How I Would Improve Pocket

(Hint: it has nothing to do with data products and machine learning)

Last week, I sent my 10-page plan for improving Pocket to it’s parent company Mozilla, and all I got was a robot response. 

The irony! My plan for revamping reading in Pocket was a reply to their recently posted vacancy for a Product Manager data products and machine learning. And then an algorithm promptly rejected me. Ha!

For context: Pocket is basically a web extension & app that allows you to store online articles for when you have time to read them. I love Pocket and use it all the time. So when I saw they were hiring a Senior Product manager data products & machine learning, I just couldn’t stay silent. 

Because I love Pocket and I think it can be a lot better than it is, right now – but the data products are definitely not where I would start. Data products and recommendation engines and algorithms are exactly what’s wrong with most online ‘destinations’ today. Not to mention that Pocket is decidedly not a destination, at least not for me, mostly because the reading experience is pretty poor. I read a lot in Pocket, and I’m grateful it exists, but the actual reading is not great.

So if Pocket’s going to focus on anything, in this devoted fan’s mind, it should be improving the core product, not discovery.  Not yet, at least.

Seeing how I love Pocket so much, I thought I’d write my ideas down and send them to Pocket, using their hiring process as a way to get their attention. But then the robot recruiter shot me down. 

The project plan was a labor of love, and took me a couple of evenings. And now nobody at Pocket is ever going to read it. So as to not let fun writing go to waste, I thought I’d share my product plan here instead. Sure it’s a long read, but hey, just save it to Pocket for later!

Executive summary: Pocket saves democracy

Pocket has an excellent brand and good product market fit. But it needs to double down on improving the core reading and sharing experiences, and its monetization efforts. Pocket can position itself as the open web alternative to the tech giants’ walled content gardens by enabling writing and commentary to flourish alongside reading. 

Machine learning will have a role in improving the discovery experience, but only once the crucial improvements in reading, sharing, and monetization have been implemented. And then only as a tool to achieve a goal: to help position Pocket as a destination. A Pocket focused on enabling better reading (and writing) more, will become a place where you come to think. This in contrast to the big content aggregators, where the public is held captive by angry content so they’ll see more ads. 

Below is my 12-month plan for achieving those objectives. After that, I will explain why these changes are needed. 

The Pocket 12-month plan

Were I a product manager, in my first 12 months I’d aim to implement the following: 

3 months6 months9 months12 months
Improve SavingLimit storing to 25 articles for non-subscribed users. All items older than the most recently saved 25 articles will be ‘archived’: they disappear from the reader’s queue, but will become available again immediately after subscribing.
Improve ReadingCreate a stellar ‘article reading’ screen for both subscribed and non-subscribed users, with the help of designers and focus groups.Improve the highlighting functionality and create a stellar note-taking and writing feature for all users. * Start a marketing campaign about ‘making time to read’ and putting away our phones more often. * Display Kobo notes in the Pocket app.  
Improve SharingCreate a ‘Digest’ feature that allows tier 2 subscribed users (see Improve Monetization, 9 months) to curate lists of articles and commentary that can be shared via Pocket, link, RSS and/or a newsletter. 
Improve DiscoveryIntegrate relevant Digest lists and commentary into people’s ‘Discover’ tab.
Improve MonetizationIntroduce new subscription pricing and USPs:* Pocket: for $2,- a month, subscribers can store an infinite amount of articles, and can download that database at any time.Start a brand advertising initiative (non-programmatically) to monetize non-subscribed users.Introduce new subscription tier:* Pocket Digest: for $4,- a month, subscribers get all the benefits from the normal Pocket subscription, and can also use the new ‘Digest’ features.Start an affiliate program selling Kobo ereaders, as part of a campaign to get people to read more and get away from their phones.

Why are these changes needed? Why not start with Discovery and machine learning immediately? I think the Pocket core product suffers from shortcomings that an improved Discovery tab won’t fix. 

I will start by defining and explaining the Pocket user persona, mission, and product features. I will also highlight what is currently lacking or underdeveloped in the Pocket product, thus illustrating the need for the changes proposed in my 12-month plan.  

The Pocket persona

Who is Pocket for? Even without doing user research, we can at least say the following things about Pocket users: 

  • They like to read. So much so, that they save articles for later. 
  • They like to read, and so they’re probably curious about the world around them.
  • They are web-savvy, given that they know how to install and use browser extensions.
  • Knowing all this, they’re probably well-educated, part of the higher middle-class, exercise regularly, vote Democrat, are worried about where their country is going, and don’t trust Big Tech. 

