← All posts

The Average Launch: The State of Indie Launches, August 2026

Published 5 September 2026

Apologies for the late issue. We were busy with our own Product Hunt launch this week.

What does the average launch look like?

18,465 indie launches analysed in August 2026. This month's feature is 28,760 launch screenshots averaged into single images, alongside the top of every category's stack as a build sheet.

The average launch page is a nav bar, a button and a centred headline.

We've spent four issues looking at the technology behind indie launches. This month we looked at the pages themselves.

We averaged 28,760 screenshots, separating light pages from dark ones so the layouts would stay visible. Across product categories, the same pattern appeared: a logo at the top left, a button at the top right, and a centred headline. All sixteen category composites share that layout, and 45.5% of the pages behind them render dark.

Five things stood out

A quick word on methodology

The scored launches come from Product Hunt (84.5%), Hacker News Show HN (9.4%) and PeerPush (6.1%). A launch is counted against the board it launched on, even when our own crawl found the site first. Every launch in this set was crawled during August 2026. The crawl method is unchanged. Tech detection used 42,773 fingerprints at the end of August, up from 15,759 at the end of July, mostly through the addition of small vendors found by the discovery crawl. Full methodology at /methodology.

This issue uses three datasets, and each number says which one it comes from. The monthly cohort is 18,465 scored launches. The composites use 28,760 screenshots, sampled across the whole catalogue. The stack analysis uses every classified scored launch with at least one current detection, 70,234 launches: we map detections to the twelve layers used on /stacks, then calculate the leader and its share within each product category. How the composites were made is explained where they first appear.

Denominators, as ever: category and per-tool counts are sub-corpus counts and read lower than the all-time totals on the linked /tech pages.

The cohort

Source Launches Share
Product Hunt 15,611 84.5%
Hacker News (Show HN) 1,735 9.4%
PeerPush 1,119 6.1%
Total scored 18,465 100%

August's 18,465 scored launches compare with 19,772 in July. Show HN's share slipped from 9.7% to 9.4%.

The average launch

Here is the average light launch page: 15,670 screenshots, one image.

Composite of 15,670 light-mode launch screenshots averaged into one image. A pale page with a dark horizontal band across the top, a darker mass at the left end of the band where logos sit and a dense dark blob at the right end where the sign-up button sits, and a soft grey block of headline text centred below.

The dark bar across the top is the nav. The blob at its right-hand end is the sign-up button. Zoomed in, the top 90 pixels look like this.

The top band of the light composite, enlarged. A dark smudge at the left where logos and wordmarks cluster, a thin line of nav links across the middle, and a solid dark oval at the right end where thousands of Sign up and Get started buttons have landed on the same pixels.

Across fifteen thousand pages, the logo, headline and call to action keep landing in roughly the same places.

The dark composite has the same basic layout, plus something else.

Composite of 13,090 dark-mode launch screenshots. A near-black page with a faint lighter nav band across the top, a soft grey hero mass in the centre, and a faint pale rectangle in the middle of the page with two rounded button shapes side by side at its foot.

That pale rectangle with two buttons is a popup. We'd noticed it in July's games sample and assumed, wrongly, that it was an age gate or a "play now" overlay.

How we made the images

The idea comes from "average face" portraits: line up photographs and blend them into one image. We used screenshots captured at the same size, sampled at random from launches with a high-confidence product category (0.9 and up). We sampled up to 2,000 per category, 28,760 screenshots in all.

A plain average turns grey because light and dark pages cancel out. So we sorted the screenshots by mean brightness, averaged the two groups separately in linear light, then stretched the contrast to make the structure visible.

The composites show where visual elements repeatedly line up. Common layouts leave a blur; an identical element in an identical position can stay sharp even if relatively few sites use it.

The popup is Google's

In the 1,297 dark games screenshots, the popup we'd seen in July was still there: a bright centred rectangle with two pill-shaped buttons at the bottom.

The clue was its sharpness. Headlines blur because each site puts different text in a slightly different place; this didn't. We split the dark screenshots by whether the crawl detected a consent-management tool and averaged each group separately.

Two dark composites side by side. Left, the 11,992 dark launch pages with no consent tool detected: a plain dark page with a faint nav band and hero, no popup. Right, the 1,098 dark pages with a consent tool: the same dark page behind a sharp, legible Google consent dialog, with the headings Personalised advertising and content, Store and/or access information on a device, and two buttons reading Manage options and Consent.

On the left, the 11,992 dark pages with no consent tool detected: no popup. On the right, the 1,098 with one: Google Funding Choices, the consent dialog that ships with AdSense, legible down to the button labels.

Funding Choices appears on 4.0% of classified launches overall, including 9.4% of directories, 8.5% of games, 7.7% of design tools and 5.3% of content sites. In AI, the rate is 2.2%, and the composite has no visible popup.

The dark-mode gradient

Sorting the screenshots also gives us the share of sampled pages in each category that render dark.

