DIREKT Technology and Connectivity Research¶
Status: Open; no final minimum SDK, device profile or network assumption approved.
Decisions controlled¶
- minimum supported Android version;
- target device performance class;
- app download and update size budgets;
- image capture and compression rules;
- offline draft and sync requirements;
- upload retry/resume behaviour;
- map caching and low-bandwidth presentation;
- OTP/session recovery behaviour;
- accessibility and input methods;
- remote tester distribution feasibility.
Participant data to collect¶
Collect only with consent and without device identifiers:
- manufacturer/model family or broad device class;
- Android version, if known;
- RAM/storage constraints described or observed;
- available storage band;
- camera capability and common image problems;
- screen size and font scaling;
- battery/charging reliability;
- number of SIMs and network switching;
- shared-device behaviour;
- primary connection type;
- data-bundle and Wi-Fi availability;
- frequency/duration of offline periods;
- app-update behaviour;
- digital confidence and accessibility needs.
Do not collect advertising IDs, IMEI, SIM serial numbers, contact lists or unrelated diagnostics.
Field task set¶
Use synthetic content to measure:
- open app from cold start;
- select area/category;
- load list and map results;
- inspect provider and trust details;
- create a structured enquiry;
- capture two document pages and one premises photo;
- interrupt the upload by disabling connectivity;
- resume or correct the upload;
- recover after process death or device restart;
- enlarge text and navigate with accessibility settings where relevant.
Metrics¶
Record bands rather than false precision during research:
- time to first useful content;
- search result load time;
- image upload time;
- data used per task;
- upload failure/retry count;
- storage required;
- steps requiring assistance;
- permission-denial recovery;
- completion rate;
- participant confidence;
- device heat/battery concern;
- offline recovery success.
Design hypotheses to validate¶
- list view must remain useful when map tiles are unavailable;
- location can be manually selected without GPS;
- images should be compressed before upload while preserving evidence readability;
- uploads require persistent progress, retry and action-required states;
- provider onboarding must save drafts locally;
- customer search should cache recent areas/categories without presenting stale availability as current;
- large media should load only on demand;
- text and icons must remain usable with font scaling;
- notifications are helpful but cannot be the only way to retrieve status;
- phone-number changes need recovery and re-verification rules;
- shared-device sessions require explicit sign-out and privacy protection.
Minimum SDK decision template¶
| Candidate minimum | Estimated coverage | Capability benefit | Exclusion risk | Security/maintenance impact | Evidence IDs |
|---|---|---|---|---|---|
| Pending |
The final decision must use observed target devices and current Android/Play requirements. It must not be copied from another project.
Performance budgets to approve later¶
| Budget | Research input needed | Approved value |
|---|---|---|
| Initial download size | data-cost and storage findings | Pending |
| Cold start on low-end reference device | device sample | Pending |
| Search payload | bandwidth tests | Pending |
| Thumbnail size | readability/data trade-off | Pending |
| Evidence image upload | document readability tests | Pending |
| Offline draft storage | provider workflow | Pending |
| Retry backoff | connectivity patterns | Pending |
Connectivity research questions¶
- Which mobile networks are used in the candidate area?
- How often do participants switch SIMs for cost or coverage?
- Is Wi-Fi available to providers?
- Are outages brief, frequent or prolonged?
- Can SMS arrive when mobile data is unavailable?
- Are OTP delays or recycled numbers common concerns?
- Are data bundles purchased daily, weekly or monthly?
- Do users disable background data?
- Do app stores update automatically?
- Are devices shared between business staff or household members?
Remote testing sequence¶
- static synthetic prototype through GitHub Pages;
- Android CI artifact to the core technical team;
- Firebase App Distribution after package-name and credentials approval;
- controlled low-end device cohort;
- Google Play internal/closed testing later.
Firebase distribution is not a substitute for field device coverage; testers need Google/Firebase access and may differ from ordinary users.
Exit requirement¶
Phase 1A must produce enough evidence to define constraints for Phase 1B prototypes. Final build values may be fixed in Phase 2 only after observed device data, prototype performance and current platform requirements are reviewed.