All of this is to Pocket’s advantage, because the Pocket brand has the following associations: 

  • It’s simple to use.
  • It’s playful and approachable.
  • It celebrates the open web, and is not in any way tied to Big Tech.
  • It’s respectful of privacy.

So it aligns with the values of our users. But what does Pocket do, exactly? 

Pocket’s value proposition: a bad deal

We can define Pocket’s mission statement as follows: 

For people who find more good articles on the internet than they can read, Pocket provides a simple, reliable and playful way to store those articles, and resurface them for later consumption. 

  • Unlike Instapaper, Pocket’s playful brand is untarnished by the association with Amazon and its inhumane labor practices (Amazon Kindles come with Instapaper pre-installed). 
  • Unlike Readwise, Pocket is approachable and more about reading for pleasure and understanding, than reading for some sort of capitalist Bildung-for-productivity’s sake. 

Pocket, in other words, is a dependable and playful web utility with moral values and a quirky taste in design – a rare thing in the history of the web, and rarer still is that it is still around, 15 years later. 

In summary, strategically speaking, the Pocket brand is extremely well positioned. It does what users think they want, and it aligns with their core values. However, despite its clear use case, historically Pocket has struggled to formulate a convincing argument for why users should pay for its service. 

Below I list the five benefits users supposedly get for buying a subscription (USPs), and my arguments for why I think we could improve our sales pitch. (For more information, visit

(0) Save, read, watch & listen: This is the main use case of the app, and the subscription page admits I can already do these things as a non-subscribed user. Not a reason for buying, not even a USP. 

(1) Permanent library of everything you’ve saved: Telling potential subscribers at the same time that (a) the library of non-subscribed users is non-permanent *gasp* without going into detail about when my items will start disappearing, and so not capitalizing on the instilled fear, and yet at the same time, (b), calling to mind that, historically, idealistic web utilities have folded, wiping all their user-generated content as they go (Google Reader, etc.). Yeah you’re right, conversion funnel: on the internet, no consumer service is permanent. So why pay for Pocket’s archive? It’s going to go away next month probably, anyway. Not a reason for buying. 

(2) Suggested tags: For what exactly? I genuinely wonder who uses this. Instead of neatly stacking my articles into tag-based lists by hand, my articles get labeled but automatically? And probably with labels that almost but not quite match my own system? We will return to this idea of useless productivity when we explore how to improve the Reading experience. Not a reason for buying. 

(3) Full-text search: So the Search functionality for non-subscribed users does not search through articles? That explains why it’s so poor at finding what I’m looking for. Providing a bad user experience upfront, however, does not fill consumers with confidence that things will be better on the other side of the paywall. Not a reason for buying. 

(4) Unlimited highlights: There’s a limit to how much I can highlight? Somehow, the idea of charging for the digital equivalent of neon ink feels like excessive nickel-and-diming. Not a reason for buying. 

(5) Premium fonts: The connection is never made explicit, but this benefit is supposed to speak to consumers who desperately – and understandably – want to improve the reading experience. Again, a bad consumer experience does not make a convincing argument for paying. But aside from that, the font is not the main problem with the reading experience, as we will see when we explore improvements. Not a reason for buying. 

(6) What to pay for instead: If these USPs aren’t convincing, what will people pay for? Why are people currently paying for Pocket? I think that, in spite of the USPs mentioned above, people currently paying for Pocket do so because they want to feel like they contribute. They’re paying to keep a great part of the internet alive. And that’s great – except that if that’s our main reason for conversion, I’d expect it on the subscription page as a USP (The Guardian does this very effectively). 

That’s why I think people currently subscribe. But let’s first look at what functionality Pocket actually provides and if we can find valuable features and improvements there. 

Pocket’s jobs to be done, and further improvements

For Pocket users, the application has 4 general jobs to be done, that I list below in order of priority: 

  1. Saving 
  2. Reading
  3. Sharing
  4. Discovering

I will describe how Pocket has chosen to address those jobs (its feature set) in some detail and suggest improvements.  


Pocket’s main job to be done is saving that interesting article for later consumption, because I can’t read it right now. Pocket handles that by providing save extensions and browser plugins on all platforms, coupled with a user account, and a landing page or app where logged-in users find all their saved articles. 

