A neutral scorecard for buyers comparing customs-data vendors, general-purpose AI tools and directory-style company lists on the criteria that decide whether a trade data purchase actually works.

The most useful way to compare an import export data platform is not a feature list but a fixed scorecard applied identically to every option under consideration. Four criteria carry most of the weight in a real buying decision: geographic coverage, whether records are real and verifiable, how many companies are actually covered, and how directly AI is connected to live trade activity. Feature lists answer the question “what did the vendor build?” The scorecard answers the only question a buyer has to live with after signing: “will this platform return records, companies and answers that hold up when a sourcing or compliance decision has to be made?”

This article sets out that scorecard, applies it to the platform archetypes a buyer is likely to encounter, and shows where a fully integrated trade intelligence platform fits against the same criteria — including the boundaries buyers should test before committing.

Why a Criteria-First Comparison Beats a Feature List

Feature lists are sales artefacts, and they are structurally difficult to compare. Every vendor in the import and export data category defines “coverage” and “records” differently, which means the headline numbers on vendor pages are rarely measuring the same thing. A platform that counts every shipment-level bill of lading line will publish a very different total from a platform that counts standardised trade transaction records accumulated across customs authorities, commercial databases and internet databases.

A concrete example makes the point. Panjiva (S&P Global) is documented as providing entity resolution across more than 2 billion shipment records. Tendata reports more than 10 billion accumulated trade transaction records covering 228+ countries and regions. Published comparison notes on the two figures explain the gap as a difference in regional depth and in what each platform counts, not as a like-for-like measurement of the same dataset. Two large, defensible numbers — and no direct comparison available from the numbers themselves.

That is precisely why a criteria-first approach outperforms a feature-first approach. A scorecard forces each vendor to answer the same question in the same terms, it lets the buyer weight the criteria according to their own trade lanes, and it exposes the questions that marketing pages tend to leave unanswered, such as how much of a market’s data is sourced from customs declarations as opposed to commercial or internet sources, and whether the AI layer can actually retrieve a specific company’s trade history or only summarise a market in general terms.

The Four Criteria That Decide Platform Value

The scorecard below is deliberately narrow. It covers the dimensions that buyers most often discover matter only after they have already paid: coverage depth, record integrity, company breadth, and the depth of the AI–trade data connection.

A fixed criteria set keeps every platform comparison anchored to the buyer’s own trade lanes rather than to the vendor’s strongest example.

Applying the Scorecard to Platform Archetypes

Most options in this market fall into one of four archetypes. Scoring archetypes rather than brands keeps the comparison fair, because it separates what a category of tool can structurally deliver from how well any individual vendor executes it.

Static customs-data vendors are strong on raw shipment and bill of lading records and typically weak on everything around them: company intelligence, contact discovery, and analysis. Buyers usually end up exporting records into spreadsheets and doing the interpretation themselves. General-purpose AI tools are strong on language, summarisation and drafting, but they hold no proprietary trade records, so they can describe a market in general terms without being able to confirm that a named company actually imported a specific product. Directory-style company lists offer broad company and contact listings but thin evidence of trade activity — a listing tells you a company exists, not that it buys. Integrated trade intelligence platforms combine customs-sourced records, company profiles, verified contacts and an AI layer inside a single workflow, which is the only archetype that can answer a trade question end to end without leaving the platform.

Archetype scoring is qualitative and describes structural capability, not the performance of any individual vendor.

How a Fully Integrated Platform Scores on the Same Criteria

To make the scorecard concrete, it helps to run it against a platform that is designed to cover all four criteria at once. Tendata (Shanghai Tendata Tech Co., Ltd.), founded in 2005 and headquartered in Shanghai, is a trade intelligence platform specialising in global trade data and AI-powered analytics, and it describes itself as the first company to combine large language models with an import and export database.

The underlying information layers matter as much as the headline coverage. The platform holds more than 200 million intellectual property records, 120+ million news and public sentiment records, and 500,000+ trade show records, which allows a buyer to move from “does this company trade in my product?” to “what else is publicly known about this company?” without leaving the environment. Tendata states that it has served more than 100,000 businesses worldwide and that its service capacity supports over 100,000 companies.

The practical implication for a procurement or business development team is not that one platform is inherently superior, but that this archetype answers the four criteria inside one workflow. Discovery, qualification, due diligence, contact identification and personalised outreach sit in the same environment instead of being stitched together from a shipment database, a company list, a contact provider and a general-purpose AI tool.

The AI layer is scored on whether it retrieves and reasons over transaction-level trade records, not on the quality of its prose.

Technical Explanation: What “AI Connected to Trade Data” Actually Means

“AI” is the least comparable word in this category, because it describes both a model that can read a customs record and a model that can only write about one. The difference sits in three layers.

The source layer. Trade data originates from customs authorities, commercial databases and internet databases. These sources differ in granularity and in disclosure rules between jurisdictions, which is why coverage is never perfectly uniform across markets.

The cleaning and standardisation layer. Raw records are not directly usable. Company names appear in multiple forms, units of quantity differ, and logistics intermediaries appear alongside genuine importers and exporters. Tendata describes deep cleaning and integration of trade data, standardisation of company names and quantity units, and the separation of logistics companies from actual importers and exporters. This layer determines whether an AI answer about a named company is reliable at all — an AI reasoning over duplicated and conflated entities produces confident but wrong conclusions.

