Mota — Open Questions¶

Purpose: Decisions that must be made, experiments that must be run, and assumptions that must be tested — each with an owner and a date.

Rule: an open question without a decision date is a wish. Every item below has a date and a decision owner.

1. Blocking decisions — must be resolved before Gate 0¶

#	Question	Why it blocks	Options	Decide by
Q1	What legal entity operates this?	Required for Paynow merchant onboarding and POTRAZ data controller registration	(a) Register a private limited company (~US$150–300) (b) Operate as a sole trader initially © Defer until revenue	Gate 0 end
Q2	What is the product's final name?	Store listing, domain, brand assets, and PDF footer all depend on it	"Mota" is a working title. Check trademark and domain availability for Mota, Mota ZW, Motacare	Gate 0 end
Q3	Where is data hosted, and is cross-border transfer lawful?	CDPA restricts transfers outside Zimbabwe. Supabase regions are ZA/EU/US. Getting this wrong is a regulatory issue, not a technical one	(a) ZA region + documented adequacy assessment (b) Local Zim host (costly, limited) © Legal opinion before launch	Gate 1 start
Q4	Is OTP delivery reliable on all three networks?	It is the entire authentication design	Test 20 sends per network (Econet, NetOne, Telecel); measure delivery and latency	Gate 0, week 1
Q5	What is the scope of "severe service" intervals?	Determines the reference data and therefore the product's core recommendation	Zimbabwe conditions: dusty, short trips, heavy loads. Apply manufacturer severe-service by default? Or normal with a prompt?	Gate 1 start

2. Decisions for Gate 1¶

#	Question	Options	Decide by
Q6	React Native (Expo) or PWA + Capacitor?	Expo: reliable push, store presence, heavier bundle. PWA: lighter, no store, unreliable push on cheap Android. Recommendation: Expo, because push reliability is the retention mechanism	Gate 1 start
Q7	How many vehicle models in the reference table at launch?	20 (faster, covers most of the fleet) vs 40 (better coverage, more research). Recommendation: 20 at Gate 1, expand to 40 by Gate 2	Gate 1 start
Q8	Does the PDF need a QR verification link in v1?	Adds a server-side read-only page. Small effort, meaningful trust gain	Gate 1 mid
Q9	Free-text fallback when a model isn't in the table?	(a) Allow with generic 10,000km/12-month interval (b) Block and request the model © Allow with no intervals until reviewed. Recommendation: (a) with a confidence downgrade	Gate 1 start
Q10	What's the odometer update nudge cadence?	Weekly is annoying; monthly is too sparse. Recommendation: prompt on app open if >30 days stale, max once per week	Gate 1 mid

3. Decisions for Gate 2¶

#	Question	Options	Decide by
Q11	Lead pricing: per lead or per booking?	Per booking (US$3, zero risk to garage) vs per lead (US$1, higher volume, garage bears risk). Recommendation: per booking	Gate 2 start
Q12	How many garages in the free tier's monthly lead cap?	5 is a guess. Too few and they can't feel value; too many and there's no reason to upgrade. Test 3 vs 5 vs 10 with founding garages	Gate 2 mid
Q13	Do garages want to see customer due dates?	This is the upgrade driver. If garages don't care, the subscription has no hook and we're a lead-fee business only	Gate 2 end
Q14	How is garage quality verified?	(a) Existence + contact only (b) Site visit © Customer ratings. Recommendation: (a) with a complaints process; ratings later	Gate 2 start

4. Decisions for Gate 3¶

#	Question	Options	Decide by
Q15	Subscription vs lead-fees as the primary model?	Depends entirely on Gate 2 evidence about which the garages respond to	Gate 3 start
Q16	Should the price index include crowdsourced (unverified) prices?	More data, but lower trust. Recommendation: yes, visually distinct from partner prices	Gate 3 mid
Q17	Maximum sponsored slots per category?	More revenue vs index credibility. Recommendation: 2, clearly labelled	Gate 3 mid
Q18	Annual prepay discount?	~17% (US$400 vs US$480) improves cash and reduces churn but cuts revenue	Gate 3 mid

