I’m about to kick off a major project with a client that is turning a handful of interviews into a dashboard uncharitably, and analytics more broadly.

I have spent my whole career, even pre-Draft, circling the topic of data display like a curious leopard. Data should tell a story. Most of it, by default, does not.

Now that we’re sold through most of the year, it’s worth talking about what’s worked for us.

Building on our free letter, in this lesson we’re covering some of the more subtle, consultative aspects of critique. This has shifted over the past few years alongside new technological developments & the urgent necessity to restore design to a consultative position.

Announcement 1: Thanks to everyone for your interest in our open consulting slot. We are now sold through late October. Grateful, as always, for your support of value-based design.

For those of you who wish to increase your revenue in 2027, you may begin your application by replying to this email.

In a paid lesson from way, way back, I talked about how to pull heat & scroll maps every month and analyze how behavior changes over time. Sometimes it changes in response to the improvements you’re making; sometimes it changes alongside the behavioral characteristics of your addressable market. In either case, you should build these changes into a coherent narrative that explain what your customers are really doing.

Recent developments have made it much easier to start with this.

I was on a fun chat with a fellow consultant recently who does sales. I am unfond of doing sales. It’s nice talking to people who like doing sales, because we have a weirdly symbiotic relationship.

Every software business should ask their customers what drove them to join and how they can help. Post-signup surveys should be a part of onboarding, and they should be periodically analyzed for design improvements & product direction.

In this paid lesson, we’ll walk you through what & how.

I made a mistake recently. I was hanging out with my pal Quinn in Mexico and thanked her for referring a client to us recently. I said I hoped the box full of snacks I sent her was delicious.

“Box?”

I am, in some ways, the worst person to answer this question, since we contractually dictate tooling. However, since we have fluency in describing tooling, in knowing the systems that surround tooling, that may provide us with a perspective that could be valuable to you in the present moment.

It is every room talking exclusively of LLM. It is every podcast being about LLM. It is writing a reasonably well-read newsletter, and all of a sudden, single installment that is not about LLM ends up with half the open rate. It’s a lot of fear and anger and frustration in the face of a bunch of people who appear to have got the religion. It’s the biggest room of LLM-pilled people universally admitting that they have no idea what they’re doing. It’s a bunch of propaganda from a few companies that may not exist in five years. Business as usual, I guess.

In this lesson, we’ll talk about what we’re presently doing at Draft to synthesize design research. This is keeping in mind a few things:

• We only use locally-hosted, open-source models

• We check our work, since LLMs are still prone to making errors

• We don’t use LLMs to create final comps, instead using them in the first 80%-ish of design: research & synthesis

This has made our jobs a little easier. A little.

When I wrote last week’s letter, I didn’t mention LLM, because I didn’t have to. Then I went to a conference that I love dearly. I love it so much that, when the previous edition ended, I immediately bought a ticket for next year. Later, it pivoted entirely to discussing LLM during the day sessions.

This was a blessing in disguise. On the one hand, there is no way on earth that you will ever make me go to an all-LLM conference on purpose. On the other, I already know most of these people. I’ve even worked with some of them. So no matter what, I’ll be down for the hallway track. I know that won’t be entirely about LLM. I’ll find my people. I’ll also be able to take the temperature of the room. And I’ll do it all in the best food city in North America.

In another dimension, I would have come back throughly LLM-pilled, spending the rest of my life teaching you how to Claude up the joint. I know this will shock all of you, but that did not happen. I have been around the block long enough to both take new developments seriously and also know how to incorporate them into my practice. I literally stare at software like this for a living. I point out the failings of everything in a way that is infuriatingly correct, forever.

Those who’ve been around here for a while already know this, but: LLM is not the totality of your practice. In fact, it is not the practice of most people. Most people have just been handed a tool. Your job is not a tool. Your job has never been a tool.

What happens when someone generates a front-end comp for you and says this must be the way? Of course this hasn’t happened with any of our clients, since this is unallowable into the field. But we’ve noticed this happening in other organizations, and we read this post about the survivor of a toxic workplace recently, so we feel called to speak out.

In my 20s I had a bunch of chances to move out to the bay area or New York, work at a big company, play the startup lottery. I never did, clearly. One company approached me with a creative director role and I said I’d love to accept, and then asked them to move the company to Chicago. They did not reply to my inquiry.