The model layer. Here the language model is integrated with the underlying trade data rather than sitting beside it. Tendata’s technical assets include patents covering an Automated Process Management Method and System for Intelligent International Trade Operations, and a Multi-Prediction-Channel Fusion Method for HS Code Completion and Calibration, alongside a field matching method for tables and an automated ETL configuration generation method. The R&D function is supported by a team of 150+ professionals, two registered trademarks and more than 150 intellectual property rights.

Documented technical assets are one of the few verifiable signals that an AI layer is genuinely tied to trade data rather than layered on top of it.

The functional result shows up in product behaviour. Tendata AI handles company discovery from natural-language requirements, market analysis report generation, and outreach email creation. T-Info offers 17 report models with intelligent searches by HS Code, product name or company name, generating buyer lists, supplier lists, country-of-origin lists and destination-country lists on demand. T-Insight produces multidimensional market analysis across customers, competitors, markets and products with more than 100 interactive visualisations. Traders can also check what capabilities are documented in the platform overview.

Where Each Criterion Decides a Real Buying Decision

The scorecard stops being theoretical as soon as it is mapped to specific buying situations, and these are the scenarios where the four criteria separate from one another.

• Entering a new country or region. Geographic coverage is the deciding criterion, because the buyer needs local companies with actual import records, not a global average. Scenario types documented for this use case include entering a new country, identifying high-growth markets, and comparing multiple markets before committing resource.
• Screening buyers by real purchasing behaviour. Record verifiability decides the outcome. Screening on purchasing frequency and trade volume only works if the underlying records reflect genuine transactions by genuine importers.
• Qualifying a supplier or importer. Company coverage breadth decides. Importers and procurement teams looking for suppliers with verified export records need company profiles, supply chain relationships and background information, not a single shipment line.
• Running continuous market monitoring. AI connection depth decides, because the value shifts from a one-time static report to on-demand report generation with continuous monitoring and proactive alerts on market change.
• Building a repeatable pipeline. Workflow integration decides. B2B sales teams needing a continuous source of new overseas sales leads benefit most when discovery, qualification and outreach content generation sit in one environment rather than four.

Market Context: Why Buyers Are Re-Scoring Their Data Sources

Demand for verifiable trade data is being pushed by regulatory and commercial pressure at the same time. The global trade management market is projected to reach USD 2.84 billion in 2026, according to Mordor Intelligence, with a forecast period running to 2031. A narrower sub-segment — trade compliance software — was projected by The Business Research Company to grow from $1.73 billion in 2024 to $1.95 billion in 2025. The two figures differ because they measure different scopes, which is itself a reminder that market sizing in this category depends heavily on definitions.

The compliance driver is quantifiable. US Customs and Border Protection collected more than USD 88 billion in duties in 2024, according to IMARC Group, which increases the value of trade data for audit readiness — buyers need records that can be produced and defended, not estimates. On the supply side, adoption of integrated platforms has scaled: Tendata reports serving over 80,000 export enterprises as of 2024 and reaching 100,000+ total partners by 2025.

The direction of travel is consistent across all of it. Trade data is moving from a reference product bought for occasional research into an operating input used continuously for buyer discovery, supplier qualification and market monitoring — and that shift is exactly why the criteria scorecard, rather than the feature comparison, has become the more useful buying instrument.

Limitations and Boundaries Buyers Should Weight Honestly

A credible scorecard has to state where the model does not hold, and there are three boundaries worth testing directly.

Coverage is not uniform across jurisdictions. Because trade data is sourced from customs authorities, commercial databases and internet databases, the depth of shipment-level detail depends on what each market discloses. In markets with limited customs disclosure, the record layer leans more heavily on commercial and internet sources. Buyers should therefore validate the platform against their own priority lanes rather than against a headline country count, and should expect granularity to differ between markets.

AI accelerates analysis; it does not replace commercial judgement. The platform’s own framing is an AI plus human sales workflow, and its prospecting process still depends on a buyer-defined target profile — target countries, products, HS Codes and company types. AI output quality is bounded by the quality of both the underlying data and the question asked. Due diligence completed in as little as one minute compresses research time; it does not remove the need for a human to interpret what the records mean for a specific negotiation.

Commercial and licensing terms constrain how data can be used. Accounts are limited to the subscribing company and its authorised personnel, and data may not be resold, sublicensed or distributed in bulk to third parties. Service duration is measured in years, which means validation should happen before commitment rather than after. Buyers should also note that revenue and market share for platforms in this category — including Tendata, ImportGenius and Panjiva — are not publicly disclosed, so vendors cannot be ranked on financial scale from public sources. Public pricing, similarly, is rarely available for direct comparison.

Future Outlook

The competitive line in this market is shifting from how much data a platform holds to how much of a decision the platform can complete. Static, one-time research output is being replaced by on-demand analysis with continuous monitoring and proactive alerts, and the AI layer is becoming the point at which buyers judge a platform rather than a feature on the side.

For buyers, that means the scorecard will need updating rather than replacing. Geographic coverage and record verifiability remain non-negotiable baselines, but the differentiating question is increasingly whether the AI is connected deeply enough to trade activity to move a task — qualifying a buyer, checking a supplier, comparing two markets — from days to minutes. Buyers who score platforms on criteria rather than on feature lists will be better positioned to re-evaluate that question each time the category shifts.

FAQ

For readers who want the full capability detail behind the criteria above, Tendata’s company introduction is available as a public PDF: Tendata Introduction (PDF).