skin analysis API for direct sales
Skin Analysis API for Direct Sales: The Selfie as the Reseller's Tool
How a skin analysis API for direct sales gives the reseller an objective profile of her customer — and gives the brand standardized data to feed catalog, CRM and repeat purchase.
In beauty direct sales, the person talking to the customer is not the brand: it is a network of resellers, consultants and representatives with wildly different levels of experience. A skin analysis API for direct sales solves a problem specific to that model: turning the assessment of skin tone, undertone and condition — today done by eye, differently by every person — into standardized data the brand controls, delivers inside the reseller app and reuses across the catalog, the CRM and repeat purchase.
The flaw in the model: every reseller has her own criteria
A consultant with ten years on the road gets the foundation shade right almost every time. One who joined last month gets it wrong, swaps it, and sometimes pays for the exchange out of her own pocket. The brand never sees that friction: it sees an order, a return and a reseller who disappears the following quarter.
The blind spot is that nothing was ever recorded about what was observed on the customer. The information lives in the reseller's head, at best in a WhatsApp note. When she leaves the network, the customer starts from zero with the next one.
A selfie-based skin analysis delivered via API fixes this on two levels. For the reseller, it becomes an objective second opinion during the appointment. For the brand, it becomes a structured field in the database: tone, undertone, apparent skin type and condition, hair attributes, each with its own confidence score.
Where the skin analysis API for direct sales fits in the flow
At the first appointment
The reseller opens the brand's app, starts an appointment and asks the customer for a selfie — taken right there, in person, or sent through the conversation. The photo goes to the API and comes back as JSON with the beauty profile. The app does not show the JSON: it shows the identified tone, the undertone and, right below, the catalog SKUs that match that profile.
The practical gain is that the conversation changes register. Instead of "I think you're on the lighter side," the reseller says "the photo analysis indicated a warm undertone, and in our catalog these two shades are the closest." That is easier to defend and easier for someone who started yesterday to reproduce.
In follow-up and repeat purchase
A saved profile is a reusable profile. Three months later, when the brand launches a new line, the reseller does not have to redo anything: her customer list is already tagged by tone and skin condition. The message stops being "there's a new foundation" and becomes "there's a new foundation, and shade 230 is yours."
For brands that want to automate part of this, the profile also feeds AI-assisted service. That is the role of MaIA Beauty Advisor, which reads the beauty profile and recommends products from the brand's own catalog on WhatsApp and on the product page — useful when the reseller is offline and the customer wants to decide the same day.
Architecture: one brand app, many resellers
Identity and scope
The pattern that works is simple: the API key lives in the brand's backend, never in the app distributed to the network. Every request from the app passes through your server, which identifies the reseller, records the consumption and forwards the call. That way you avoid a leaked key in an app installed on thousands of devices, and you gain a per-person log.
Usage control across the network
Credits are bought by the brand, not by the reseller. Since the backend is yours, you set the rules: a monthly quota per consultant tier, a limit per appointment, a block on resending the same photo. This control usually matters more than it seems — large networks test limits fast.
Where the profile lives
The API response is a document; its home is the customer record in the brand's CRM, with date, origin (which reseller, which appointment) and version. The reseller sees only what she needs. The brand sees the whole base and can cross profile with order history to plan inventory and launches.
Consent, transparency and limits
The photo belongs to a third party and is sent by someone who is not a brand employee. That requires three explicit things. First, the customer's consent recorded inside the app, before capture, with clear wording about purpose and retention period. Second, training for the network: the reseller needs to be able to explain in one sentence what will be done with the image. Third, clarity about the scope of the analysis — it describes visible attributes in a photo, it is not a medical assessment and does not replace a health professional.
It is also worth aligning internal expectations: a photo with poor lighting, a filter or heavy makeup returns lower confidence in some fields. Teach the network to repeat the capture when the app flags that, instead of moving ahead with a weak result.
Makeup: from reading the skin to testing the color
A skin profile answers "which shade matches." For the next step — "how does it look on me" — you bring in virtual makeup try-on. The API version receives the photo and returns the image with lipstick or foundation applied, with the palette calibrated to the brand's real products; it is the format that fits a network app well, because processing happens on the server. The browser SDK, in early access, covers thirteen products and the Full Look in live camera mode, rendered locally — an alternative for when the brand has a web version of the reseller app.
What to measure in the first cycles
Start with four indicators: the share of appointments with a captured profile, the shade-related exchange rate on orders coming from appointments with a profile versus without, average order value in both groups, and adoption by reseller seniority tier. That last one reveals whether the tool is actually leveling the network or only being used by those who already sold well.
If you want to start on the technical side, MaIA Scan's selfie-based skin and hair analysis is delivered as an API, with a model trained on Brazilian skin and a confidence score per field.
Create your account, buy credits and generate your API key in the documentation — or talk to our team to design the flow for your reseller network.
Frequently asked questions
Does the reseller need her own API key?
No. The key stays in the brand's backend and the network app calls your server, which identifies the reseller and forwards the request. That avoids an exposed key in apps installed on many devices and produces a per-person usage log.
Who buys the credits in a direct sales model?
The brand buys the credits and sets the usage rules: a quota per consultant tier, a limit per appointment and a block on resends. Since control sits in your backend, you decide how consumption is distributed across the network.
What exactly does the analysis return?
A JSON with attributes visible in the photo — skin tone and undertone, apparent type and condition, hair attributes — and a confidence score per field. It is a description of visible characteristics, not a medical assessment, and it does not replace a health professional.
How should the end customer's consent be handled?
Consent must be collected and recorded inside the app, before capture, with a clear purpose and retention period. Since the reseller is the one operating it, train the network to explain in one sentence what happens to the image.
Can the shade be tested in the same conversation?
Yes. The try-on API receives the photo and returns the image with lipstick or foundation applied, with the palette calibrated to the brand's real products. The browser SDK, in early access, extends this to thirteen products and the Full Look in live camera mode.