Churn, retention, dark UX: welcome to unsubscription hell

I cancelled two subscriptions last week. 

The first was for a major weekly newspaper with 1.5 million subscribers. It's one of those subscriptions you can't cancel online (how is this OK?). 

I had called a few months earlier and told them I wanted to cancel. They said they had made a note of it and the subscription would run out in August and not renew. 

Come September, sure enough there was a new charge on my card and they kept delivering the paper. I cancelled again and was now refunded the pro-rated (!) annual amount they had initially taken. Fine.

This morning I get an email from them asking for my reasons for cancelling. 

Fair enough, I think. They want to know how to get better. 

So I go through an automated funnel that, after every option I could have cancelled for tries to counter the argument. In the end I select "Other" (which was true, I cancelled because of their editorial stance) and the funnel culminates in: "Sorry to see you go, you can send an email to" 

Oh. They didn't really want feedback. What they wanted was to get a last shot at keeping me from churning. Bastards.

The other subscription I cancelled (or tried to!) was for a big online subscription startup. 15 million users. I had started the free trial a few days ago, but wasn't using it. So I wanted to cancel before it charged me an annual fee.

I went to their site. Under "Account" there was "Subscription Management." Great, I thought.

But it turned out that under "Subscription Management" there was really no way to manage my subscription. Instead it sent me to a Zendesk flow that had me answer multiple questions about wanting to cancel. Fine, I thought, maybe they haven't gotten around to building it yet (after 15 million users? Man I'm naïve sometimes). 

At the end of the Zendesk flow it told me to write a message to customer support (seriously?).

So I emailed customer support. They were fairly responsive (~12 hours) and very friendly and said they couldn't find my trial in the system. Instead they sent me a long email about all the benefits I got from their product. 

I emailed from the same email I had signed up with, so... seriously? That's like dark UX social engineering (aka "lying"). 

I ended up forwarding them my confirmation email to prove that there was a trial and asked them to cancel. So that's the end of that, I thought. Sure enough I get a Zendesk email asking for feedback (nope, not making that mistake again). 

Deliberate misdirection, invisible unsubscribes, forced subscription continuity... these are just some of the dark UX patterns we can see subscription startups employing to rope people into something they don't want anymore.

In my case it's just a nuisance. Just thirty minutes of my life I won't get back, but the $100 wasted doesn't matter. But for other people, it could really matter. So how is this ok? 

Of course this stuff goes beyond subscription: check out @DarkPatterns on Twitter for some prime examples. 

In the end, it always comes down to how ethical management is. And that mostly comes down to culture, how they were raised, and the source of their ambition. It's hard to see this type of greedy behavior from companies that I respect in an industry that I love. 

But some of it may also be driven by competition: if my competitors can pay higher acquisition cost because their lifetime values are higher because they've managed to reduce churn through dark UX, then I may have a real problem. 

I'm a limited government guy, so I'm not sure whether regulation is the best answer here. Perhaps the market will self-regulate through public shaming. In any case here are some best practices in my view:

- if I can sign up online, I should be able to cancel online 

- let me know when my renewal comes up, so I have the choice

- if I accidentally renew (you can see I'm not using the product), refund me (at least pro-rated)

- allow people to transfer/sell their subscription to others at any time during the life of the subscription

- don't mislead, trick, shame, or otherwise goad people into buying something they don't really want

I've always been reluctant to subscribe to stuff via Apple because I find the 30% tax (reduced over time, I know) egregious. But if it gives me better control over my subscriptions, I might just start doing that in the future.

Data collection and privacy, a rant on facial recognition software

The first time I walked through automated passport scan and facial recognition gates at an airport, I felt a bit uneasy. 

Eric Schmidt said "if you've done nothing wrong, you've got nothing to worry about." And if you know anything about privacy, you know that this "nothing to hide argument" is common but poor. 

The best snarky counter I've heard: it's like saying the right to free speech is only important if you have something to say. The whole point of individual liberty (aka Western civilization!) is that the government's intrusions into your life need to be justified. 

The more sophisticated argument runs: the point is not that there's something to hide, but that the person has the power to hide something. The discovery of this ability - to shield the self from others - is the wellspring of individuality. 

