The IKEA Effect, Industrialised

A founder opened her laptop to me on a video call in May 2025, sharing her screen before she had finished saying hello. The screen showed Cursor with a few thousand lines of TypeScript, a home feed that rendered, a broken auth flow and a CSS variable she had named --nonsense-green. She ran a small import business in Auckland and had spent her weekends building what she was showing me. She shared her screen with a small deliberate motion that I have since learned to recognise. It was the motion of a person handing you something they had made.

Her name is not relevant here. She had a view on which corner of the navbar the search icon belonged in, and the view was not negotiable.

I opened the prototype and shared my screen. I clicked around for about twelve minutes while she watched. Three things happened in those twelve minutes that had never happened on a call in 2018, or in 2014, or in 2011, and I want to name them precisely.

First, I noticed her defend a decision. I had navigated to the filter modal and tutted slightly, an engineer’s tut, involuntary, at a dropdown that did not need to exist. She said, before I said anything, that the dropdown existed because three of her suppliers used different names for the same product category and she had added it so she would stop mentally renaming things each morning. She was explaining rather than asking permission.

Second, I noticed I could not tell her the prototype was wrong in the usual way. In the old engagement, a scoping conversation began with me finding holes in the brief and pointing at them. Here there were no holes, because there was no brief. The product was the brief. The brief ran on her laptop. When I showed her the broken auth flow she nodded and said yeah, that is where I ran out of Saturday. Instead of a negotiation between her idea and my interpretation of her idea, the conversation was the two of us looking at something she had already made.

Third, and this is the one that matters: when I suggested that the dropdown could probably go, she paused, visibly, and I saw the small flinch of a person being asked to part with something. She said the dropdown could stay until we had data that said otherwise. I said that was fine. We moved on.

We used to manufacture love. Now we inherit it.

I want to draw a contrast with a client from 2018 who I will call Marcus, and who is a composite whose details I have combined to preserve anonymity. Murray in the first chapter is an illustrative reconstruction of the same translation problem. Marcus ran a small group of boutique fitness studios in Sydney. He had a couple of hundred paying members and an idea about a class-booking app that would, in his words, feel like his brand and not like a generic booking tool. He came to Sliday with a long PDF written by a consultant he had paid well. The PDF was genuinely good. We read it twice. We ran two discovery calls. We extracted user stories, sketched flows, agreed acceptance criteria, signed a fixed-scope statement of work and built the app in sixteen weeks.

What Marcus wanted, and what neither he nor the consultant had managed to put on the page, was a home screen that felt like walking into one of his studios at seven in the morning — the wall of named regulars, the class on the board, the hand-chalked quote above the desk. The PDF had asked for a dashboard summarising upcoming classes and recent activity, and the team had built exactly that: a tidy card grid with the next three classes, a members-joined counter and a neat little line chart. It was the correct implementation of the correct sentence in the correct document. It also answered a question Marcus had never actually asked, which was how is my business performing today, instead of the question he lived inside, which was who is in the room this morning. One line in a spec had absorbed the entire emotional register of the product, and nobody had noticed, because the line was not wrong on its own terms.

I flew to Sydney to demo it to him in a glass-walled meeting room above one of his studios. He was polite through the booking flow. He was polite through the membership page. He reached the home screen and paused. He leaned back, folded his arms and said, in almost exactly these words: this is not what I had in mind.

Marcus was Murray’s failure mode half a decade later: the translation arrived correct and the customer did not recognise it.

The app was an excellent implementation of the document we had written together, which was itself an excellent interpretation of a brief his consultant had produced, which was in turn a plausible guess at the picture in Marcus’s head. Three translations. Each one faithful. The cumulative effect was a product he could use and did not love. We finished the engagement, delivered the app, invoiced for the remainder, and Marcus quietly moved to an off-the-shelf booking tool within eighteen months. There was no falling-out. There was just an absence, where an advocate should have been.

Every studio has a Marcus. If you run a studio you have more than one.

There is a small body of peer-reviewed research that names what was happening in Sydney and what was happening on my screen in May 2025. The name for it is the IKEA effect, which Michael Norton, Daniel Mochon and Dan Ariely documented in a set of experiments published in the Journal of Consumer Psychology in 2012.¹ The finding, paraphrased: when people assemble something themselves (a box of Lego, an IKEA storage unit, an origami frog) they rate the finished object as more valuable than the identical object assembled by someone else. Their labour attaches. They bid more to keep it. The effect holds whether the thing they made is beautiful or lopsided, and it disappears if the build is left unfinished or taken apart again. Their builders were amateurs, and the comparison objects were expert-made; whether an expert attaches to her own work the same way is a question the experiments did not put. The attachment is to a completed act of assembly, not to the effort alone. I use this study to describe a possible mechanism, not to establish a measured effect in software founders.

