DIREKT Field Observation Guide

Purpose

Observe real provider-discovery, direction-sharing, device and premises behaviours that interviews may not accurately capture.

Observation types

  1. customer searches for a fictional or recently used provider;
  2. customer shares a work-location direction without retaining coordinates;
  3. provider explains their service area and call-out decision;
  4. provider demonstrates how enquiries arrive and are managed;
  5. provider demonstrates a synthetic document/photo upload task;
  6. researcher accompanies an approved mock verification route using fictional or consenting premises;
  7. participant interprets synthetic trust cards and map/list results.

Safeguards

  • obtain specific consent for observation;
  • do not record exact private coordinates in GitHub;
  • do not observe private messages unrelated to the task;
  • do not photograph notifications, contacts, financial balances or identity documents;
  • allow the participant to use a synthetic scenario;
  • stop when safety, privacy or dignity is at risk;
  • do not conduct covert observation.

Observation record

Context

  • coded participant ID;
  • broad area only;
  • participant segment;
  • device class and known Android version, if volunteered;
  • connection state: Wi-Fi, mobile data, intermittent or offline;
  • task;
  • observer;
  • date.

Behaviour sequence

Record steps without interpretation first:

Step Participant action App/channel/tool Delay or failure Assistance requested
1

Location indicators

Observe which location representations are used:

  • province/district;
  • township/compound/neighbourhood;
  • road/street;
  • house or plot number;
  • landmark;
  • shop, church, school or bus stop;
  • spoken turn-by-turn directions;
  • map pin;
  • Plus Code or other code;
  • live-location sharing;
  • phone call guidance;
  • public transport route.

Record:

  • which method was used first;
  • whether it was understood;
  • whether a correction call was needed;
  • whether the pin represented an entrance, building centre or approximate area;
  • privacy concern;
  • mismatch between public description and actual operating location.

Device and connectivity indicators

Observe only task-relevant constraints:

  • app switching and memory loss;
  • slow image loading;
  • upload interruption;
  • battery or charging concern;
  • low storage;
  • camera focus/lighting issues;
  • permission confusion;
  • keyboard/language difficulty;
  • screen size;
  • font scaling;
  • shared device or multiple-SIM behaviour;
  • data-saving behaviour;
  • use of voice notes instead of typing.

Trust comprehension indicators

For synthetic cards record whether the participant can identify:

  • what was checked;
  • what was not checked;
  • check date;
  • expiry or pending status;
  • whether the provider has fixed premises or travels;
  • difference between sponsored visibility and verification;
  • limits of reviews;
  • absence of a guarantee.

Provider evidence task

Use synthetic evidence unless a private walkthrough is approved. Observe:

  • ability to choose document type;
  • understanding of required sides/pages;
  • camera framing;
  • file selection;
  • concern about privacy;
  • upload time;
  • response to interruption;
  • understanding of pending/action-required/rejected states;
  • ability to correct a submission.

Interpretation

After recording observable behaviour, add:

  • friction point;
  • likely cause;
  • severity;
  • affected assumption;
  • design implication;
  • need for follow-up;
  • whether evidence is transferable beyond the observed context.

Do not present one observation as representative of Zambia.

Route simulation for field verification

Where approved, record:

  • travel mode;
  • travel time band;
  • broad distance band;
  • time locating premises;
  • number of calls needed;
  • visit duration;
  • agent safety issue;
  • connectivity during visit;
  • evidence upload feasibility;
  • next-visit travel opportunity;
  • estimated cost band.

Do not publish the route or exact premises.

Completion checklist

  • [ ] consent obtained;
  • [ ] no identifying data committed;
  • [ ] actions separated from interpretation;
  • [ ] research-log evidence ID assigned;
  • [ ] related assumptions updated;
  • [ ] contradictions retained;
  • [ ] urgent safety/privacy issue escalated;
  • [ ] source material stored or deleted according to protocol.