The launch readiness framework: eight dimensions and forty checks

Why a score beats a checklist

A checklist tells every reader exactly the same thing, and a reader who disagrees with an item simply skips it and moves on. A score does something a checklist cannot: it forces you to make a judgement about your own work, then hands you back a number you probably will not like. The argument you have with that number is where the value is.

Consider the check "Is your pricing public?". On a checklist it is a tick box, and most founders tick it because a pricing page exists. In a scored assessment you have to choose between No, Partly and Yes, and the moment you consider Partly you start thinking about the "contact us" on the top tier and the usage limits written down nowhere. Partly is a more useful answer than a tick.

The second thing a score does is rank. A checklist presents forty items as equals, but a positioning failure and a missing UTM convention are not the same category of problem, and treating them as equals is how teams spend launch week on the wrong work. Weighting produces an order, usually the reverse of the order in which work feels urgent.

The eight dimensions and what they weigh

The framework splits launch readiness into eight dimensions, each carrying a fixed weight that sums to 100. The weights are opinionated and they are the argument this framework is making.

What each dimension actually measures

Positioning and message

This dimension asks whether the words on your homepage do the job of explaining the product to somebody who has never heard of it. The representative check is whether five or more people outside your company have repeated your value proposition back to you correctly, after thirty seconds on the homepage and without help. Most founders have never run that test, and most who run it fail it the first time. The positioning work that has to happen before launch is cheap to do and expensive to skip.

Demand evidence

Whether anyone has demonstrated, with something more costly than an opinion, that they want this. The check I care about most in this group is whether anyone has paid you or formally committed to pay before launch. A waitlist signup costs a stranger four seconds. A signed pilot agreement or a deposit costs them something real, and only the second kind of evidence predicts revenue.

Site and conversion path

The mechanics of turning attention into a signup or a conversation. A representative check: have you walked the entire path, from ad or post through to activated account, on a mid-range Android phone on a real mobile connection. A desktop browser squeezed to 400 pixels wide does not count, because it will not show you the 4G latency, the autofill failure or the button your thumb cannot reach.

AI visibility

Whether answer engines can retrieve you, understand what you are, and name you when somebody asks the question your product answers. The representative check is direct: ask ChatGPT, Perplexity and Gemini your main category question and see whether your product appears. Supporting checks cover third-party mentions, a comparison page naming the tools buyers weigh you against, AI crawler access in robots.txt, and Organization and Product schema with sameAs links. Semrush puts brand-owned websites at only 5 to 10 per cent of the sources AI search references, and Muck Rack found in 2026 that 84 per cent of AI citations come from earned media. Your own site is necessary and nowhere near sufficient. We go deeper on the mechanics in answer engine optimisation for product launches.

Offer and pricing

Whether the commercial shape holds up. The check that catches the most teams is whether the value metric grows as the customer succeeds. Per-seat pricing on a product that reduces headcount is a structural problem, not a pricing page problem, and it does not get easier to fix once you have customers on it. There is more on this in the guide to pricing decisions you cannot easily reverse.

Product and activation

Whether the product carries a new user to value without a human in the loop. The representative check is time to first value under ten minutes for a brand new account. Related checks cover whether the activation event is instrumented, whether an empty account teaches anything, and whether the product stays up under a traffic spike. Launch traffic converts once, so a product that needs a human in the loop produces signups rather than customers.

Search foundations

The technical and structural work that lets you be found in month nine rather than launch week. Indexability and correct canonicals, largest contentful paint under 2.5 seconds on mobile, pages targeting how buyers search rather than what you call the product internally, and per-page titles instead of templates. None of it produces launch day traffic, which is why it gets skipped.

Launch ops and measurement

Whether the launch has owners, dates and a definition of success written down in numbers before anything ships. The representative check is whether analytics and attribution are live and verified before launch rather than after. Teams that verify attribution after launch spend the following month arguing about which channel worked, with data that cannot settle it. The 90 day pre-launch runbook covers the sequencing.

Why the weights sit where they do

Positioning and demand carry 15 per cent each because no amount of launch day execution rescues a launch that nobody understood or nobody wanted. Every other dimension improves a launch; these two decide whether there was a launch worth having. When I look at the ways launches actually fail, the failure usually traces back to one of the two, and it was usually visible three months earlier to anyone willing to ask.

