work for users, and the COps team is much more aware of the changes we’re shipping and how they work.

9 Likes

Incredible reply! :blue_heart: :mondo:

1 Like

Pineapple on pizza, yes. Cats, they’re ok.

We’re currently building a new help screen for new users which is better at helping users that just opened an account. This changes our approach to help significantly and it’s a hypothesis we’re testing… Up until now we’ve been building the help screen with very powerful features like machine learning search, article suggestions, etc. More recently we’ve given the search bar more space on the help homepage, this works really well for the majority of users. But an empty search bar can also be quite scary.

That’s why the new help screen for new users takes the approach of “help on rails”. Instead of giving you all the different options to find the content you want, we make it more prescribing by placing the content we think is most relevant for new users more accessible (after analysing thousands of conversations from new users). We don’t remove any of the existing functionality, but we simplify what sits on the homepage.

I’m really excited about this change for two reasons. 1) We’re ramping up user growth in the coming months, and new users are twice as likely to get in touch than the rest of our user base. If we suddenly experienced explosive growth the COps team would be under significant strain to keep response times acceptable. So this is a project that can have a real positive impact in the COps team. 2) We’re testing a hypotethis that we haven’t before, and it’s change of mindset in how we approach help, so I’m very excited about that as well.

We’re shipping this on Android as a user experiment next week. Only new users signing up will be allocated on the experiment, and some of them might have a slightly different version than the screen below.

image

6 Likes

What’s your opinion on bees?

I love bees.

1 Like

Looks really good thank you for the reply!!

Also, sure there is some Joke around the bees, but I love bees… Im from Manchester :smiley: So even got a bee tattoo on my wrist!

1 Like

Great question! Our priority is making sure users have access to their money. An ‘incident’ can range from feed items not being created in real time, to a full blown outage where all payments decline. We have a few things we can do internally to minimise user impact, depending on the severity of the incident. What we can do also depends very much if it’s an internal incident or from another company that we interact with (like Mastercard or other banks).

Ultimately if there’s something impacting all of our user base, and they can’t access their money for whatever reason, we would consider sending a message with a push notification to all users letting them know what’s happening, so they’re not caught off guard. It might seem like an overkill to many users, but people will be sitting in a restaurant without being able to pay their bill, or at the supermarket till with a trolley full of groceries (and might not even have food at home). So first line of action is being transparent about what’s happening and keeping users looped in while the engineering teams resolve the issue.

Internally we would immediately start publishing overtime shifts that anyone can pick up and jump straight to the conversation queues, and for the following days if we expect the issue to last longer. Being a customer centric company everyone at Monzo is also trained on how to use our internal tools and answer customer queries. So if things are really bad, we’d ask other teams to jump on the customer support queue.

In less serious incidents we have other tools we’d use to minimise user impact, or COps impact (from the additional influx of queries). Something very straightforward to do is just adding new content to the help screen. ‘Emergency’ content appears at the top of the suggested articles (below the search bar), also appears on the top of the search results, and also in the screen where users first type their message as one of the ‘suggestions’. This is very effective at answering questions that do not impact user’s money, but might create confusion in app for some reason (like a duplicate feed item, or a slightly delayed bank transfer). These are the issues that there is no action from our COps team except explaining what’s happening, which the help screen is also effective at doing.

We also use StatusPage to publish incidents. If we add an incident on StatusPage users that signed up for SMS notifications on their website would be notified (albeit it’s a very small number of people). StatusPage also allows us to set a severity, and depending on what we choose it can automatically add an alert banner throughout the Monzo app, with the description of what’s happening, and enable an alert box that would appear to anyone that tries to contact us during the incident (to tell users we’re aware of the issue and they don’t need to get in touch, but they can still do it).

What’s really interesting is that from each incident we do a “post-mortem” to discuss how we could have dealt with it better. More often than not we realise we could change our systems to stop some of the issues before they affect all of our users. As we see more of these we become better at preventing large scale incidents, and at minimising user and COps impact.

4 Likes

A month or so ago we gave a bit more space to the search bar in the help homepage. Everything that used to be on the help screen homepage was still there, but when you first landed on that screen you would see search front and centre.

This was done as an experiment and had a reduction of 12% in user conversations. We didn’t prevent anyone from getting in touch, but we nudged them to try searching first, and it this was very good at helping users solve their problem faster. So this is now standard for all users on both iOS and Android.

Transaction related queries represent 10-15% of all our conversations. We’ve done a few things in the past to help users resolve their transaction queries without needing to get in touch, and I don’t think we’ve nailed this yet.

These type of queries can be very different from one another, and can be difficult for users to self service a resolution. The vast majority of transaction questions don’t actually need any action if users are okay with waiting a few days, as they almost always sort themselves out with time. But naturally this sometimes affect the available balance shown in the app (although the actual balance of the account is right), so even when users understand what’s happening, sometimes they send a message “just to be sure”. We will continue to work on this problem until we get it right.