Facial recognition software is being rapidly embedded into camera sensors everywhere. The price of surveillance, hardware and software, is falling further and further. Tracking is ubiquitous and inevitable. 

The only cogent regulation I can come up with, given how pervasive this tech will become and the power of the companies and governments deploying it, is to regulate its use cases: 

- You cannot use emotions derived from facial expression to deny insurance cover, but you can use them to prompt secondary screening at immigration. 

- You cannot use mobile phone patterns as probable cause, but you can use them to surveil an already suspicious suspect. 

- You cannot use browsing history to construct an ad targeting profile, but you can use sites where the user is logged in. Etc.

This is certainly a partial answer. The better solution would be to stipulate that all devices and software must remain fully transparent to the end user any time it communicates - expanding the inspection principle of free software to proprietary code. It's hard and costly, but it could work. 

The introduction of the GDPR shows how difficult this stuff is. Whereas parts of the directive are well-intentioned, the implementation is that very little has changed. It's just more paperwork and a weak demonstration of the illusion of power in the EU. Oh yeah, and governments are largely exempt (wtf?). 

In the meantime, we've increased our collective anxiety a bit further. We've lost a bit more about what it means to be human. And we've inched more towards the dystopian totalitarian future that science fiction seems so hell-bent on prescribing for us.

Seriously, what was the last future-positive AI literature you've read? Star Trek? My point exactly.  

When to pivot, when to fail

Fred Wilson has a good post about a frequently encountered scenario today: Pivot or Fail? The gist is that, while the pivot seems universally celebrated in our industry (think Twitter, Slack), there are great reasons for failing gracefully.

To summarize (though you should read his original post), a hard pivot means potentially taking the wrong team, the wrong investor base, and too much dilution into your "next" thing, when what might be best is to just start again from scratch. 

I can empathize with that position. I've been involved in companies post-pivot which where just less interesting than what the founders and we had originally set out to do. But thankfully my "interest" isn't the only thing that matters. As Fred rightly says: "...the harder path is often the best path. And the easy path is often the harder one."

A hard pivot can be incredibly messy. You may have to replace much of your team. You may have to rebalance the cap table. You may have to re-sell your shareholders on what you want to do next. How tempting, especially for very talented founders, to accept failure, shut the business down, and hand back the cash. 

And yet, I do think there are circumstances under which the hard pivot is justified and can work.

Anecdotally, one of the best teams in my portfolio, by little fault of their own, had navigated themselves into just that kind of position. Their original business had worked pretty well. The company had scaled to millions of mobile users and a few million in revenue. We were profitable and pretty unhappy. 

The market had turned out to be much less attractive than we had assumed. New competition - substitutes and direct competitors - was emerging all the time. Paid marketing was no longer viable. 

We had strategic concerns about the long-term value of what we were building. But most importantly there was one thing we were certain of: we couldn't scale this thing to anything approaching a "significant" outcome. 

In economic parlance the "opportunity cost" for everyone around the table was high, but most of all for the founders. 

Having come to the conclusion above, we discussed the "graceful failure" scenario: if we put the company up for sale now, the acquirer would want to hire the team for at least 12-24 months. Including mandating an advisor, running the process, the founders taking some leave after the earn-out, and coming up with a new idea, it would be three to four years before they'd be back with another company. 

But they were raring to go. And most importantly, we had backed primarily them, the team, first at Seed and then at Series A. What they were doing was always a secondary consideration to us.

So around two years ago, we pivoted - hard. They pitched each VC partnership their new idea and we jointly gave them the go-ahead. All investor-shareholders took a significant haircut (15-20% I believe) to re-up the founders and reset the option pool. The founders restructured the team, keeping only those employees who were motivated by the new venture and could contribute. 

We then made the existing profitable business a subsidiary with its own Managing Director in order to run it for cash while exploring a sale. We ended up selling it around a year after that, netting several million which went straight into the founders' new business.

Sound messy? It certainly had the potential to be. But because we all made sacrifices to align ourselves it worked. It's a testament to the strong bond the founders had built with each investor, the investors' belief that this was a very talented team worth backing through thick and thin, and the collegiate, cooperative nature of the syndicate itself.