AI visibility sits at 13 per cent, level with the conversion path and above traditional search, and this is the weighting people push back on most. Two reasons. First, it is where a 2026 launch is most likely to be invisible without anyone noticing. A search problem shows up in Search Console within weeks; an AI visibility problem shows up as nothing at all, because there is no dashboard for the answer you were never included in. Second, it is a separate game from ranking. BrightEdge and Ahrefs both put the overlap between pages cited in AI Overviews and pages ranking in the organic top ten at somewhere between 17 and 38 per cent. Most AI citations go to pages that are not winning the same query in classic search, so doing traditional SEO well and expecting AI visibility to follow is an assumption the data does not support.

Search foundations sits at 10 because it is the compounding channel rather than the launch channel. Its payoff arrives in month six, which makes it easy to defer and expensive to retrofit.

Launch ops carries the lowest weight for a simple reason: good operations cannot save a weak product. It is also the easiest dimension to score full marks on, since every check in it is a decision rather than a constraint, which makes a low ops score a choice rather than a limitation.

How the scoring works

Forty checks, five per dimension. Each is answered No, Partly or Yes, scoring 0, 1 or 2 points. A dimension score is the points you earned as a percentage of the ten points available in that dimension. The overall score is the weighted average of the eight dimension scores using the weights in the table above.

Partly exists because binary answers push people towards Yes. Given only No and Yes, the founder with a pricing page that half explains the product ticks Yes and learns nothing. A dimension full of Partly answers scores 50 per cent, which is an accurate description of a thing that is half done.

You can run all forty questions yourself from this page. The launch readiness assessment exists to do the weighted arithmetic and rank the gaps, which is tedious by hand and easy to get wrong in the direction that flatters you.

What the five bands mean

• Launch Ready, 85 and above. Go. This band is rare and usually belongs to a team that has launched before and been burned. Tighten your two weakest checks and ship.
• Launch Capable, 70 to 84. You can launch and it will work. The gaps will not stop you, they will cost you upside you never see and therefore never miss.
• Fragile, 55 to 69. The launch will feel flat. Nothing here is fatal, but enough is weak that the result will underperform what the product deserves. Fix the top three before you set a date.
• High Risk, 40 to 54. Delay four weeks. Launching into these gaps burns attention you get once and cannot buy back.
• Not Ready, below 40. The problem sits upstream of the launch. Work the weakest dimensions before planning launch mechanics at all.

A low band is a statement about preparation rather than a verdict on the product, and preparation is the one variable you fully control.

How to use the result

Work the weakest checks in weighted order, not in the order they appear on screen. A No on a positioning check is worth more than a No on an ops check, because positioning carries half again the weight and the downstream effects are larger. Working by hand, multiply the points you are missing in each dimension by that dimension's weight and start at the top of the resulting list.

Do not chase 100. The last fifteen points cost more than the first fifty and buy less. A team that moves from 48 to 72 in three weeks has changed the outcome of the launch. A team grinding from 88 to 96 is polishing, which is a comfortable way to avoid shipping. Set a target band rather than a target number.

Rescore after the work is done, then again a month after launch. Scores drift on the visibility and search dimensions in particular. The Digital Bloom found that content untouched for more than three months is over three times more likely to lose AI visibility, so a dimension you scored well in April can be materially worse by August with no change on your side.

What this framework will not do for you

Three limitations, stated plainly.

Self-assessment is generous. People score their own positioning about a band higher than an outsider would. If you want the number to be useful, answer Partly whenever you find yourself constructing an argument for Yes. Better still, have someone outside the company answer the positioning and conversion sections while you stay out of the room.

The weights are opinions. They are informed opinions drawn from watching launches go wrong in repeatable ways, but a hardware launch, a regulated fintech and a developer tool do not really share a weighting. If you disagree with the AI visibility weight, reweight it and rerun the arithmetic. The dimensions are more durable than the numbers attached to them.

It does not know your market. The framework has no view on whether your category is growing, whether a funded competitor is about to launch the same thing, or whether your timing is wrong. A product can score 90 and fail because the market did not want it in 2026, and no self-assessment catches that.

Two situations suit it badly. A feature launch inside an existing product inherits its positioning and demand, so the real questions there are about migration and communication. And when there is serious money on the line it does not replace a paid audit, because a scoring tool cannot ask the follow-up question an experienced reviewer would. For a first launch it gives you a defensible order of work three months before you need it. The rest of the launch guides go into each dimension in more depth.