Onboarding is a fashionable topic at the moment. You see lots of onboarding articles and tear-downs shared these days. But onboarding is just the first date. If someone installs your app or registers on your site, you’re not guaranteed a long term relationship — far from it. One study quoted by Luke Wroblewski, paints a pretty clear picture:
“26% of all apps downloaded were opened only once and then never used again (…) 13% are opened only twice, 9% are opened only three times.”
So, what happens after onboarding is crucially important, and keeping a user actively involved in a long term relationship is much harder than impressing them on a first date.
Encouraging users to progress from first-timer to loyal advocate is something I call “Ramp Up”. It’s a really useful concept for designers to think about, and something we don’t talk about enough. Before we start, let’s take a look at some of the things that happen when you do it wrong.
The abandonment antipattern
Up until just recently, Evernote’s web app was the perfect example of the abandonment antipattern. When the user finished the registration process, this is where they ended up:
After registering on the website, Evernote would show users this blank page. Given the depth and complexity of the Evernote ecosystem, this is like dumping a person in the middle of a desert with no survival training. It’s abandonment.
The front-loading anti-pattern
In front-loading, the user is forced through a short tutorial when they install the app. Most mobile apps make this mistake.
The problem with front-loading walkthroughs is that they try to teach by telling, like a teacher in a 19th Century classroom. They don’t actually engage the user in any of the actions and help them through the steps. Also the timing is totally off — after installation, most users just want to try out the new app, not read a bunch of instructions.
The Zone of Proximal Development
In psychology, there’s a concept called the Zone of Proximal Development (ZPD), which explains that learning is helped when you’re fed the right level of complexity for your current level of development, and that learning is blocked if the materials are either too easy for you or out of your reach.
It’s a simple enough idea, but it’s non-trivial to implement in a piece of software — you have to quantify the difficulty of the actions you want the user to complete, and you have to track the user’s progress through them.
In modern educational frameworks, it’s common to break down subjects into small learning objectives, and to arrange them into a sequential order. For example Pearson have a framework called the Global Scale of English. The scale consists of thousands of objectives that lead in a ladder from complete beginner all the way up to fluency. For example, a learner is taught how to recognise the letters of the alphabet before they’re taught how to read and understand prices. One of the applications of the framework is in eLearning platforms, where a decision-making engine (much like a Ramp Up Management System) works out what to show a learner next — something that stretches them, but is still within their reach.
I’m sure you can see what I’m getting at here. Of course, the difference is that if you’re not in the education sector, your users won’t see themselves as learners. In most cases, users expect to become proficient at using your service without ever having to stop and think. Helping users Ramp Up at just the right time and in just the right context is really hard.
It’s about persuasion as well as learning
Here’s the other thing. We regularly need consumers to just do us a favour with the hope that they’ll gain some value in return. Review our app. Check out our marketing campaign. Import contacts. Upgrade your account. And so on. We need them to do this stuff to be successful, and they’ll do it if the relationship capital is sufficient — if trust has been earned. If you ask them too soon or at the wrong time, it’s a big turn-off.
There are various models of persuasion out there, with BJ Fogg’s BMAT model being one of the most well known. The principle is simple — a user won’t bother doing something if they don’t simultaneously experience the desire to do so (motivation), the means to do so (ability), and awareness of the call to action (a trigger).
Twitter’s Ramp Up Management System
Twitter‘s approach involves injecting cards into your feed according to your behavioural profile. When you follow a low number of users, Twitter injects more “follow” cards. Once you start following a high number of users, Twitter injects different Ramp Up cards, as shown below.
The Google Analytics Ramp Up Management System
Google Analytics uses a straightforward notification panel with an inbox. Their Ramp Up Management System appears to work pretty much like Twitter’s — they monitor user behaviour, they have a decision-making engine that works out what notification to show, and they track which actions are completed, to make way for new ones. Their notifications appear to mainly consist of configuration actions, diagnostic warnings and links to educational resources.
Since Google Analytics is a workplace tool, they don’t bother taking any steps to “mix the medicine in with the food” in the way consumer apps do. It’s important to recognise their approach as being suitable for a specific niche — people pay for Google analytics training and certification in order to then be paid as an expert to use it. Users expect a bit of hard work and pain as part of the journey, which is what makes their formal, no-nonsense notification approach palatable for users. If you’re working on a consumer product focused on enjoyment, where the “journey is the destination”, you’d do well to steer clear from such a direct approach.
Clippy: why timing and pacing is crucial
It’s easy to be dismissive of Clippy, but there’s actually an interesting story to tell. Clippy’s inventor, Eric Horvitz, originally gave it decision-making engine that was powered by a Bayesian network, and this engine was very restrained in triggering Clippy to appear on-screen. In other words the original, pre-release Clippy was reasonably smart and respectful towards users.
As the story goes, the Microsoft management team were disappointed with the idea of hiding Clippy away so they stripped out the Bayesian decision-making engine and replaced it with a crude system that made him appear on-screen a great deal.
Clippy failed for a number of reasons, but this lesson stands out — one of the most important things about any Ramp Up strategy is knowing when and where to show users Ramp Up suggestions. If you have a crude Ramp Up system, you will probably get it wrong and end up being “Clippy-level” annoying. Timing and pacing is crucial.
Invasive notifications and disrespectful design
In Mike Ryan’s article Don’t Be A Notification Bully, Mike talks about invasive notifications such as “The Ruse”, where you’re tricked into visiting a sales page, “The Gang” where you’re hit with lots of different notifications and no means to opt out from them; and “The Passive Aggressive” where weasel wording tries to make you feel bad for not doing the thing you’re being asked to do. In some cases they spill over from just being pushy into full-on dark patterns.
There’s a lot in common between Clippy’s failings and the failings of the Ramp Up Management Systems in today’s apps, even though they are decades apart.
Anatomy of a unified Ramp Up Management System
When I worked as UX Lead on The Telegraph’s digital edition app a few years ago, I worked out that we had about a dozen different types of call-to-action to show users, depending on whether they were on a free trial or an extended free trial, whether they were a paying subscriber, whether they’d yet swiped from page to page, whether they’d written a review, turned on iOS notifications, and so on.
The app was at risk of becoming a horrific nag, even though all the user research showed that users wanted a refined and grown-up reading experience. They literally just wanted to read the paper in peace. So I came up with the idea of a Ramp Up Management System. Most of the logic resided in a decision table I called the “nudge matrix”, which began life as a simple spreadsheet. Triggers were listed in one column, calls-to-action in the next, and parameters in subsequent columns. A simplified version of it is shown below:
The idea was for the system to analyse each row in a sequential order. When a row was triggered, it caused the associated call to-action to be displayed. I’m sure there are smarter ways of doing it, but this worked for us, and had the bonus of being easy to understand, allowing all the different departments involved to contribute without messing it up.
In the end the Ramp Up Management System was implemented with two decision tables, one for infobar calls-to-action and one for modals, and great care had to be taken with the logic to make sure it didn’t get inappropriately stuck on any of the rows.
If you’re familiar with marketing automation or advertising platforms, you might be wondering what’s new here. Those industries have had this sort of functionality for years. For example, take a look at Salesforce’s customer journey builder. It‘s pretty neat. But the fact is, they use it very differently to the way a product or service designer would. In marketing it’s mainly about sales and discount codes, and in advertising it’s about following you around the web with creepy ads. There’s so much more you can do when you’re motivated by a honest desire to make users’ lives better.