Choosing a stack is choosing failure modes
“One codebase halves cost” rarely survives year two. Choose React Native vs native by interaction complexity, OS APIs, performance budgets, store risk, and team skills. We build a tech-risk list in discovery—not keyword matching on job posts. Below is a scoreable decision table. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Product shape beats framework taste
Content, forms, moderate motion B2B tools often fit RN; camera/Bluetooth/background/heavy animation lean native or hybrid. List tech touchpoints on critical paths first. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Performance budgets and list complexity
Long lists, complex gestures, live camera need early prototypes. RN can excel with skilled optimization; without specialists, risk hits the schedule. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
OS integration and China OEM variance
Push, background, payments, sharing vary across Android OEMs. Budget native modules and OEM adaptation. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Team skills and hiring reality
Strong React teams ramp on RN faster; deep iOS/Android experts favor native. Mixed teams need module ownership so everyone-touches-nothing does not happen. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Release cadence and OTA bounds
RN OTA has policy and risk bounds—it does not replace store review. Compliance features still ship in store builds. Document the release train. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Hybrid is often the answer
RN shell + critical native modules, or native shell + embedded RN, beats ideological purity. Boundary via interfaces and version pins. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Two-week selection spike
Spike the riskiest path: lists, payments, push, offline. Decide on evidence—not framework wars. Fold this into KSX discover→design→build→launch checks on staging so verbal alignment does not drop the requirement. Fold this into KSX Studio discover→design→build→launch: named owner, Definition of Done, staging proof, and rollback so verbal alignment does not drift.
Checklist
- 1List OS API touchpoints on critical paths
- 2Define mobile performance budgets
- 3Assess skills and hiring gaps
- 4Clarify OTA vs store release strategy
- 5Complete a two-week riskiest-path spike
Key takeaways
- Choose on failure modes and team—not slogans.
- China Android OEMs make integration cost real.
- Hybrid with clear boundaries is usually pragmatic.
FAQ
- Can RN be high performance?
- Yes with skill and budget. Do not assume automatic speed without specialists.
- What about Flutter?
- Score with the same dimensions; team skill and OS integration beat hype.
- Reuse web code?
- Share domain logic and tokens; do not force one UI everywhere.