five signals.
one week.
the surface is
the business._
asked what craft means, 57 percent of designers said visual polish and attention to detail. not speed. not volume.
figma's state of design 2026, published in march, asked working designers what craft means. the top answer was visual polish and attention to detail, at 57 percent. thoughtful problem solving came second at 47. neither answer mentions output. a profession that has spent two years being told its value is production said, when asked plainly, that its value is judgement about finish.
this is good news with a bill attached. if craft is polish and problem selection, then craft is the last place you can still buy an advantage, and almost nobody is buying it. they are buying more output at a lower price and calling the result a refresh. the advantage is sitting in the part everyone is defunding.
the cheap version of this is not a rebrand. read your homepage out loud, find the first sentence that says what you sell to a person in their own words, and if it is not the first sentence, move it. then take the three places the eye lands first and finish them properly. that is one day of work and it is the highest return day available to you this quarter.
68.01 percent of us google searches between january and april ended without a single click.
sparktoro's study, published 9 june, puts the us zero click rate at 68.01 percent for january to april 2026, against 60.45 percent in 2024. a 7.56 point move in two years. ai overviews now appear on more than 20 percent of searches and cut click through on those results by close to 60 percent. the search result stopped being a doorway. it is the answer.
the cost is not a traffic number, it is a strategy that quietly expired. every dollar in your site is allocated on the assumption that a person arrives, reads and decides. increasingly the thing that arrives is a model deciding whether to mention you at all, and it does not scroll, does not feel your typography, and cannot cite a feeling. if the page is atmosphere, there is nothing in it to quote.
write the specifics down in plain sentences where they can be retrieved. what you do, who you do it for, what it costs, where you operate, what you will not take on. one page, no hedging, no cleverness. a company that publishes facts gets cited. a company that publishes a mood gets summarised into somebody else's paragraph.
air canada argued to a tribunal that its own chatbot was a separate legal entity.
a passenger asked the airline's website about bereavement fares. the chatbot described a refund policy that did not exist. when the passenger tried to claim it the airline refused, arguing among other things that the customer should have read the real policy the bot had linked to. canada's civil resolution tribunal rejected that in february 2024 and ordered the airline to pay 650.88 canadian dollars plus interest and fees.
the money is not the story, the finding is. everything on your surface speaks with your authority, including the thing somebody added in an afternoon because a vendor said it would deflect tickets. you cannot put a widget on the front of house and hold it at arm's length from the back of house. the gap between what the surface says and what the machinery does gets discovered by exactly one person, the buyer, at the worst possible moment, and now occasionally by a tribunal.
go and read what your own interface tells people about refunds, timelines, availability and price. then go and check whether the people behind it can deliver that. where the two disagree, either the machinery wins or the promise changes. there is no third option where the disclaimer saves you.
the surface is all that is left.
the model reads it. the court reads it.
the visitor comes last.
three public release ledgers count the first two weeks of september at six, seven and sixteen models. they also disagree on which day openai shipped its flagship.
llm-stats, benchlm and digitalapplied all publish dated release ledgers. for 1 to 14 september 2026 they report six, seven and sixteen releases. two date gpt-6 astra to the third and one to the fourth, and each lists models the others do not carry at all. these are not opinion pages. they are the trackers people cite when they tell you the pace is accelerating.
the absurdity comes with a bill. a company waiting for clarity before it acts is waiting for something that does not exist, and the wait is not neutral. every quarter spent evaluating is a quarter a competitor spent shipping something mediocre and learning from it. the evaluation is the delay. the delay is the decision, and it is being taken by default rather than by anyone in the room.
stop evaluating the field and evaluate one workflow. take the handoff that hurts most, put something on it this month, and measure the handoff rather than the model. the model will be replaced twice before your process changes once, which is the entire argument for fixing the process first.
the output is now the cheapest thing in your company.
price the parts of your business against that. anything that is production is heading for the floor, and fast enough that pricing it as skilled labour is a bet against the calendar. anything that is judgement about what to make, what to cut, what good looks like and when to stop is heading the other way, because it is the part that did not get automated and the part a client cannot check for themselves.
the companies in trouble are the ones whose price was a proxy for effort. when the effort disappears the price has no story left. the ones that are fine were being paid for the call rather than the hours, and most of them have not noticed that this is now their entire position.
take the last three things you shipped and ask what you removed from each one. if the answer is nothing, you were producing rather than deciding, and the market will price you accordingly inside a year. the discipline is subtraction. start charging for it out loud, as the thing itself.