Today we are happy we pivoted. The new business is thriving, having added hundreds of customers and generating more revenue than the old business ever could have. The future looks 10x bright and the sky is the limit - as it should be. And yes, we had a few more soft pivots along the way.

If I look at the timeline, the founders would be nearing the end of their earn-out at some corporate around now. For two years they would have worked on a business they had long stopped believing in, in a setting they never really chose for themselves. 

While they could have walked away from it all, I'm glad they didn't. At the key juncture, we demonstrated loyalty to them and their journey. And they in turn have been very loyal to us. There's nothing misguided in that.

The venture business has high failure rates baked into its business model. But it's also flexible enough to allow for things that aren't in the playbook. Things that support long-term relationships that look beyond what is profitable for each party or makes economic sense in the near-term.

So is this type of hard pivot the exception to the rule of failing fast and gracefully? I'm not so sure. There's a lot to be said for clean endings. But when all parties are excited about continuing to work together and they have a shot at greatness? Then failure isn't really an option. 

"No/But" vs "Yes/And" kind of pitch meetings

I had a candid discussion with one of our founders the other day about what makes for a good VC interaction in a fundraising process. 

There were a lot more detailed points but what it came down to was this:

"As a founder, you're constantly doubted by everyone. And you need to sell to everyone. So when a VC comes along and says: you don't have to sell to me, I understand what you do, and I will help you make this great - that's the person you want on your side." 

This is interesting because I have "first pitch" meetings that can go both ways. Ones in which I'm very critical because I see weaknesses in the model very quickly and I feel like the founder doesn't really address them. And so they stay at a "No/But" kind of level, where the founder insists on one thing and I don't quite agree and bring up lots of stuff that contradicts or challenges them.

And then there's the second type of meeting, where I get to "Yes/And" very quickly. Those meetings are great because the mental models the founders and I work from are similar. And we ping off each other in a good way and I can see how we could help. 

I never get to the point where I invest if the first meeting is a "No/But" interaction. And I don't by any means invest in all the founders I have "Yes/And" meetings with. 

But it did get me thinking about how I show up. As a former entrepreneur turned VC, of course you want every single company you meet with to succeed. And one of the big challenges in venture, for me, is sitting in judgement of people, having to assess models, when what you really want to do most of the time is just appreciate the great thing that they're trying to do.

I'd like to have more "Yes/And" meetings even if that means suspending disbelief for half an hour to get the energy/mindset/attitude into that place. I will work on that.

Operationalizing product-market fit at Superhuman

Rahul Vohra of Superhuman has a great post on First Round's excellent Review platform: How Superhuman Built an Engine to Find Product/Market Fit

It's about operationalizing the product-market fit construct using Sean Ellis' (he of LogMeIn, Dropbox, and Eventbrite fame) customer development survey. The approach centers on the question: how would you feel if you could no longer use the product? And digging down on the further responses of "Very Disappointed" and "Somewhat Disappointed" users. 

Sean's heuristic is that you need to get the "Very Disappointed" number to 40% of users to be at product-market fit.

A lot has been said that's critical about the customer development process. And I agree with some of it (it's moderately unhelpful on the Thiel path of 0 to 1 rather than 1 to n).

But Rahul's implementation is one of the best narratives about it that I've seen. A lot of Sunstone's seed stage startups use a similar process and we encourage any seed-stage company to consider using it. It makes progress measurable while you're still in prototype stage.

Measure what matters - this does. 

Ryan Sweeney's Ode to Qualtrics

Sometimes the best love stories are those that you only catch a glimpse of from afar.

I had the pleasure of meeting Ryan Smith, the co-founder and CEO of Qualtrics, in 2013 at Founders (thanks, Paddy!). And I met Ryan Sweeney briefly when I worked at Accel in 2012. Exceptional people.

Now Qualtrics has been acquired by SAP for $8 billion (21x LTM revs!) and it has defined a new category: experience management. 

But what struck the deepest chord in me is Sweeney's poetic post about the acquisition. It's unlike anything I've seen a VC write about an exit. A founder relationship with soul indeed. 

