AnjHub Free Trial
AI & Technology

Let Clients See the Haircut Before They Book: A Salon Owner's Guide to AI Hairstyle Previews

Let Clients See the Haircut Before They Book: A Salon Owner's Guide to AI Hairstyle Previews

It's 3:15 on a Thursday. She holds her phone up to you with a picture of a blunt bob and honey balayage, and says the sentence every stylist knows: "Something like this, but not so... short. And maybe warmer?" You ask three questions, pull out the color chart, explain what her hair will and won't do. She nods. Then: "Let me think about it." The chair stays open.

That moment isn't a pricing problem and it isn't a skill problem. It's a translation problem. She can see a picture; she can't see herself in it. And most of the software sold to salons does nothing at all about those ten minutes.

The friction is before the appointment, not after it

Here's what gets lost in a lot of "AI will fix salons" write-ups: clients are mostly happy once they're in the chair. MODERN SALON's own color research found 98% of salon color clients describe themselves as happy or very happy with their haircolor. So any vendor leading with "everyone hates their haircut" is selling fear, not software.

The friction sits earlier, in the guessing. A 2014 survey commissioned by Toni & Guy (1,000 Americans, reported by This is Money) found half of respondents had regretted a hairstyle at least once, and half of the women surveyed said they chose not to tell their stylist they disliked a cut because they didn't want to hurt their feelings. That's an old, brand-commissioned poll, so please don't quote it as an industry rate — but every owner recognizes the behavior. People guess, they go quiet, they put off the decision.

Then there's the reference photo itself. Salon educators have been writing about the "Pinterest vs reality" gap for years: different density, ring lights, a fresh blowout, extensions, and now inspo pictures that were generated by AI in the first place. One stylist sums it up neatly — "Pinterest is inspiration, not a promise." A good consultation turns a photo into a plan. The hard part is doing that with words alone while your next client is already waiting.

Rule of thumb: a preview isn't there to prove what the hair will look like. It's there to give both of you something concrete to say yes or no to.

What's actually out there (and why most of it isn't built for your chair)

I went through the tools an owner or a client is likely to run into. Three buckets, and the differences matter a whole lot more than the adjectives on the box.

1. Consumer try-on apps

L'Oréal Professionnel's Style My Hair was a 3D hair color app — shades followed the natural flow of your hair, and at the end it pointed you toward a L'Oréal salon. No cut simulation, no booking, and it needed an install; the iOS listing was pulled from the App Store in October 2024 according to AppBrain's archive. FaceApp has hairstyle filters behind a Pro subscription (ZDNET's walkthrough puts it around $10 a month), but it's a selfie toy for social, not a consultation record. Old-school web tools like TAAZ ask you to manually rotate a hair piece over a photo — a 2D overlay, not a re-render.

Common thread: the client does it alone, on her own phone, and the result never reaches your booking page. If she has to download an app in the middle of a consultation, you've already lost her.

2. Enterprise try-on APIs

Perfect Corp / YouCam is the serious one. Its AI Hairstyle API explicitly distinguishes generative re-rendering — rebuilding hair and lighting while keeping the face — from traditional AR color overlays, and it names salons as a fit. You can take it as an API, an SDK, or a hosted page. But it's a component you integrate and pay for in credits, its page doesn't publish a per-call price, and its natural home is a product page or a brand microsite rather than the ten minutes at your station. Good tech. Just not something a two-chair shop is going to wire into its booking flow on a Tuesday.

3. The booking platforms you already pay for

This is the part that surprised me. Vagaro's own AI page says it plainly: "AI isn't going to cut and color hair." Their AI is for marketing, the front desk and operations. Booksy's AI is a receptionist. Mindbody's is an assistant / front-desk support. Fresha's client tools are consultation forms and client photos. I could not find a native generative hairstyle try-on on any of their published feature lists. Client cards that store a photo of a past cut are useful — they just aren't a preview of a future one.

So if you want this inside the consultation, you either bolt on an API or it has to be built into the client portal you already use. That gap is the whole reason we put it in AnjHub.

Five things a preview tool actually needs in a salon

How AI Look works inside AnjHub

There are three ways into it, depending on who's holding the screen:

The flow is deliberately short. Take or upload a front-facing photo in even light (JPG / PNG / WEBP; very large files get rejected up front), then pick from 21 presets: seven cuts (Bob, Long Bob, Pixie, Long Layered, Bangs, Curly, Blunt), eight colors (Natural Black, Chestnut Brown, Honey Blonde, Copper Red, Balayage, Ombre, Ash Blonde, Burgundy) and six updos (Classic Bun, Ponytail, Side Braid, Chignon, Top Knot, Half Up). There's an optional ethnicity / country selector and a 240-character custom prompt in any language, which is where your own words go — "keep the length, soften the fringe, warm the tone". Out comes a Before / After pair you can share straight to the client's phone.

