TikTok Product Intelligence

Spot product momentum early — with evidence, not hype.

TikTok Product Intelligence turns a product idea into an auditable evidence pack across organic TikTok posts, buyer-intent comments, TikTok Shop commerce, persistent history, breakout detection, optional Creative Center ads, optional supplier sourcing, public supplier due-diligence signals, and optional landed-cost scenarios.

It is a product-intelligence engine, not a black-box “winning product” score and not an automatic buying decision system.

What should I enter?

Start with the product idea you already want to investigate. You do not need to know an exact TikTok product ID or URL.

Good inputs include:

• a specific product or model: TOZO NC20 Pro;
• a product category or niche: electric bike, portable blender, walking pad;
• a supported hashtag, creator, TikTok video/profile URL, or TikTok Shop product/seller URL when you already have one.

TikTok Product Intelligence evaluates the idea or scope you give it. It does not autonomously scan all of TikTok and choose today's “best products” for you.

What you get

A run can produce structured records for:

• TikTok organic posts and creators;
• comments classified into buyer intent, price, availability, shipping, product questions, and sentiment;
• TikTok Shop product, seller, price, sold-count, rating, and review evidence;
• conservative product clusters with explicit REVIEW_REQUIRED states;
• transparent commercial signals with visible formulas;
• persistent same-scope snapshots;
• change, velocity, and breakout evidence after enough history exists;
• durable monitoring state and optional signed webhook alerts;
• public TikTok Creative Center ad evidence;
• optional supplier offers, MOQ, quantity tiers, and observed wholesale costs;
• public supplier due-diligence signals for human review;
• optional user-declared landed-cost scenarios;
• compact RUN_SUMMARY.json output.

Start small

Use the default Store example for a useful first run without the optional higher-cost sources:

{
  "query": "electric bike",
  "country": "US",
  "maxPosts": 5,
  "maxCommentsPerPost": 3,
  "includeShop": true,
  "shopMarket": "US",
  "maxShopProducts": 5,
  "includeAds": false,
  "includeSuppliers": false,
  "includeLandedCost": false,
  "monitor": false
}

Optional ads, supplier sourcing, and monitoring can add upstream cost. Review the first dataset before enabling them. The small default limits also keep Apify Store health checks fast and inexpensive.

Supplier sourcing

Supplier sourcing is optional and disabled by default.

{
  "query": "electric bike",
  "includeSuppliers": true,
  "maxSupplierOffers": 20,
  "supplierTargetQuantity": 10
}

The Actor normalizes public supplier rows such as price ranges, quantity-price tiers, MOQ, supplier profile/trust fields, certifications, and provenance. Automatic links require product-specific evidence. Brand-only, generic, or ambiguous supplier rows remain unresolved.

When retail and supplier evidence already use the same currency, the Actor can calculate:

retail price - observed supplier unit cost

That is a gross spread proxy, not landed margin or profit.

Supplier due diligence

For product-linked supplier offers, the Actor can expose observed marketplace/profile signals such as:

• Trade Assurance;
• supplier badges or assessment flags when present;
• years on platform;
• review score/count;
• response rate;
• certifications.

These fields are evidence for human review. They do not mean the supplier has been independently verified by TikTok Product Intelligence.

The Actor does not independently verify supplier identity, manufacturing capability, product compliance, sample quality, or transaction safety, and it does not rank suppliers.

Landed-cost scenario

Landed-cost analysis is optional, disabled by default, and requires supplier sourcing.

Example:

{
  "query": "electric bike",
  "includeSuppliers": true,
  "supplierTargetQuantity": 10,
  "includeLandedCost": true,
  "landedCostCurrency": "USD",
  "sourcingDestinationCountry": "SI",
  "freightPerUnit": 35,
  "customsDutyPerUnit": 18,
  "nonRecoverableImportTaxPerUnit": 0,
  "otherImportCostPerUnit": 7
}

The formula is deliberately simple and visible:

landed cost per unit
= observed supplier unit cost
+ freight per unit
+ customs duty per unit
+ non-recoverable import tax per unit
+ other import cost per unit

The import-cost amounts are user-declared inputs. The Actor does not determine customs classification, duty, VAT/tax treatment, freight quotes, or regulatory obligations. There are no hidden numeric defaults; enter 0 only when you have evidence that an amount is zero.

Supplier cost, retail evidence, and declared amounts must use the same currency. The Actor performs no FX conversion.

When the scenario is computable, the Actor can show:

retail price - landed-cost scenario

as a landedGrossSpreadProxy.

This remains a scenario, not gross profit, net profit, ROAS, or a forecast. Selling fees, payment fees, fulfillment, returns, ads, compliance, warranty, overhead, and other operating costs are outside the calculation, so profitability can remain UNKNOWN.

Evidence-first rules

The Actor intentionally preserves uncertainty:

• weak Shop matches remain REVIEW_REQUIRED;
• generic organic posts/comments are not forced onto products;
• brand-only ads or supplier rows are insufficient for product linkage;
• variant-family cumulative counters are not summed without evidence;
• one snapshot is not a trend;
• cumulative counter decreases fail closed as reset/coverage changes;
• Creative Center does not represent every TikTok ad;
• supplier listings do not prove identical product identity or supplier reliability;
• user-declared landed-cost amounts remain clearly labeled as user inputs.

Monitoring and breakout detection

Repeated runs with the same canonical scope build persistent history. Once enough comparable observations exist, TikTok Product Intelligence can produce change and breakout evidence instead of treating a single snapshot as a trend.

Monitoring is disabled by default. When enabled, the Actor retains durable breakout-alert state across repeated same-scope runs. For recurring execution, create an Apify Schedule outside the Actor using the same input. The public Actor intentionally does not create or modify account schedules, so it can run with limited permissions. Optional webhook delivery uses environment secrets rather than public input.

Output

Use recordType to filter the default dataset. Important records include:

• summary
• product_cluster
• commercial_signal
• snapshot
• change_signal
• monitor_event
• ads records and reports
• supplier_offer
• supplier_cost_signal
• supplier_verification_evidence
• landed_cost_scenario
• landed_cost_report
• performance_report

The default key-value store also exposes RUN_SUMMARY.json for a compact run-level overview.

What this Actor does not claim

TikTok Product Intelligence does not promise or infer:

• guaranteed winning products;
• guaranteed sales;
• profit or ROAS;
• independently verified suppliers;
• exact total market demand;
• complete visibility into all TikTok advertising;
• customs, tax, or compliance advice.

Cost controls

The core evidence pipeline has fail-closed request-volume limits. Supplier sourcing has a separate collection budget and is opt-in. Landed-cost calculation itself uses already collected supplier evidence and does not add another upstream collector.

Requested limits are not billing claims. Real cloud cost depends on configured upstream Actors and result counts.