The experiments were done with furniture and paper models. The mechanism is older than that. Marcus did not assemble his booking app, and the app did not stick to him. The Auckland importer assembled her prototype, and she bid — instinctively, without asking — for the dropdown to stay in the nav. Tomo, the CEO of Genomelink, put it plainly when we spoke: everyone on his team, designer, scientist or engineer, now owns a feature from the research stage through to prototyping. He had built a revenue-quality MVP largely by himself, the sort of product a younger version of him would have paid someone else to make. The ownership was not delegated. It was performed.

The psychology is forty or a hundred or a thousand years old. What is different in 2026 is that the act of assembly has collapsed, for software, into something an operator with no engineering background can do over six weekends. Cursor, Lovable, v0, Bolt and Claude Code (the tools are interchangeable for this argument, and the argument will outlast whichever three of them are still trading in 2028) let the founder perform the assembly step that, for all prior methodologies, the studio performed on the founder’s behalf. Every hour the founder spends in the tool is an hour of labour she attaches to the product. By the time she opens her laptop to us on a call, she has run the IKEA effect at industrial scale on herself.

This is not a flattering observation. I am not saying empowered founders are more deserving, more visionary, more anything. I am saying their emotional position relative to the product has changed, and that change is the single biggest silent advantage the new model has over the old one. It collapses the failure mode I described in Sydney. A founder who has assembled the prototype does not walk into a room and say this is not what I had in mind, because what is in her mind is the repository on her laptop. The mismatch has nowhere to live. Łukasz, a creative technologist based in Poland, watched something similar happen in a single afternoon: a writer friend with a specialist science background insisted she had no use for AI, certain her field was beyond its reach. He generated a sample article in her subject area. She was surprised enough to publish it on her company site after minor edits, and within minutes of doing so she had sent him five more topics to try. The fear and the re-engagement arrived almost simultaneously. Once she had shaped the output, ownership followed.

It is worth saying, to any founder reading this, that the attachment is not a free good. The same labour that binds you to the prototype will, later in the engagement, bind you to features that should be cut, to flows that tested your patience but will test a user’s more, to the dropdown you added at two in the morning because it solved a problem you, specifically, had. The IKEA effect is indifferent to whether the thing assembled is any good. Every founder we have worked with since 2024 has at some point defended a decision that, on the evidence, did not survive contact with real users. The job of a studio worth hiring is to name those moments without flinching, and the job of a founder worth working with is to hear them. The love is an asset and, managed badly, a tax on judgement. Most of the work of the first month is the two sides learning which is which.

For a studio owner reading this, the commercial consequences are not subtle. The IKEA effect is what lets a studio invoice for hardening rather than for guessing, and charge the rate the old engagement could not quite justify. When the founder already loves the product, several conversations that used to happen late in the engagement either do not happen at all or happen early, before the contract is signed. Scope arguments compress, because the founder has an opinion on every feature that survives and very few opinions about features she never built. Pricing can be anchored against the prototype rather than against a nebulous future spec, which gives both sides a shared ruler to measure work by. And (this is the harder one) a studio gains the ability to say no to pet features precisely because the founder’s love is pooled around the whole artefact, not around any one menu item; losing one screen does not threaten her sense of authorship, and she will often concede a deletion she would have fought tooth and nail under the old contract. The chapter on pricing makes the full argument, and this is only the leading edge of it.

For an engineer reading this — and specifically, for an engineer who has already had the experience of being handed a founder’s prototype and asked to finish it — none of what I have written implies that you ought to be grateful for the founder’s love, or deferential to it, or careful around it. It is a feature of the engagement, not a constraint on your judgement. The dropdown in the Auckland importer’s navbar may in fact need to go in month three, when real usage data arrives, and it will be part of our job to tell her so. The founder’s attachment is data about what she values and how hard she will fight to keep it. We respect it the way we respect any other input to the work, and we overrule it when the evidence says we should. Adults treating other adults as adults. The IKEA effect is a fact about the room, not a brief for the engineers.

What the effect does do, reliably, is give us a starting position we never had before. We used to arrive at the beginning of an engagement with a brief and a founder who could not yet picture the thing. We spent six weeks translating. By week six we had a shared mental model and a product-in-the-post that nobody had yet met. Now we arrive at the beginning of an engagement with a repository and a founder who has been picturing the thing for four weekends straight. We spend six weeks subtracting, hardening and shipping. By week six we have a product both sides already recognise.

Love is the cheapest engineering input we have ever had access to, and the founder brings it pre-installed.