Bar chart of the share of launch pages that render in dark mode by product category. Crypto leads at 66% and games at 65%, then content, social, developer tools and AI at 50% to 55%. Finance, marketing, health, design and education sit around 40%, and HR, productivity, e-commerce and directories are lowest at 36%.

Overall, 45.5% of the sampled pages render dark. Crypto (66.4%) and games (65.1%) lead. Content, social, developer tools and AI sit around half, while HR, productivity, e-commerce and directories are around 36%.

AI is almost evenly split at 50.2%. Despite the association with black backgrounds and gradient text, its pages are much less likely to render dark than crypto or games pages.

We first measured this on a hundred screenshots per category in July. With up to twenty times the sample this month, the ranking has barely moved.

The same skeleton in every category

Here are the light composites for AI (993 pages) and HR (1,271 pages), side by side.

Two light composites side by side, AI on the left and HR on the right. Both show the same layout: a dark nav band across the top with a logo mass at the left and a button mass at the right, and a centred grey block of headline text below. They are close to indistinguishable.

We could not tell them apart without the labels, and we made them. Across all sixteen categories, the same layout appears, with small shifts in colour temperature: warmer for e-commerce and health, cooler for developer tools and crypto.

Same pixels, same package.json

The technology rankings are similarly consistent. For each category, we took the leading detected technology in twelve stack layers. Ten layers have the same leader in all sixteen categories.

This grid shows eight of those layers.

Grid of sixteen product categories against eight stack layers. The column headers read Tailwind CSS, shadcn/ui, React, Next.js, Vercel, Cloudflare, Resend and Google Analytics. Every cell in a column carries the same technology name as its header; only the percentage changes, from 38% to 65% for Tailwind, 10% to 26% for Resend, and 31% to 56% for Google Analytics.

Vite and Stripe also lead everywhere, though they aren't in the grid.

The exceptions are backend-as-a-service and auth. Firebase leads thirteen categories; Supabase takes HR, crypto and directories. Google Sign-In leads fourteen, with Privy ahead in crypto and Auth.js in directories. Detected shares in these layers are low, so relatively small differences can change the leader.

The bigger variation is in adoption rates. Tailwind ranges from 38% of the "other" category to 65% of HR tools. Resend runs from 10% of games to 26% of marketing tools, and Google Analytics from 31% of social apps to 56% of e-commerce.

This resembles the split we described in May: Next.js and Vercel are strongest in HR, AI, marketing and finance; WordPress and Hostinger are more prominent in e-commerce, games and "other".

The average launch, as a build sheet

Here are the leaders across the full stack-analysis dataset: every classified launch since April with a current detection, not just August's cohort.

Horizontal bar chart of the leading technology in each of twelve stack layers across 70,234 classified launches. Tailwind CSS 53.2%, shadcn/ui 21.0%, React 35.8%, Next.js 34.6%, Vite 23.2%, Vercel 31.6%, Cloudflare 42.9%, Firebase 4.3%, Google Sign-In 4.0%, Stripe 4.8%, Resend 19.4%, Google Analytics 39.7%. The runner-up in each layer is named underneath in small type for comparison.

The gaps are large. React leads Vue.js 35.8% to 2.1%, Next.js leads Astro 34.6% to 2.4%, and Tailwind leads Bootstrap 53.2% to 2.0%.

Some of the less obvious tools also have substantial shares. shadcn/ui appears on 21% of classified launches and 27% of AI launches. Resend is detected on 19%, and Vite on 23%.

These are separate rankings, so the sheet doesn't tell us how often all these tools appear together.

Where categories differ

Below the shared leaders, some category preferences emerge. This table picks out the clearest examples: technologies with high adoption relative to their overall rate, restricted to tools detected on at least 3% of a category's launches.

Category Over-indexes on
Developer tools GitHub Pages 8.0% (2.1x), Fastly 9.2% (1.6x), Astro 3.7% (1.5x)
AI Google Sign-In 6.3% (1.6x), Clerk 3.7% (1.5x), PostHog 9.0% (1.3x)
Marketing Facebook Pixel 9.7% (2.0x), Google Workspace 22.5% (1.5x), Hostinger 9.6% (1.4x)
Finance Supabase 4.6% (1.5x), Stripe 6.2% (1.3x), Lovable 5.3% (1.3x)
Health Apple App Store badge 8.1% (3.7x), Lovable 5.9% (1.5x)
Education Bootstrap 3.6% (1.8x), Firebase 6.8% (1.6x), jQuery 6.2% (1.4x)
Content YouTube embeds 6.2% (1.3x), Google AdSense 6.4% (1.3x)
HR Supabase 5.3% (1.7x), Brevo 5.5% (1.6x), Clerk 3.9% (1.6x)
Social Apple App Store badge 4.6% (2.1x), Firebase 8.0% (1.9x), Lovable 6.3% (1.6x)
Games Google Funding Choices 8.5% (2.1x), Google AdSense 9.6% (2.0x)
Design Pinterest tag 4.8% (2.4x), Vue.js 3.8% (1.9x), Microsoft Clarity 9.6% (1.5x)
Crypto WalletConnect 14.7%, wagmi 9.9%, Privy 5.2% (all 50x and up)
Directories Plausible 7.5% (2.4x), Google AdSense 10.7% (2.2x)
E-commerce Shopify 9.7% (32x), WooCommerce 6.9% (13x), jQuery 15.2% (3.5x)
Other WordPress 18.0% (3.3x), Elementor 10.2% (3.8x), Google Maps embeds 6.2% (5.1x)
Productivity Nothing above 1.4x

Shares are of each category's classified launches; multipliers compare those shares with the overall rate. Unlinked entries are embeds or badges.