5. Experiments queued¶

#	Experiment	Hypothesis	Method	Success measure
X1	WhatsApp group launch	A useful PDF artefact will outperform an app advertisement	Post the service-record concept (not a link) in 5 car groups, then the link	≥30 signups per group
X2	Garage-first onboarding	A garage will onboard its own customers if the console saves it work	Onboard 3 garages, watch whether they tell customers	≥5 customer signups per garage within 2 weeks
X3	Price index as content	Sharing a "what this should cost" snapshot drives installs	Share a CVT fluid price comparison in 3 groups	≥20 signups per share
X4	The resale test	Owners value the PDF most when selling	Interview 5 people who recently sold a car	≥4 say a service record would have raised their price
X5	Notification-to-action	Users act on reminders rather than dismissing them	Instrument reminder → log within 7 days	≥30% action rate
X6	Odometer accuracy	Users know their current mileage	Ask 15 owners; compare to their dashboard	≥12 accurate within 500km
X7	Record value test	Users with a substantial history return more often than users with almost none	Compare 90-day return rate for users with ≥5 records vs ≤2 records	Return rate at least 2× higher for the history-heavy cohort. If not, the moat is imaginary and R-04a is existential
X8	Competitor watch	ZiMINDEX or ZWPRICE will add a vehicle-adjacent category	Quarterly check of `zimindex.co.zw/products` and `zwprice.com/stores` for automotive terms	Any automotive category appearing = activate R-04a/R-04b response

6. Known unknowns we are choosing to ignore for now¶

Explicitly recorded so they don't get rediscovered as surprises.

Unknown	Why deferred	When it matters
Whether garages will report accurate job outcomes	Unverifiable in v1; trust-based	When lead disputes exceed 1/month
Optimal service intervals for Zimbabwe conditions	Requires aggregated data we don't have	After 5,000 records — then compare recommended vs actual service dates
Whether users will pay for anything at all	Deliberately not testing; the model doesn't require it	If the supply side fails, this becomes the fallback hypothesis
Whether the price index can be automated (receipts, OCR)	Manual garage submission is sufficient at v1 scale	Above 500 price entries
Whether insurance partnership is viable	Requires volume and a relationship	Year 2
Real demand outside Harare	Physical garage network is a Harare constraint	After 25 Harare garages

7. Assumptions to validate with data, not opinion¶

#	Assumption	Current basis	How it gets validated	By when
V1	Owners will log a service in <30s if designed well	Design principle, untested	Timed usability test, 10 users	Gate 1
V2	3–5% of active users request a garage per month	`[ESTIMATE]` — no evidence	Instrument and measure	Gate 2
V3	30% lead → booking conversion	`[ESTIMATE]` — no evidence	Garage-reported statuses	Gate 2
V4	Garages will pay US$40/month	`[ESTIMATE]` — 2 of 5 interviews at Gate 0. Supported by analogy: ZWPRICE monetises suppliers, not buyers	Actual conversions	Gate 3
V5	25–35% 30-day retention is achievable for a maintenance app	`[ESTIMATE]` from category norms	Measured retention	Gate 1
V6	The interval table can be built for 40 models in <2 weeks	`[ESTIMATE]`	Actual build time	Gate 1
V7	Garage onboarding takes ~1 hour	`[ESTIMATE]`	Time the first 10	Gate 2
V8	The record is valued more than the transaction. Users return for their history, not just for a quote.	Unvalidated and now load-bearing. ZWPRICE can route quotes; only a record gives a reason to return. The entire defence against R-04a rests on this.	Ask Gate 0 interviewees what they'd miss if the app disappeared; measure 90-day return rate among users with 5+ records vs 1–2 records	Gate 1

V2, V3, V4 and V8 are the load-bearing assumptions of the entire business model. If any is off by 2×, the revenue projections in 11-business-model.md are wrong by 2×. They must be measured before Gate 3, not assumed.

V8 is the one that arrived with the competitor research. It is the assumption that decides whether ZWPRICE is a threat or an irrelevance, and it can be tested cheaply: compare retention between users with a substantial service history and users with almost none. If history doesn't predict return visits, the moat is imaginary.

8. Questions we've answered (for the record)¶

Decisions already made, so they don't get relitigated.

Question	Decision	Rationale
Should consumers pay?	No	Market evidence; global free alternatives
iOS in v1?	No	Low share, doubles cost
Build vs buy the backend?	Buy (Supabase)	One team, no differentiation in auth/DB ops
Microservices?	No	Operationally unjustifiable at this scale
ML for service prediction?	No	Deterministic intervals are more accurate and explainable
Real-time tracking?	No	Different product, different buyer
Freemium for consumers?	No	No consumer revenue at all
Which idea from the set?	Car Care	`01-ranking.md` — 7.80 vs 6.05 next best
Build the grocery price finder?	No, absorb the mechanic into car parts	Six live competitors including ZiMINDEX; category is occupied
Build the ZESA app?	No	Load-shedding absent 304 days; ZETDC ships an official app
Partner with ZWPRICE or compete?	Partner, at 2,000 users. Compete before that by out-executing on the record	ZWPRICE routes quotes; Mota holds history. Complementary, but do not hand them the lead flow early
Adopt ZWPRICE's trust model?	Yes, immediately	Three-tier listing labels, price-completeness flagging, "not the seller" disclaimer, OTP before lead
Should the price index show crowdsourced (unverified) quotes?	Yes	ZiMINDEX demonstrated that contribution flows work; label it distinctly and never rank it above verified data