Face-shape analysis is a seperate little feature and, to my mind, the more interesting one technically. It runs on the device using MediaPipe face landmarks, classifies oval / round / square / heart / diamond / oblong, and only sends back the shape and the ratios — the photo itself isn't uploaded for that step. A repeat analysis for the same client on the same day is charged once. Six shapes won't replace a stylist's eye, but it gives a nervous client a starting vocabulary.

On the generation side, the server applies an explicit scope guard: it may change hair and only hair. Facial features, face shape, skin tone and texture, ears, neck, body, hands, clothing, jewelry, glasses, hats, background, lighting, color grading and framing are all off-limits. That constraint matters more than any model name, because a client who sees her own face slightly "corrected" will never trust the preview again.

And the money part, in one sentence: $10 buys 100 previews, one-time, no subscription and no expiry. One generation costs one credit, a failed generation costs nothing, and the charge comes off the client's own balance first, then the shop's shared pool if needed. Shops can switch AI on or off, hand out sub-quotas to individual stylists, and see a log of every generation. The feature is off by default, so someone has to enable it before the entries appear. You can see how it sits alongside everything else on the plan page.

Where I'd draw the line — and say so out loud

No AI preview predicts the finished cut. Vendors and independent reviewers both list the same failure modes: hairlines getting "filled back in", 4a–4c texture being loosened or straightened, identity drift on deeper skin tones, side lighting, extreme angles, glasses and hats, color accuracy, profile shots. None of them tell you how a style behaves on day three, how it grows out, or how long it takes to style at 7am. Any preview is a direction and a tone reference. Say that sentence to the client before you show her the image and you'll sound like a professional instead of a salesperson.

Privacy deserves the same plain talk. The face-shape check runs locally; the photo used to generate the preview does go to a backend image model, because that's what generating an image means. Ask before you take it, use a front-facing shot the client is happy to share, and don't upload anything she'd rather keep on her phone. That's just good hygiene with any AI tool, ours included.

One more small honesty: the little progress bar is a front-end animation, not a live percentage from the model. Give it a minute on a busy day instead of promising "30 seconds".

A five-minute consultation script that actually uses it

  1. Ask first, show second. "What are you trying to change — length, color, or how much time you spend on it?" You'll get a better answer than from a photo alone.
  2. Show two or three options, not twenty-one. A wall of previews just creates a brand new version of decision paralysis.
  3. Name what the image can't show. Density, texture, grow-out, upkeep, and what your water or her previous color will do to the tone.
  4. Use the shared image to close. "If this is the direction, I'd book two and a half hours, in two sessions." Then put it on the calendar while she's still looking at it.
  5. Keep the reference with the client record. Next visit you start from that image instead of starting over.

None of this replaces the consultation. It just gives the consultation a shared picture, which is a heck of a lot faster than a shared vocabulary you have to build from scratch every time.

What to measure after a month

I'd skip the vendor case studies. There isn't good independent data on AI try-on changing salon booking conversion, redo rates or no-shows — most of the numbers floating around come from a makeup e-commerce test or a supplier's own blog. Instead, watch your own book: how many "let me think about it" consults turn into a booked slot on the spot, whether clients arrive pointing at the same image you agreed on, and whether you're spending less time re-explaining a look at the second visit. Those are your numbers, and they're the only ones that matter.

The bottom line

Consumer apps are for playing with your own face on a Sunday night. Enterprise APIs are for brands with a dev team. The booking platforms, for now, use AI to run the front desk rather than the consultation. If a preview is going to earn its keep in a real shop, it has to sit in the same place the client already books, cost a fixed amount per preview, leave the face alone, and be upfront about what it can't show. That's the idea behind the client-facing side of AnjHub — the preview lives next to the appointment, not in a different app.

And if you only take one thing from this: the goal was never a perfect picture. It was getting her off "let me think about it" and into the chair, with both of you looking at the same idea.

AI hairstyle try onsalon consultationhair color previewsalon softwareclient expectations

Ready to get booked online?

Start your free 14-day trial — a professional website with built-in booking.