The Saving experience works very reliably. Though on the web, for the stressy mindset that motivates users to click their browser extension, it is somewhat cluttered. I mean that I have no desire to organize my tags from a miniscule ‘save successful’ confirmation popup as I hurriedly move the cursor towards the ‘close tab’ button while my manager’s shadow slowly spreads itself across the surface of my standing desk. On mobile, the confirmation popup doesn’t dismiss itself automatically. Imagine you’re a squirrel, and you have a nut that sticks to your little claws when you want to store it away. How would you feel? Also, is that magpie following us? Also, there seems to be a persistent issue with opening iOS share sheets from in-app Safari webviews, but that’s not Pocket’s fault. 

Saving needs to be reliable, fast, and hassle-free. Pocket is almost there, but not quite. 


Saving is for many users as much aspirational as it is actually useful. Like squirrels scouring the forest to find nuts that they then methodically store and only sometimes consume, so too do Pocket users browse the web, conscientiously save articles, and then sometimes come to Pocket to read those articles. In short: it is my conjecture that people save more articles than they read. 

However, I will also bet good money that most Pocket users want more time to read. If they could spend more time on Pocket and just read, like in the old days before the Internet mushed their brains, they would. It’s just that they feel drawn to their darn phones so much. 

Let’s help the poor suckers! Let’s explain how they can wean themselves off the dopamine drip. Help them block time for reading in their calendar. Educate them on deleting apps, or setting up Focus Time, blocking access to sites, or putting away their phone. Hire someone to try out every digital detox method and write about it. Send digests and article lists on Saturday morning. Encourage printing articles for offline reading, or send out personalized printed feeds yourself. Sell our users Kobo ereaders! 

Real reading is more than just reading more, though. Active reading requires highlighting and note-taking, and revisiting those notes, and being able to gather up those notes, and take them with us to our desk to use in our writing (see also: Sönke Ahrends, ‘How to Take Smart Notes’). Highlights and notes, at their best, are ready-to-go building blocks for writing new articles. But Pocket highlights in their current implementation are not easily translated into new writing, because they are not even stand-alone notes. Instead, on the web, the ‘Pocket highlights’ button brings you to a list of all articles that have highlights in them… In the meantime, fucking Readwise moves into the ‘saving web articles’ space, people! – The solution is obvious. Enable people to write about the articles they read on Pocket: first, present highlights as standalone input for the writing/thinking process. Then allow more complex note taking with a full-on focused writing environment, then integrate with Kobo’s highlights and notes feature for ebooks, collecting those inputs in Pocket for elaboration, then allow for ebook reading directly from Pocket, then randomly resurface noteless quotes to gobble up Readwise, then Goodreads, then buy Kobo (because “people who are really serious about software should make their own hardware”, to quote Alan Kay). 

I haven’t even gotten to the core reading experience, which is… not good. It’s not good, partly because of reasons beyond our control (GDPR, paywalls), but also because of some poor but seemingly deliberate choices in fonts, for example. 

By the way, have you considered creating RSS feeds out of those article listening experiences? So people can listen to their Pocket queue from their podcast app? 


People want to share the crazy stuff they read online. With friends, for a laugh, or in their semi-social bubble or in the company Slack, to impress. Sending a link is good enough between friends – but why not help out the ambitious reader a bit? Automatically send out an email digest of their favorite articles each week to their subscribers, complete with notes and commentary if they want. Maybe we even allow people to monetize their curation, as Ghost and Substack and now Twitter-owned Revue do. 

Then, when people use Pocket to easily curate reading lists, and compose and share digests, are we truly laying the groundwork for Pocket to become a destination – which is where ML comes in.


Squirrels don’t need more nuts. They know where to find fresh nuts. The trouble is in locating the nuts they buried last week. The same goes for our users: new content is not the issue. Current Pocket users already know what sources to check for articles. Improving the Discovery part of the app is not crucial for those core users. 

The people that need help discovering new articles are those that currently get their content via Facebook, Twitter, Reddit. For them, Pocket can become a destination, a safe haven for normies realizing that they long for stress-free reading and thinking. However, the normies will come to Pocket and not see anything in their queue. That group of people, we can serve with machine learning, by surfacing new Pocket-made curated lists and think pieces, most saved and most shared articles, friends’ digests, etc. 


I think I’ve shown convincingly that despite Pocket’s great brand and mission, its monetization strategy and functionality are deeply, deeply flawed. We’ve explored where and how to improve its feature set, and at the top you find my 12-month plan. 

By implementing my 12-month plan, I’m confident we can take Pocket to new heights: it will help active readers to share their voice, it will finally get the success it deserves as a great consumer product, and it will inspire young entrepreneurs that love the open web to build great new tools. Pocket, basically a small, old, dearly beloved web utility, can be an ally in the fight against surveillance capitalism and the death of democracy, and in the process, it can become insanely profitable. 

Let me know if I can help.