Designing Specialty Intake Flows on WhatsApp: IVF, Orthopaedic and Dermatology Compared

This post is for clinic managers and marketers who already have WhatsApp live for enquiries and want to collect better information before a human takes over. It assumes you know how to set up the channel; if not, start with our walkthroughs on WhatsApp Business API for clinics and WhatsApp marketing for healthcare. Here the question is narrower: how should the intake conversation be designed, and why does the right design differ for fertility, orthopaedic and dermatology practices?

An intake flow has one job, to give the coordinator enough to make the next call useful without making the patient feel interrogated. Too few questions and the coordinator starts again from scratch. Too many and patients drop out at question four. The balance depends on what the specialty needs to know before a first conversation, how sensitive the subject is, and whether media such as photographs or scans help.

We have deployment data from our own work on drop-off by step, but we do not publish it here, and it would not transfer to your clinic anyway. Instead the post gives the design logic for each specialty and a method for measuring drop-off in your own flow.

Two ways to build the flow

Conversational questions. The bot asks one question per message and the patient replies in free text or taps a button. It is flexible and feels natural, but each step is a separate message, and free-text answers need parsing.

A structured form inside WhatsApp. WhatsApp Flows lets a business show a multi-screen form within the chat. Meta's Flows documentation lists components, endpoint, encryption and monitoring guides. Its media upload guide says the photo picker can take the camera or gallery, supports up to 30 files at a default of 25 MiB each, and that uploaded media are encrypted and held on WhatsApp's CDN for up to 20 days, so your endpoint must download and decrypt them promptly. It also warns that WhatsApp does not guarantee customer files are free of malicious content, which matters for a clinic that opens patient uploads on staff computers.

In practice, use a hybrid: conversational opening to qualify, then a structured Flow where you need several fields and a photo or document, then a human handover. Check Meta's current documentation for component and version requirements before you build.

Principles that apply to every specialty

• Lead with the patient's goal, not the form. The first message acknowledges why they wrote.
• Ask only what changes the next step. For each question, name what the coordinator does with the answer. If nothing, remove it.
• Order from easy to sensitive. Name, city and preferred time come before medical history.
• Offer skip or "tell me later". Optional is better than abandoned.
• Say how the information will be used and who sees it, in the first screen.
• Never give clinical advice in the flow. It collects and routes; the clinical team advises.
• End with a clear promise, such as who will contact them and when, and keep it.

Consent and data

The information collected in intake is health-related personal data, and often sensitive in context. India's government explainer on the DPDP Rules 2025 says a Data Fiduciary must give a clear notice explaining the specific purpose for which data is collected, and lists purpose limitation and data minimisation among the Act's principles. So build in a short notice step at the start, record the consent with a timestamp, and collect only what the clinic will use. Your legal team must approve the notice. Our compliance and opt-in management service and the WhatsApp healthcare guide cover related practice.

Photographs and scans raise the stakes. Decide in advance who may view them, where they are stored, how long they are kept, and how they are deleted.

IVF and fertility: privacy first, few questions, no media by default

Patients writing to a fertility clinic are often anxious and private. Some are writing without telling family. The flow should respect that.

Question order:

1. Consent and privacy notice, in plain language.
2. First name and preferred way to be contacted, including a safe time and whether voice calls or messages are fine. This is a courtesy and a risk control, as an unexpected call can disclose a patient's enquiry to someone else.
3. Who is enquiring: a woman, a man, a couple, or someone helping.
4. City, and whether they can travel for visits.
5. Stage, as a simple choice: exploring options, already treated elsewhere, referred by a doctor.
6. Preferred consultation mode: in person or video.

What to leave out: detailed medical history, test values and years of attempts. Collect those in a clinical consultation or a secure portal, not in a chat.

Media: do not ask for reports in the first flow. If previous reports matter, offer to collect them after a human contacts the patient and agrees a method.