4 Likes

I think this answers you question:

2 Likes

The most rewarding thing of being a product manager in the Scalers team is seeing the impact we can have on the COps Team (Customer Operations). It’s a team of nearly 200 people, and we have the opportunity to make their daily work more enjoyable, and stress free. It’s a constant battle between a fast growing user base (more user conversations coming through) and the Reductivity squad trying to minimise the number of questions we receive. If the speed of these two different dynamics gets out of sync the COps team is the first to feel the pain of user growth.

I also believe we should focus the time of the COps team in solving the wicked problems our customers face. If someone loses their job and suddenly can’t pay their overdraft, this can create a lot of anxiety to peoples lives, and the COps team does an amazing job in dealing with these situations. Or if someone gets robbed in a random country on the other side of the world and is left without a phone or a wallet, at this time we should dedicate as much time as necessary in solving these problems, and going above and beyond to make sure our users can carry on with their lives without needing to worry about money.

Solving the wicked problems, the problems that require high touch high empathy, makes the daily life of the COps team more challenging, and more enjoyable. It’s very rewarding as a pm to have an impact in this.

Right now we still get many questions that don’t necessarily need to be resolved by the COps team. And we can prioritise work in the Reductivity squad in the Scalers Team to make sure users don’t need to get in touch in these situations. Part of our team’s mission is to make sure users don’t need to get in touch, but can still do so at any time! So if someone prefers to chat with us because their moving house and need to update some of their account details the COps team is here to resolve this through chat, but many people that write in about these things would prefer to do it themselves in the app.

Ultimately as a company we want help over 1 billion users to not having to worry about their money. Making sure we can reach that goal while still providing amazing human interaction to our users is what makes me come to work every day.

Prioritisation is always the most difficult thing. We can do many different things, but prioritising work to make sure we get the maximum impact on the shortest period of time is a challenge. Most all of our work will have interdependencies between multiple domains (design, backend engineering, iOS/Android engineering, user research, content, operations, data, etc) so making all of this make sense for the entire team on a day-to-day basis requires a lot of planning.

4 Likes

There’s no single trick to make this work, reducing number of conversations is a consequence of many different efforts, thousands small changes that collectively help us move towards that goal.

We’re a very data driven team and have a very good picture of what users write to us about, what’s the response we give to users, and what actions were taken on a user’s account (if any). We can see this at an aggregate level over time with some real magic from the Data team. We use this information as a guiding star to our roadmap.

Simply put we pick the most common problem users have, and make whatever change we need to in the app to help that group of users. We do this every week, and keeping a laser focus on that list has worked very well for us in the past - it’s our Hit List.

Some problems are easy to solve, like updating your phone number, we just need to allow users to do this themselves in the app and that will solve 80% of the conversations about this (we’re shipping this on Android next week!). The other 20% of users is more difficult to solve, the flows are never perfect or 100% discoverable, and it requires a lot of work to fix these problems for the long tail.

3 Likes

How do you plan to do that? New golden tickets? New “network-based” features? Can you give any examples?

1 Like

As far as I know it’s not a term we picked up from an industry standard. It simply means fixing all of the problems that can block our ability to grow our user base, from a customer support perspective. So it’s the “COps Scaling” team if you’d like. Also, I think it sounds cool :slight_smile:

It’s a balance between the two, and it depends on where we are at any time. For example when we started rolling out the current account last year, there were a few issues that created a big strain in the COps team. At one point mostly everyone that opened an account was getting in touch, that was a time that we needed to be reactive.

In the past 6 months we’ve halved the number of conversations that come through, which buys us some time to be proactive in solving issues that would otherwise not be possible to prioritise.

2 Likes

We need to make sure the basic flows for self-service are available. In the past months we’ve done a lot of work in this and mostly everything can be self-served. Then the question is making sure users can find those flows when they need them.

There’s a lot of work in this area, and we’ve covered some of this in previous blog posts. Things like machine learning to predict what you’re going to ask, dynamic article suggestions, better discoverability, etc.

Have a read at the blog post @Neal posted this week explaining the role the Data Team plays in this process. https://monzo.com/blog/2018/08/01/data-help/

2 Likes

Spot on! Bringing more network based dynamics into the Monzo app is one of the ways we’re going to dial up on growth. These can be golden tickets (referral leaderboards, rewards for inviting more friends, etc) or can also be features that makes you want to have your friends on Monzo, like being able to split your flat bills with your flatmates in an elegant way).

The core Product team is focusing on growth and there are 2 different squads focusing just on this (not in Scalers).

1 Like

Cool. Are there any other examples, and do you have a daily new account target in mind?