• Introduction

Data entry

A person has something to give the system — a date, a name, a measurement — held in their own loose, human form. Data entry is the move of getting that expression in, across whatever surface hosts it: a form field, a conversational turn, a control rendered in the inline interface, a value typed into a command menu bar. It resolves the tension between the messiness of human expression and the system’s need for structure, capturing intent with minimal effort — frictionless by default.

Strategies

There are two main strategies for handling input, depending on the nature of the data and the system’s capabilities.

Forgiving format

Permit users to enter inputs in a variety of formats and syntax, and have the system interpret it intelligently.

  • Data flexibility: Data can be represented in multiple valid ways (e.g., dates: “tomorrow”, “12/12”, “Dec 12”).
  • Error prevention: Users frequently make formatting errors that a computer could easily fix (e.g., phone numbers, credit cards).
  • Unified input: Users might enter ID, name, date, or status in a single field.

TODO: Example (date picker accepting “next friday”)

Structured format

The structure of the control gives the user a clue about what kind of input is being requested. Use this when ambiguity is dangerous or the format is strict.

  • Rigid expectations: The format is familiar and well defined (e.g., credit card information, local telephone numbers, license strings).
  • Error reduction: Breaking data into chunks (2-5 characters) makes it easier to transcribe or memorize.

TODO: Example (also date?)

AI & verification

When shifting from “user does the work” to “AI does the work, user verifies”, a verification loop is critical to ensure accuracy and build trust.

  1. Extraction: System normalizes unstructured input (OCR from image, transcript from voice, loose text).
  2. Confidence: System flags low-confidence interpretations.
  3. Review: User confirms or corrects the data.
  4. Rationale: In complex tasks, the user verifies why the system made a choice, maintaining ownership of the decision.

Affordances

Helper patterns that guide the user towards successful entry.

Input hints

A phrase or example that explains what is required or gives additional detail about what is being requested.

Hints can be provided in two ways:

  • Helper text: Text below or near the input.
  • Placeholder text: Text inside the control (if supported), though this disappears when typing.

Placeholder

Prefill a field with an example input or instructional text that helps the user with what to do or type.

When to use:

  • No good default: When you can’t guess a reasonable default value but need to explain the field.
  • Clarification: When the purpose or format of a control isn’t immediately clear from the label alone.
  • Input - the basic component used.
  • Select - picking one value from a short, bounded set when the actor already knows the answer.

Resources & references

  • Tidwell, Brewer, Valencia (2020) Designing Interfaces, 3rd ed.
  • Reicherts et al. (2025) AI, Help Me Think—but for Myself (arXiv:2504.06771)

Related patterns

Enacts

  • Formality — every data entry field is a choice about how formal the input has to be

Complements

  • Validation — the entry the message responds to; a forgiving format prevents errors this move would otherwise have to report

Alternatives

  • Good defaults — use when you can make a good guess to avoid data entry entirely.

Related

Preceded by

  • Localization — locale-aware input parsing and format tolerance

Enabled by

  • Autocomplete — autocomplete as a way to reduce input friction
  • Autofill — autofill as a strategy for frictionless input
  • AI completion — AI completion as the AI-driven end of the input strategy spectrum

Hosted by