Denver October 2024 to now

Cutting more than 40% from the cost of every photo the product generates

An AI tool that turns a handful of uploaded selfies into a set of studio-quality portraits.

web app ux/ui
website design
webflow website
ai image pipeline
programmatic seo

The Bettershot website on a laptop, showing the LinkedIn photoshoots headline beside a grid of generated headshots

40%+: lower cost for every photo the product generates
64 pages: running on eight page types we built, out of 90 live on the site today
"legitimated our startup": what the co-founder said the work did for the company

The challenge

Bettershot was bought, not founded. Two marketers acquired the tool on Tiny Acquisitions and ran it without outside investment, and by mid-2024 it worked: $29 bought 120 headshots inside three hours. What it did not have was a reason for a stranger to trust it with a folder of photos of their own face. The site was a homepage, a few comparison pages and a blog that had barely started. It read as a tool someone was trying out, which is a hard thing to sell against Aragon, HeadshotPro, Secta, BetterPic and Photo AI, all chasing the same searches and all making the same promise. Two problems pulled against each other underneath that. People judge an AI portrait on whether the face still looks like theirs, so output quality was the entire argument. But every image the product makes costs real money to make, and 120 images an order tied the cost of a sale to volume rather than to what the customer valued. Improving the photos and improving the margin turned out to be the same piece of work. Sitting under both was distribution. A product competing on those terms has to be findable, and there was almost nothing on the site for search to index.

The Bettershot homepage on a laptop, with the professional photoshoots headline above a row of photoshoot styles

A Bettershot gallery of generated winter photographs, each with a thumbs-up count, above share and download controls

What we did

1Redesign the paid photoshoot flow
The web app is where the money changes hands, and it asks for something unusual before it delivers anything: a folder of photos of your own face. We designed that experience end to end, from the first upload through choosing outfits and backgrounds to getting the set back. We also added a step the product did not have, live from November 2024. A customer can share a finished set so friends and colleagues vote on which shot is best. That takes the awkward part of the purchase, judging your own face, and makes it something you do with other people. It also puts the product in front of someone who has not heard of it.
2Rebuild the website in Webflow
We designed the site and implemented it in Webflow, which is still what bettershot.ai runs on today. The rebuild changed what the site was for. The old one explained a tool. The new one had to make a company running on its own revenue look like a business you hand your photos to, which is the job Rachelle Brempong describes when she says it legitimated the startup. Webflow also mattered for what came after. The founders are marketers, so every page and every line of copy had to stay theirs to change, and the page system we built on top of it had to be theirs to extend without waiting on us.
3Make each photoshoot cheaper to produce
Generation cost is the whole margin in a product like this, and it scales with every single order. We reworked the pipeline on three fronts. Fewer generations per order, better prompts behind them, and a flow structured so the work happens where it changes the result. The cost of producing a photo came down by more than 40%, and the output got better rather than worse. The change shows on their own pricing. In August 2024 an order was 120 headshots delivered within three hours. Today it is 32 or 40 images delivered in minutes. Fewer photos, better ones, produced faster and for less.
4Build the search page system
With no brand recognition and a category full of competitors chasing the same terms, Bettershot had to be found. We built page templates rather than a pile of one-off pages, so opening a new vertical or a new city is a content entry instead of a design job. Eight page types now run 64 of the 90 URLs on the site: headshot verticals including LinkedIn, lawyer, teacher, military, ERAS and AMCAS, six US cities, comparison and alternatives pages against every named competitor, plus guides and use-case articles. Each keeps its own title and description. The 17 articles on the blog are ours too, written with an AI agent in the loop.

The Bettershot website's three-step explainer: upload a few photos, select outfits and backgrounds, then view, edit and download

The call we made

We went the other way on the one number the whole category advertises. Fewer images per order, better ones, with the saving spent on output quality and on a page system that gets the product found.

The obvious play in a saturated market is to match it and then outspend it. More photos for the money, and paid traffic on the terms everyone else is bidding for. Neither half was available here. Bettershot was bought rather than funded and runs on its own revenue, so outbidding better-capitalised competitors was never realistic. And volume pricing ties the cost of a sale to output, which makes the margin worse exactly as the business grows. Cutting the set paid twice instead. A smaller set can be a better set, and it costs a fraction to produce, which is where the 40% came from. Attention then had to be earned rather than bought, and that is what the page system is for.

What we built

Platforms: Web
Stack: Webflow
Design system: Eight page types behind 64 of the 90 pages live on the site
Timeline: October 2024 to now, still running

What’s next

We are still working together. The engagement started in October 2024 and has not ended, and Bettershot brings us back as new work comes up across the product, the site and the page system.