Compliance care: fertility treatment is regulated and the content you collect and say may be constrained by law and professional rules. Ask your legal advisers before adding any question about, for example, preferences that touch sensitive regulatory areas. Do not suggest outcomes or success rates in the flow.

Where drop-off tends to occur (a hypothesis to test, not a reported result): at the question that feels medical or intrusive, and at any step requesting a phone number if the privacy assurance was weak.

Orthopaedics: structured and visual

An orthopaedic enquiry is often about a joint, a mobility problem or an injury. The coordinator needs to route to the right sub-specialty and, ideally, see imaging.

Question order:

1. Notice and consent.
2. Name, city, preferred time.
3. Which area: knee, hip, shoulder, spine, hand or foot, ankle, other. Use a list.
4. What best describes the reason: pain, injury, replacement consultation, second opinion, follow-up.
5. How long, in broad bands, so the patient taps instead of types.
6. Optional: upload of existing scans, X-ray images or reports.
7. Mobility or insurance preferences, if relevant to routing: cashless or self-pay.

Media: imaging and reports are valuable and frequently available on the patient's phone as photos or PDFs. Use the document or photo picker for them, and tell the patient it is optional. Request upload after the qualifying questions, not before.

Clinical safety: include an urgent-problem branch: a recent severe injury, inability to bear weight, or numbness are the kind of points your clinicians may want to route to immediate human contact. The clinical team defines these. Our AI chatbot service page covers how escalation to staff fits the wider build.

Where drop-off tends to occur (hypothesis): at the upload step if files are hard to find, and at a long free-text description.

Dermatology: photographs are the point, and the risk

Skin concerns are visual, and a photograph speeds up triage. But images of the body are among the most sensitive data a clinic handles, and cosmetic and aesthetic dermatology carries advertising and consent constraints.

Question order:

1. Notice and consent, with a specific line on photographs: what they are for, who sees them, how long they are kept.
2. Name, city, preferred time.
3. Concern type, as a list: acne, pigmentation, hair, nails, rash or itching, skin growth, cosmetic procedure, other.
4. Duration, in bands.
5. Optional photo upload, with guidance on lighting and distance, and a statement that the patient should not send images of intimate areas through chat. Offer an in-person route for those.
6. Previous treatment, as a short free-text or a skip.

Media: make the photo step explicitly optional and separately consented. Photos should go to a controlled store, not remain on a staff member's personal phone. Disable auto-download in staff devices where possible.

Compliance care: do not display or reuse patient images in marketing without explicit written consent, and see our note on before-and-after content for cosmetic practices.

Where drop-off tends to occur (hypothesis): the photo step, especially where guidance is missing or the consent line is alarming.

Comparison at a glance

Measuring drop-off in your own flow

You cannot improve what you do not measure, and every step needs an event.

1. Name each step in your flow: notice, name, city, concern, duration, upload, final.
2. Log an event when a step is shown and when it is completed, with a timestamp and a flow version.
3. Compute the funnel: the share of starters who reach each step. Calculate completion, and the median time per step.
4. Segment by entry source (click-to-WhatsApp ad, website button, QR code), because intent differs.
5. Look at the biggest single drop, not the average. Then change one thing, such as the wording, order or whether the step is optional, and compare versions with enough volume to see a real difference.
6. Pair numbers with transcripts. Read twenty abandoned conversations. They explain more than the percentage.

A tool such as AiTextBotz, which lets patients ask questions on WhatsApp and hands over from AI to your team, can host such a flow; whichever platform you use, insist on step-level events.

Where to start

• Pick one specialty and write the question list, with the purpose of each question next to it.
• Draft the notice and consent text and get legal approval.
• Build the flow in the order above, with media optional.
• Instrument every step before launch.
• Review abandoned chats weekly and change one variable at a time.
• Keep the handover promise: respond within the time you stated.

For related services, see intake form automation and the WhatsApp healthcare guide.