Productivity, the largest category at 11,082 launches, stays close to the overall technology mix. Among tools meeting the 3% threshold, its largest lift is the App Store badge at 1.4x.

The adoption clock

Since June we've tracked each email tool's "tech-share": the percentage of its users classified as technical builds. We're testing whether that share falls as a tool reaches a broader audience.

Email tool Users Tech-share
Cloudflare Email Sending (beta) 474 38%
Postmark 124 28%
SendGrid 1,940 27%
Mailgun 845 26%
Resend 13,605 24%
Amazon SES 2,552 24%

User counts are within the classified dataset of 70,234 launches with a high-confidence product category, rather than the all-time totals on the linked pages.

Cloudflare Email Sending rose from 0.45% of the June cohort to 0.90% in July and 1.27% in August, or 235 launches this month. Its tech-share fell from 48% to 42% to 38%. It remains the most developer-skewed email tool in the table, but that skew has narrowed in both monthly comparisons.

Resend has levelled off. Last month we reported a four-month slide from 24.9% to 22.6%. The current figures show 22.9% in June, 23.4% in July and 23.0% in August, against technical baselines of 25.0%, 25.1% and 25.3%.

July's figure rose from the 22.6% we published as late crawls came in. For now, Resend is holding roughly two points below the technical baseline.

Risers and fallers

Field July 2026 August 2026 Δ
Vercel hosting share 30.4% 29.1% -1.3pp
Tailwind CSS adoption 51.1% 49.6% -1.5pp
React adoption 34.8% 34.5% -0.3pp
Next.js adoption 34.0% 33.8% -0.2pp
Resend share of email 19.0% 17.8% -1.2pp
Cloudflare Email Sending 0.90% 1.27% +0.37pp
Named AI-builder cohort 7.3% 7.1% -0.2pp
Lovable share of named cohort 49.4% 44.5% -4.9pp
Avg Vibe Score (PH) 29.8 28.0 -1.8
Avg Launch Readiness (PeerPush) 92.7 93.3 +0.6
Avg Vibe Score (Show HN) 18.1 16.6 -1.5

Auto-derived from last month's data. Monthly cohort figures, so they sit below the build sheet's shares, which cover every classified launch since April; pp means percentage points. Anything inside about two points is noise, and recent figures can change as further crawls arrive.

Lovable's share of the named AI-builder cohort fell almost five points to 44.5%, inside a cohort that barely moved. Base44 now accounts for 193 launches, against Lovable's 588, Replit's 146 and Framer's 102. We'll check whether Lovable's lower share persists in September.

Vercel's share from April to August reads 33.2%, 32.1%, 32.2%, 30.4% and 29.1%: down 4.1 percentage points overall, with a small rise in June. Tailwind is below half for the first time since tracking began in May, moving from 53.9% to 52.0%, 51.1% and now 49.6%. Each single step is inside the noise band.

Tuesday is launch day, mostly

The recurring check, run across the whole corpus of 92,971 launches. Tuesday leads at 20.7% against Wednesday's 17.6%, the weekend is the only real cliff, and the peak is still a Product Hunt artifact.

Bar chart of indie launches by day of week across the whole corpus. Tuesday is the tallest bar, highlighted in amber, with the weekend noticeably lower.

The claim funnel

93 founders have claimed their listing and triggered a recrawl, up from 75 at the end of July. Of the 77 we could compare cleanly, 47 shipped at least one observable fix. Median scores were 7.6 out of 10 before and 9.5 after, and the typical launch gained 1.0 point (medians do not subtract).

The most-added items were a Referrer-Policy header (23), llms.txt (22), security.txt (22), a privacy policy (12) and robots.txt (5).

If you launched on Product Hunt, PeerPush or Show HN in August and have not checked your listing, find it on /browse. Claiming is free. The Launch Readiness Check runs the same analysis on any URL.

What's next

This is issue #5. The recurring shape holds: a monthly feature, the adoption clock, the deltas table, the day-of-week check, the claim funnel.

Next month is about WordPress. It is a small share of launch-board launches and the biggest single thing on the wider web, and for the first time we will put the two populations side by side: which plugins and themes each runs, who hosts them, where the sites are, what they sell, and whether the plugin stack changes by country. Issue #6, Around the World in WordPress.

Bookmark /blog.

Thanks to the PeerPush team for the ongoing integration support.

Explore the August data