The full post is here on Accel's website

Defending the indefensible & the state of the media

Hussein Kanji is a prolific social media link poster. Next to his side gig as perhaps Europe's most underrated unicorn maker at seed stage with Hoxton Ventures (Deliveroo, Darktrace, Babylon), his main endeavour day in and day out is the curation of my Facebook feed. I have no idea how he does it - I'd usually wager some sort of combination of Zapier and Buffer, but perhaps he actually does read all of the stuff. In which case he's the most well-informed person I know. This post is not about him.

Sometimes something so egregious slips through the cracks of my Facebook newsfeed that it bears mentioning. This post is about that. It is on the border of not wanting to comment on politics (I did last week and it went poorly). 

The once venerable Newsweek has a piece on the Acting AG of the US, Matthew Whitaker, with the salacious title ACTING ATTORNEY GENERAL MATTHEW WHITAKER ONCE SAID JEWS, MUSLIMS AND ATHEISTS SHOULD NOT BE FEDERAL JUDGES. Yes, all caps. 

Wow, I think, and take the clickbait. The article, written by Nicole Goodkind, starts off promising:

New acting Attorney General Matthew Whitaker said in 2014 that judges without a New Testament “biblical view of justice,” should not serve on the federal bench and suggested that he would block the appointment of non-Christian judges if given the chance.

While Whitaker singled out atheists in particular as being unfit to serve, his comments also extended to Jewish and Muslim Americans.

Woah, I think. Really? That a man of such blatant bias would make it so far in the Justice Department is unfathomable. But these are strange times, aren't they? I read on.

Whittaker's two competitors [ed.: in the Republican primary for Iowa Senator], Sam Clovis and Joni Ernst, said that they would use faith-based criteria and make sure they acknowledged “natural law.”

Natural law is the belief that legal rights and systems of morals were given to humans by God and were not derived from the rules of society. The concept is often used in religious communities to deem certain acts immoral and “unnatural” like the use of contraception or same-sex relationships.

Oof. Dear Newsweek, that's not actually the definition of "natural law" at all. I mean, Wikipedia knows more about it than you. The point of natural law is that it derives from, well, nature. It does not comment per se on the metaphysics of that nature, if any. And atheist derivations of ethics from natural law abound, from Plato and Aristoteles onwards. Jewish and Islamic and Catholic philosophies of natural law exist. 

And - dare we say it - much of common law jurisprudence is founded on natural law. The Declaration of Independence bases its validity, the entitlement of the "separate and equal station", on the "Laws of Nature" (and to be fair, "of Nature's God").

Take "Thou Shalt Not Kill." A common view in theology is that this is a natural law, God-given perhaps not so much via a miraculous event involving stone tablets but because it was discovered as a central natural rule without which humans could not live together. Every fiber in our being knows it's wrong to kill another person. If we permit it to happen, "society" fractures (also, you get Horcruxes). That is the meaning of "natural law."

Reading on, we learn about Matthew Whitaker's criteria for selecting judges to the federal bench:

"What I’d like to see is things like their worldview.… Are they people of faith? Do they have a biblical view of justice? I think that is very important.”

The moderator interrupted Whitaker and asked “Levitical or New Testament?” which is an indirect way of asking whether people of the Jewish faith should be banned from serving as federal judges.

“I’m a New Testament,” responded Whitaker to laughter. “And what I know is as long as they have that [New Testament] worldview that they’ll be a good judge.”

One wishes Newsweek had expanded that first elipsis. We would know more about what Whitaker actually said. But by substituting [New Testament] in brackets, the journalist Nicole Goodkind is putting words into his mouth and then determines that those words are anti-semitic. 

As an aside: the interjection of "Levitican or New Testament" is a reference to the (wrong-headed) view that the God of the Old Testament is a vengeful deity and the God of the New Testament is forgiving and loving. It's a frequent Christian trope, hence the laughter. 

But back to the main event. Yes, Whitaker seems to have said that he wants to understand a potential judge's worldview. Clearly a person of faith is something he likes because he believes they understand the concept of natural law. But his framing it as "person of faith" sounds quite inclusive of other religions, to me at least.