I have regretted nothing about this decision. Chicago is one of the best food cities in America, and it’s also the best large city for community in America. I like both of these things enough to orient my life around them. I was once wandering a different city with a local there, and they said something that stuck out to me: “this is going to sound like I’m dissing [CITY NAME], but the people are unambitious here.” I instantly got it. Chicagoans have ambitions to throw the best barbeques and go to every street festival during the summer. They do not have ambitions about their jobs, only to do good work and vibe.

In short, I log off at precisely 5:01p every single day, and then I cook, garden, go to shows, and hang out with my friends. Which is exactly what being a Chicagoan is supposed to be. My friends frequently tell me that they have to remind themselves that I’m a very different sort of person in my day-to-day job. Good. My colleagues sometimes get invited to my big parties, and they’re shocked at the vibes. Good.

Besides the obvious quality-of-life elements, there is a fringe benefit to my job about doing all of this. None of my friends are early adopters of anything except for bike lanes and locally native fruits. Their phones are five years old and their screens are cracked. Every single one of them is pausing new tech purchases because Apple is a fascist company now. Some of them are asking about screen repair.

One of the big themes I had in the hallway track of MicroConf a couple of weeks ago surrounded A/B testing, which I guess I’m a little known for here. It looks like everyone is running A/B tests now. They’re building their own frameworks with LLMs, feature-flagging new ideas, and going off the results.

I actually love this. One of my favorite clients home-rolled their framework. Another one of my old clients built a whole Rails gem for experimentation back in the day. Most contemporary frameworks are too heavy-handed and too expensive to justify for the price.

But this also creates a lot of methodological questions.

If you’re an independent publisher – meaning you print & ship directly from your place of business, and don’t contract it out to a third party – can you please hit reply to this email? Got a question about international shipping that is best served by those who have similar infrastructure. Thanks!

I enjoyed reading this piece from Elizabeth Lopatto about the current technological conditions, and wanted to write a little bit about its ramifications for value-based design.

When we think of innovation in technology, what arrives? Maybe big developments like the iPhone. Maybe “disruptive” businesses like Uber. Machine learning – the purest expression of the technology – is, admittedly, on the list.

At MicroConf this past week, one of the best talks was by Amanda Natividad on the topic of inbound traffic.

You probably know that we specialize more in what happens after someone comes to your site. But inbound matters, because the quality of traffic that you get matters. And we find that people want to know everything that they can about their inbound traffic. They want to test cross-device. They want precise attribution.

None of this is possible right now, possibly ever.

In this lesson, we’ll talk about what to do about it.

With the benefit of one year of critical distance, we’ve overhauled our website. We:

• Pared back all content & offerings that do not structurally serve our positioning.
• Used as few words as possible.
• Added surgically. New speaking page. New bit about text.
• Swapped our former three-page application monstrosity with a simpler front door. Our previous positioning required high walls, a lot of boundaries, and deep field protection. Not so much in software. Kindred spirits. We belong here.
• Fixed a few cosmetic things in the nav.

The funnel now goes Roadmap → Revise → Retainer. Growing businesses who don’t yet fit into this funnel can pick up our books or get a teardown or call.

We will be concluding sales of two of our books soon. We love Draft Evidence & Store Design, but they no longer serve our positioning and are unallowable into the field. It’s best to call time on those while they’re still getting an impact. If you run a store or library and want copies, please get in touch – we love leaving giant crates of books on our front porch for our hardworking delivery logistics system.

In a recent paid lesson, I wrote:

When you’re moving from free to paid, there are still high objections on the customer’s part, and they may not be fully convinced as to your product’s value. They may also be forced to upgrade out of necessity, creating emotions of frustration or resentment.

This poses an interesting design challenge, because it effectively acts as the first point of conversion in the funnel. Free signups are not conversion. Conversion is an economic activity.

So in this lesson, we’ll go into the baseline design for a paid upsell, and what you should avoid.

We’d like to speak at more conferences & on more podcasts in 2026, so we’ve put together a little page showing what we could do for you, what we’ve done in the past, etc. Hit reply if you’d like us to make your event one to remember. In a good way. In a good way!

We also finally answered a question you may have had.

Last week we discussed tooling: specifically how tools are generally irrelevant, and adaptability to new tools is a strong signal of longevity in the industry.