Best Practices

This guide covers practical habits that improve generation quality, keep client records organized, and make your portfolio work harder over time. These are not rules -- they are patterns that reduce friction and produce better outcomes consistently.

Photo Quality: The Single Biggest Factor in Generation Results

The AI can only work with what it is given. Portrait quality has more impact on generation output than any other variable, including the style prompt.

Checklist for a strong portrait:

• Front-facing, with the face centered in the frame
• Even, natural lighting -- avoid harsh shadows across one side of the face
• No sunglasses, hats, or accessories covering the hairline or forehead
• Client's current hair visible and unstyled where possible -- pinned-up or covered hair reduces accuracy
• Plain or neutral background -- busy backgrounds occasionally bleed into the generated result
• Minimum resolution of 800 x 1000 pixels -- blurry or heavily compressed photos produce inconsistent results

In-chair photo tips:

• Position the client facing a window or soft light source rather than away from it
• Take the photo at eye level, not from above or below
• If using a phone, tap the client's face on the screen to set focus before capturing
• Natural daylight or a ring light produces the most consistent results

What to avoid:

• Screenshots of other photos (double-compression degrades quality significantly)
• Photos with heavy filters or skin-smoothing applied
• Angled or profile shots as the primary portrait input

Style Prompts: Structure Produces Consistency

Vague prompts produce unpredictable results. Specific prompts produce repeatable ones.

Prompt elements in order of impact:

1. Length -- the single most important descriptor. Be explicit: chin-length, collarbone-length, mid-back, not "medium" or "long."
2. Cut shape -- blunt ends, layered, tapered at the nape, disconnected undercut
3. Texture and movement -- straight, wavy, curly, voluminous at the crown, flat and sleek
4. Fringe -- no fringe, curtain bangs, blunt fringe, wispy fringe
5. Part -- center part, deep side part, no part
6. Color -- if not using a color preset, describe it here: warm chestnut brown, platinum blonde, deep auburn

Example of a weak prompt: "Nice medium length hair with some layers and a natural color."

Example of a strong prompt: "Collarbone-length lob with soft, blended layers, slight wave through the mid-lengths, curtain bangs parted in the center, warm medium brown."

Using the Add Details field correctly: Add Details appends to a base style. Keep it focused on one or two specific adjustments. Do not rewrite the entire style in the refinement field -- use Describe from Scratch instead if you need full control.

Custom Styles: Building a Library That Saves Time

Custom styles pay off over time. The investment is small upfront; the return compounds with every consultation.

When to create a custom style:

• You generated a result you were happy with and want to reproduce it reliably
• You have a go-to recommendation you make repeatedly
• A client returns for the same service and you want one-tap access to the exact prompt that worked

Best practice for naming custom styles: Use names that describe the outcome, not the technique. "Soft Lob with Curtain Bangs" is more useful at the chair than "Style 04" or "My Preset."

Maintaining your library: Delete or update custom styles that no longer represent your current recommendations. A bloated My Styles tab with outdated presets slows you down. Aim for a focused library of styles you actually use.

Client Records: Consistency Compounds

A well-maintained client record is a consultation shortcut. A poorly maintained one is just noise.

At every consultation:

• Confirm the correct client is active before generating
• Add a style name when saving a look -- unnamed looks are hard to identify later
• Add at least one or two tags before closing the session
• Write a brief note if anything notable came up: texture concerns, chemical history, client preferences

After the appointment:

• Add the final photo while it is fresh -- it takes 30 seconds and completes the record
• If the look is portfolio-worthy, publish it immediately rather than planning to do it later

Tags: treat them as a system, not an afterthought

Decide on a tag vocabulary before you start and stick to it. Mixing balayage, Balayage, and colour-balayage across different records creates three separate filter buckets that should be one.

Suggested tag categories to cover:

Use Settings > Tags to clean up duplicates as they accumulate.

Portfolio: Publish With Intention

Your portfolio is your strongest marketing asset. Treat it accordingly.

Only publish looks that represent the quality you want to attract. A portfolio with 5 strong entries is more effective than one with 30 inconsistent ones. Be selective.

The strongest portfolio entries have three things:

1. A clean, well-lit original portrait
2. An AI preview that accurately represents what was discussed
3. A final photo showing the real outcome

Build toward this standard for every look you publish. Looks without a final photo are acceptable early on, but the goal is to close the loop on every published entry over time.

Publish promptly after adding the final photo. The momentum of a fresh result is the best time to publish. Leaving looks in draft while planning to publish later means many never get published.

Keep your portfolio current. Remove looks that no longer represent your current skill level or style direction. Your portfolio should reflect where you are now, not where you started.

Appointment Prep: Set Expectations Before You Send the Link

Prep links work best when clients know what to do with them.

Suggested message to send with the prep link:

"Hi [name], here is your prep link for our upcoming appointment: [link]. It takes about 5 minutes -- upload a photo, try a few styles, and submit your favorites so I can review them before you come in. If you need help getting started, there is a guide here: salonvision.ai/help/client-prep-guide. Reopen the same link before your appointment to check for any notes I leave you."

Set a realistic generation cap: A cap of 3 to 5 is enough for most clients. Higher caps are appropriate for indecisive clients or complex color consultations. Caps that are too high consume quota without producing better outcomes -- most clients settle on a direction within the first few generations.

Check prep submissions before the appointment, not during it. The value of prep is arriving already aligned. If you first open the prep thread while the client is sitting in the chair, you lose most of that benefit.

Salon Teams: Habits That Protect Everyone

For stylists on salon workspace seats, two habits matter more than anything else.

Always confirm your active context before starting a consultation. A client created under the salon context belongs to the salon. A client created under your personal context belongs to you. Check the context switcher in the header before selecting or creating a client, especially when switching between personal and in-salon work in the same session.

Communicate context to new team members at onboarding. The data ownership distinction between personal and salon context is not obvious to new users. Salon owners should explain it explicitly when adding new seats -- not leave it for the stylist to discover after the fact. Point new team members to Getting Started and Salon Workspace on their first day.

What Comes Next

• Troubleshooting -- fixing issues when they arise
• Define Style Options -- full prompt strategy reference
• Portfolio -- publishing and managing your public page