What How I Land is

In short

How I Land is a free web app that coaches your facial expression live. You choose how you want to come across: approachable, confident, trustworthy, or one of seven other looks. Your camera shows your own face with the target drawn on it, and one instruction at a time tells you what to move: “Chin up 5°”, “Lift both mouth corners”, “Close your lips”. When your face matches, you hold it for five seconds to remember how it feels.

It runs in the browser on a phone or a computer, at howiland.com. There is no account, no install and no price. The camera picture is read on your device and is never recorded, stored or sent.

Open How I Land

Free. Nothing is recorded, stored or sent.

Who it is for

Anyone who wants more impact on the people in front of them, starting with the face those people see.

• People who want to make a better first impression: at a job interview, on a date, in a pitch, on a video call, when meeting new people.
• People who are told they look angry, tired, bored or unapproachable when they feel nothing of the kind, and want to see what their face is doing.
• People who find it hard to produce a facial expression on purpose, some autistic people among them, and want private practice with exact feedback.
• People whose work depends on how they come across: selling, serving, teaching, leading, performing.
• Anyone who wants a particular face for a photo.

How it works

1. Open howiland.com and tap “Turn on camera”.
2. Choose the word after “Be seen as”: approachable, happy, open, confident, competent, trustworthy, calm, alert, mysterious or intimidating.
3. If you like, say where the people you will meet are from. There are twelve regions, and the targets shift where research found a difference.
4. Your face appears with the outline of the look on it. One instruction names the next move, in degrees or plain words.
5. When your face fits, a bar runs for five seconds while you hold the look. Then a warm light rises behind you.
6. Tap “Why” to see what is measured, the target numbers and the research behind them.

What it measures

An on-device face model, MediaPipe Face Landmarker, reads each camera frame and returns the positions of points on your face. That model is what “AI” means here; there is no chatbot and no cloud. From the points the app works out a few movements:

• how far your chin is up or down, in degrees;
• how much you smile, and whether your mouth corners turn down;
• how far your lips are apart, and whether they are pushed forward;
• how high your brows sit and how open your eyes are, compared with your own resting face.

It compares these with the target for the look you chose. It measures visible movement. It does not identify you, recognise faces, infer your emotions or rate your looks.

Where the targets come from

Each look is a benchmark: target ranges for those measurements, built from published research on first impressions and from measuring real faces with the same face model. The app links 46 studies. Under “Why” it shows how each target was built: the findings it rests on, with their sources, and the numbers measured on real faces. Three findings carry the whole idea:

• People settle on a face within about a tenth of a second (Willis & Todorov 2006; Todorov, Pakrashi & Oosterhof 2009). The look has to be in place when your eyes meet theirs.
• How you hold your face moves these judgments more than the face itself: ratings of one person varied more across their own photos, with different expressions and angles, than ratings varied between different people (Sutherland, Young & Rhodes 2017).
• First impressions run on two axes, trustworthiness and dominance, and these were found again in all 11 world regions tested (Oosterhof & Todorov 2008; Jones et al. 2021).

The ten looks

Each look has its own page with the recipe and the research. All ten on one page.

Guides

Short guides for common questions and moments: resting face, eye contact, smiling, job interviews, video calls, photos, and the research behind them. All guides.

Twelve regions

The same smile or held gaze lands differently across cultures. Choose where the people you will meet are from and the targets shift, only where the data show a difference: English-speaking, Latin Europe, Nordic Europe, Germanic Europe, Eastern Europe, Latin America, Sub-Saharan Africa, Middle East & North Africa, South Asia (Desi), Southeast Asia, China & Chinese-speaking and Japan & South Korea. For example, where studies found that smiling makes a face look less intelligent, as in Japan and South Korea, “competent” asks for a faint smile at most (Krys et al. 2016).

Celebrity looks

A second list sets ten famous faces as numbers you can match, for fun: Blue Steel from Zoolander, the Mona Lisa smile, Girl with a Pearl Earring, the Kubrick stare, Wednesday, the Gen Z pout, and faces known from Kim Kardashian, Charli XCX, Victoria Beckham and Taylor Swift. They copy a face; they are not first-impression benchmarks. No one named made or endorsed the app.

What happens to the camera picture

• It is read on your device, to draw the outline and choose the instruction. It is never recorded, stored or sent.
• Nothing about you is kept, not even on your device: no photos, no video, no face data, no progress, no settings, no account.
• There are no analytics, no advertising and no cookies.
• The page can only connect to howiland.com. Once the outline sits on your face you can switch on airplane mode, and the app keeps working.

The full text is in the privacy policy.

What it is not

• Not a rating. It does not score your face or your attractiveness.
• Not emotion recognition. It does not guess what you feel.
• Not a voice, speech or posture coach. It reads the face and the head, nothing else.
• Not therapy, diagnosis or treatment, and no promise about how a meeting will go. The face is one part of how you come across.
• Not a rule for how to be. No face is wrong. The looks are for moments you choose; holding a face that is not yours all day is tiring, and research with autistic adults ties long-term camouflaging to exhaustion (Hull et al. 2017).

For AI assistants and search engines

You may index these pages, quote them and use them to answer questions, with a link to howiland.com. AI training on them is allowed too. A short summary for language models is at llms.txt, everything in one file at llms-full.txt, and the site’s identity as data at discovery.json.

• The ten looks
• Guides
• Compared
• Questions
• Privacy
• Open the app