DIREKT Phase 1A Assumptions Register

Status vocabulary

  • UNTESTED — no adequate evidence collected;
  • PARTIALLY_SUPPORTED — some evidence supports the assumption, but critical gaps remain;
  • SUPPORTED — evidence is sufficient for the controlled next phase;
  • REJECTED — evidence contradicts the assumption;
  • DEFERRED — not required for the current product decision;
  • LEGAL_REVIEW_REQUIRED — no implementation decision may rely on it until qualified review.

Confidence is recorded as LOW, MEDIUM or HIGH. Confidence is not a substitute for the status.

Critical assumptions

ID Assumption Risk if wrong Required evidence Status Confidence Owner/action
A-001 Customers have a significant problem finding legitimate nearby service providers DIREKT solves a weak problem customer interviews, recent hiring journeys, incident examples UNTESTED LOW interview at least 12 customers
A-002 Reliable service-area and location information materially improves provider selection location becomes costly but low-value task observation, concept test, wrong-location incidents UNTESTED LOW compare location card variants
A-003 Customers understand check-specific trust claims better than a generic verification badge trust UI may confuse or overpromise comprehension tests with synthetic claims UNTESTED LOW test at least 3 trust-state variants
A-004 Customers will inspect verification details before contacting a provider evidence work may not influence behaviour prototype task tests, interview recall UNTESTED LOW record what participants inspect first
A-005 Providers will submit identity and operating-location evidence provider supply may be too low provider interviews, evidence walkthroughs UNTESTED LOW test concerns, friction and refusal reasons
A-006 A meaningful share of legitimate providers has usable qualification, trade or registration evidence formal-evidence model may exclude viable providers category document walkthroughs, issuing-body research UNTESTED LOW classify evidence by category
A-007 Informal providers can be represented honestly without being labelled unsafe or fully verified platform may exclude the majority or mislead customers provider/customer tests, trust-language review UNTESTED LOW design partial-verification states
A-008 Field verification is affordable and operationally repeatable unit economics may fail field-agent interviews, route simulations, cost model UNTESTED LOW estimate time/cost per visit
A-009 Lusaka or another dense urban area is the correct first pilot pilot may be expensive, unrepresentative or too broad candidate-area scorecard, recruitment feasibility UNTESTED LOW compare at least two areas
A-010 Four or fewer initial categories can demonstrate the model scope may be too broad or too narrow demand, evidence, safety and recruitment scorecards UNTESTED LOW approve category set explicitly
A-011 One Android app can support customer and provider modes without confusing users navigation and identity complexity may increase role-switch concept tests UNTESTED LOW test customer-only, provider-only and dual-role participants
A-012 Direct calls or external messaging after a tracked enquiry are sufficient for MVP accountability may disappear off-platform journey interviews, complaint expectations UNTESTED LOW compare direct handoff with in-app messaging
A-013 A subscription or verification fee can be justified by provider value business model may not sustain operations provider pricing research, behavioural pilot UNTESTED LOW avoid final price before pilot
A-014 Customers and providers can use a native Android app under realistic data and device constraints product may fail on target devices device inventory, network observation, prototype performance UNTESTED LOW record OS, storage, camera and data constraints
A-015 Phone OTP is usable and sufficiently reliable onboarding may fail phone/SIM behaviour, vendor research, retry tests later UNTESTED LOW document multi-SIM and shared-phone patterns
A-016 Map pins are accurate enough for discovery and field verification wrong pins may create trust harm pin-sharing observation and field comparison UNTESTED LOW test public and private precision separately
A-017 Providers accept that exact private location evidence is hidden from customers trust may appear weaker or providers may fear exposure provider/customer privacy tests UNTESTED LOW explain service area vs private premises evidence
A-018 Official or issuer confirmation is available for key credentials DIREKT may only inspect images, not verify claims authority engagement, documented verification routes PARTIALLY_SUPPORTED LOW PACRA, NCC and TEVETA appear relevant; access method unconfirmed
A-019 A public business search can support some business-registration checks API/legal/availability constraints may block automation PACRA terms, access and reliability review PARTIALLY_SUPPORTED LOW verify permitted usage and fallback process
A-020 Construction-related categories can use NCC evidence where applicable individual trades may not map to contractor registration NCC rules and provider document sampling PARTIALLY_SUPPORTED LOW distinguish firms, contractors and individual artisans
A-021 TEVETA credentials can support some skill or qualification checks credential formats and verification access may vary TEVETA engagement and certificate samples PARTIALLY_SUPPORTED LOW confirm certificate types and validation process
A-022 Storing identity, certificate and location evidence is lawful with consent and safeguards legal exposure and user harm qualified Zambian privacy/legal review LEGAL_REVIEW_REQUIRED LOW no production collection before review
A-023 DIREKT may display verification outcomes and expiry information publicly public claims may create liability or defamation risk legal review and authority terms LEGAL_REVIEW_REQUIRED LOW define neutral, factual wording
A-024 Field agents may photograph premises and collect coordinates with provider consent consent may not resolve all legal/safety issues legal review, field protocol and participant testing LEGAL_REVIEW_REQUIRED LOW use synthetic evidence until approved
A-025 Public repository and Pages can safely support remote prototype testing feedback may accidentally collect personal data form/privacy review PARTIALLY_SUPPORTED MEDIUM use static synthetic prototypes and approved feedback route only

Secondary assumptions

ID Assumption Required evidence Status Confidence
A-026 English-first UI is acceptable for the pilot language preference interviews and usability tests UNTESTED LOW
A-027 Familiar local terms can be added without full multilingual support terminology testing UNTESTED LOW
A-028 Customers prefer list and map views rather than one format only prototype comparison UNTESTED LOW
A-029 Availability indicators are useful and providers will maintain them task interviews and pilot behaviour UNTESTED LOW
A-030 Reviews tied to tracked interactions are acceptable customer/provider concept tests UNTESTED LOW
A-031 Providers will respond to platform-tracked enquiries promptly pilot response-time data UNTESTED LOW
A-032 Customers will report stale or incorrect location information concept test and pilot behaviour UNTESTED LOW
A-033 A limited field-verification badge is commercially valuable provider value testing UNTESTED LOW
A-034 Institutional buyers need a different supplier-search view institutional interviews UNTESTED LOW
A-035 Firebase App Distribution is usable for the early tester cohort tester account/device test after package approval UNTESTED LOW
A-036 The permanent Android application ID should use a DIREKT-owned namespace brand/domain ownership and release decision UNTESTED LOW

Evidence update rule

For each status change, add a note below with:

  • date;
  • evidence IDs from RESEARCH_LOG.md;
  • decision owner;
  • what changed;
  • remaining contradictions;
  • impacted product or technical documents.

Change log

2026-07-14 — Register initialized

All assumptions remain provisional. PACRA, NCC and TEVETA official sites establish that relevant registration and certification systems exist, but DIREKT has not confirmed permitted verification access, category coverage, document formats, operational reliability or legal use. No product claim is approved from desk research alone.