But then he drops the big one - do they have a "biblical view of justice"? Obviously that's something that plays quite well with the conservative base in how it sounds (remember, these are Republican primaries). But Newsweek uses it to suggest that this excludes in Matthew Whitaker's view anyone from serving as a federal judge who is not Christian. 

Let's take a step back. First of all, nowhere does he say that people without a biblical view of justice won't make good judges - it seems to him a shortcut to determine a worldview that would make for a good judge. But more importantly, the journalist is once again confused about what is actual a technical term.

"Biblical justice" is a well-defined concept and has a rich tradition, something that Nicole Goodkind apparently does not realize. It starts with the Judaic concept of mishpat, to treat people equitably regardless of social status, class, race, or gender. Biblical justice is in fact directly opposed to the bias that Newsweek's headline wants to attribute to Whitaker.

But biblical justice goes further by extending mishpat to tzadeqah, perhaps best translated as "righteousness." This means a justice not limited to righting wrongs, but extended to preventative and restorative acts like generosity and charity. 

The reference to the New Testament means Whitaker personally believes that a charitable, generous, equitable justice should apply especially to those who have broken laws. But a biblical view of justice does not require the belief in God. And it certainly does not require a person to be Christian.

Oh wow, someone who wants unbiased judges that are charitable? Get out the pitchforks! 

This article is everything that is wrong with the media today. It oversimplifies. It tries to identify prejudice where there is none. It demonizes a possibly good person. It does not do its research. It simply assumes that someone is not "on the right team" and declares them fair game. 

And so I find myself defending the indefensible. I do not know whether Matthew Whitaker is a good person or the right pick for AG. But for the purposes above, it does not matter, because he did not actually say what Newsweek heard or pretended to hear. 

I do care when media tries to fan unnecessary racial and religious and political divides. Newsweek would do well to practice a bit of biblical justice. Whether it calls it that or, you know, "humanist." Because they are exactly the same damn thing.

Ex Machina (2015)

I had never watched Ex Machina, a 2015 film about the development of AGI, so I did last night. 

It's a good movie. Visually polished, engaging, sparse but well executed. I was a bit disappointed it co-mingled AGI with robotics so quickly, and except for the Turing test reference there wasn't any depth to the science behind the story. But that's expecting too much of a cinema feature.

What the film did do well was raise questions of power, creation, human frailty, gender, and manipulation. And the score was great. I think I'll watch more Alex Garland going forward. 

Arts & Letters Daily

For a very long time, Arts & Letters Daily ( was my favorite website. I think I discovered it around 2000, two years after it had launched.

It was founded and edited by Dennis Dutton, a philosopher at Christchurch (NZ) and one of the great thinkers on aesthetics. 

Its genius is simplicity: three links a day - an article, a book review, and an essay, unceasingly with a great short caption. And then a short Nota Bene section, with some current news or events, and a long list of links to other sites which, for a while, was the best blog roll out there. 

When Dutton passed away in 2010, the quality of the posts declined and it felt like ALDaily was no longer at the forefront of the more interesting debates of our time: Pinker, Hitchens, Chomsky, Singer, et al.

Over the past few weeks I feel the quality has noticeably improved, though nothing has changed (the editor, Evan Goldstein, has been in the post since 2011). 

Nevertheless, it is worth a visit especially on a weekend.

Much love,


An introduction to our framework for assessing consumer companies

My colleague Yacine has a short post on Medium about our current framework for assessing consumer internet companies

It's based on Clayton Christensen's 2016 thesis of consumers using products to get a certain job done

While that's a great starting point, we added to this four major functional differentiators: price, time, quality, and user experience. And six emotional aspirations: the desire to be special, to improve, to escape, to belong, to be safe, and to be free. 

In conjunction, these dimensions of a consumer product represent the consumer surplus derived from purchasing and using it: the delta between its cost to the consumer (not just the price!) and its benefits (not just the functional ones!). 

We'd love to know what you think about our framework. It's definitely work-in-progress. 

Please read the post and discuss with Yacine and me on Twitter.