1. Define the Market, Customer and Output
Write the country, customer problem, acceptable price band, fulfillment model, category constraints and decision owner before opening a research dashboard. The required output is a record marked reject, monitor or test, not a list of attractive products.
Set a contribution-margin floor and operational constraints from the business's actual cost model. Do not borrow a generic threshold from another seller or category.
2. Apply Hard Rejection Rules First
A score cannot rescue a candidate that fails a non-negotiable condition. Record the rejection reason so the same product is not repeatedly reintroduced by another ranking spike.
- Reject prohibited, restricted or unresolved regulatory and claim risk.
- Reject projected contribution margin below the business's documented floor.
- Reject unacceptable fulfillment, damage, return or delivery risk.
- Reject supplier inconsistency, missing required documents or unreliable replenishment.
- Reject any condition the current account, market or operating model cannot legally or reliably support.
3. Check Demand and Competitive Structure
Review category direction across more than one time window, then inspect products and shops behind the movement. Treat third-party GMV or sales fields as directional estimates unless explicitly identified as first-party data.
Separate broad demand from one campaign, one seller or one short spike. Record category concentration, dominant offers, price structure and whether a differentiated entry is credible.
4. Validate Economics, Supply and Fulfillment
Build a landed contribution model using product cost, freight, platform charges, affiliate commissions, advertising assumptions, returns and packaging. Replace assumptions with supplier quotes and account data as they become available.
Confirm lead time, batch consistency, packaging, required documents, replenishment and return exposure. Strong demand evidence does not compensate for a product the business cannot deliver reliably.
5. Evaluate Creators, Content and Differentiation
Check whether participation is distributed or concentrated among a few creators, and whether the product can be demonstrated honestly by creators the team can realistically recruit. Study hooks, demonstrations, objections and formats without copying execution.
Record caution signals rather than turning them into automatic rejection: creator concentration, seasonality, limited differentiation and heavy advertising dependence all reduce confidence. Positive signals include sustained multi-source demand, workable economics, repeatable content, stable supply and a clear customer-facing difference.
6. Produce a Filled Decision Record
The example below is illustrative, not a claim about current market performance. It shows the minimum output expected before a candidate moves into testing.
| Field | Representative entry |
|---|---|
| Candidate | Reusable pet-hair remover |
| Market and customer | One specified TikTok Shop market; pet owners dealing with fabric and furniture hair |
| Hard rejection check | No unresolved prohibition identified; supplier documents and claim wording still require owner sign-off |
| Demand evidence | Several products, shops and recent content sources observed; third-party values retained as estimates |
| Caution signals | Crowded offer and limited visible differentiation |
| Positive signals | Simple demonstration and a recurring customer problem |
| Economics and supply | Pending supplier quote, landed-cost model and packaging sample |
| Decision | Monitor, not ready to test |
| Next owner action | Sourcing owner completes cost, document and sample checks before a test brief is approved |
A filled record exposes missing evidence instead of converting uncertainty into a high score.
7. Run a Small Test with Stop Conditions
Move only candidates that pass every hard rule and have manageable caution signals into a reversible test. Define the test owner, inventory or sample limit, content hypotheses, measurement window and stop conditions before spend begins.
After the test, replace estimates with actual contribution margin, conversion, returns, creator response and operating effort. Preserve the original assumptions so the team can learn which research signals predicted useful outcomes.
8. Tool Handoff and Final Checklist
Use Kalodata, FastMoss or Shoplus only for the steps their current accounts can support: category, product, shop, creator and content discovery. Hand the shortlist to platform-native records for policy and account evidence, suppliers for cost and reliability, and the team's test log for actual outcomes.
- Market, customer, owner and output are written.
- Every hard rejection rule has an evidence owner.
- Demand is checked across sources and time windows.
- Margin, supply, fulfillment and returns use business-specific inputs.
- Creator concentration, content fit and differentiation are recorded.
- The decision is reject, monitor or small test with a next action.
- Test limits and stop conditions are approved before commitment.
Choose software only after the workflow and required output are defined.
Run the same known-entity task in both products.
Frequently Asked Questions
What should I check first in TikTok Shop product research?+
Define the market and apply hard rejection rules for policy, margin, fulfillment and supplier risk before scoring demand or content signals.
Is estimated GMV enough to choose a product?+
No. Treat third-party GMV as a directional signal. It does not confirm contribution margin, supply reliability, returns, competition or sustainable demand.
What is the difference between a caution signal and a rejection rule?+
A rejection rule stops the candidate regardless of score. A caution signal, such as concentration or seasonality, reduces confidence and changes the test design without automatically ending research.
When is a product ready for a TikTok Shop test?+
Only after it passes every hard rule, has a workable business-specific margin model, manageable caution signals, reliable supply and a reversible test with written stop conditions.
Continue your research
Third-party ecommerce tools may show modeled or estimated data rather than official sales records. Use those signals to form a shortlist, then validate the decision with current platform, supplier, margin and controlled-test evidence.