Skip to content
Interface language: English
Oriens / Signal

Conflicting data on Asian markets: how to find the source of the difference

Compare definitions, scope, units and provenance before using conflicting market figures.

Article language: English

A woman astronomer adjusts the wooden gears of a mechanism with two bright discs.
In this article

Two reports give different market values, and averaging them is tempting. That produces a third number with no definition of its own. When sources on Korea, Japan or China differ, compare the product, area, period, unit and stage of trade before assessing their authors. Two correct figures can describe different phenomena. The aim is to recognise that situation and retain a useful conclusion.

The goal is not to pick the value that looks best in a sales plan. Establish what the sources allow you to say, what remains unknown and whether the difference changes the company's decision. The procedure below helps resolve that question without pretended precision. Every figure and company in the exercises is hypothetical, used to illustrate reasoning rather than describe a current market size.

State the disputed claim in one sentence

Before opening another tab, name the conflict. For example, two documents report different numbers of businesses using a particular technology. Do not just write “the data disagree”. Add the decision the figure must support: assessing a potential customer group, choosing a channel or comparing countries. If it makes no difference to the choice, a note may be more useful than another week of investigation.

Keep both passages with the document title, URL, publication date and date accessed. For a file, record the version and page as well. A search-result snippet is not enough: it may omit a qualification in the document. The guide to reading a Korean market report (in Polish) helps establish similar checks before a number reaches a presentation. Here the task is to compare two specific claims, not assess an entire report.

Compare definitions before converting numbers

Ask what each source actually counted. It might count companies, establishments, installations, devices sold, active accounts or transactions. Then check product scope. A report on an entire machine family need not answer a question about one variant. Similar titles do not guarantee similar coverage. Sometimes reading the first footnote resolves the difference after a slide with a large number has already travelled widely.

In a hypothetical exercise, dataset A contains facilities with a particular production line; dataset B contains companies reporting that they serve the industry. One company may have several facilities, while a service provider may own no line. Simply subtracting the records cannot establish agreement. You need a definition of each unit and the relationship between entities. The same distinction helps when cleaning an Asian company database (in Polish).

Check the market boundary and where measurement occurs

Determine whether the data cover production within a country, sales in its territory, imports or companies headquartered there. These are different perspectives. A Japanese-headquartered group's global revenue does not automatically measure demand from Japanese customers for its product. Production in China may likewise serve many buyers outside the local market.

Check the stage at which value was measured. Manufacturer sales, distributor turnover and end-user purchases can include different components. Do not add them merely because they appear in separate reports. In competitor analysis, distinguish manufacturer, brand and sales channel, as in the procedure for comparing Korean offers (in Polish). Each level may be useful but answers a different business question.

The Chinese photovoltaic value chain (in Polish) illustrates this distinction: cell production, module production and additions to installed capacity describe different stages. Before comparing figures, check whether a source describes capacity, actual output or use.

Note period, unit and version

A report's publication year is not necessarily its observation year. Check whether the number covers a full period, part of a year, a forecast or a point in time. Separate current and historical values, and do not call a forecast realised sales. If you convert a currency or unit, record the method and source. Otherwise the next person cannot reproduce the difference even with the same documents.

UN Comtrade's data availability guidance explains that datasets can be updated and revised and that it maintains one version of a dataset rather than a permanent archive of successive versions. Keep the downloaded file and date. Your old copy and today's query can differ without an arithmetic error. Check the version before changing formulas.

Do not expect perfect symmetry in trade

UN Comtrade explains bilateral asymmetries: one country's export data and its partner's import data can differ. Reasons include partner attribution, valuation and differences between statistical systems. Comparing the two sides is not a simple test of which administration reported the correct figure.

For your decision, state which perspective you use and why. If imports into one country are the question, use data that answer it; treat exporters' reported values as material for assessing the difference. Do not manually adjust an official series to match the other side. The guide to HS trade data (in Polish) covers codes, periods and units in more detail. The practical point is that disagreement alone does not settle the size of demand.

Establish whether sources are independent

Three reports may repeat the same original figure. Follow their references: did each conduct its own research or cite an earlier document? If all lead to one source, they are not three independent confirmations. Search-result popularity shows circulation, not the number of measurements. A polished repetition is still a repetition.

Aeneia conceptual photography

Draw a short provenance chain: claim in a presentation, intermediate report, primary publication, table or method. If it stops at “according to experts”, record that limit. Do not invent an author or method. Monitoring East Asian markets (in Polish) should track changes in primary sources so another retelling of the same figure does not become a new market signal.

Separate explained and unresolved differences

After checking, classify the disagreement. First, the sources may measure different things; both can remain if their scopes are described. Second, an author may explicitly mark a correction to the same dataset; use the corrected version and explain the change. A later publication date alone does not prove replacement. Third, definitions may be similar but methods or data insufficient to explain the gap. The last case needs an explicit unknown, not an arbitrary winner.

For an unresolved difference, ask whether both values lead to a similar decision. If the cost of reaching the audience remains the main obstacle under both, further work may focus on the channel. If the choice depends on one number, obtain more evidence or defer the decision. Comparing a first Asian market (in Polish) should expose such conditions rather than hide uncertainty in a final score.

Exercise: two company lists

Suppose list A has twelve entries and list B eighteen. These are invented numbers for an exercise. Rather than call the larger list better, compare units. One might contain manufacturers only; the other might include distributors and entities in the same groups. Start with names, identifiers and relationships, then examine each inclusion rule.

Mark shared records, unique companies and ambiguous cases. For each difference, record only a reason supported by the material: another scope, date or application, or an unresolved issue. After that you can prepare a combined list for conversations. It is still not a directory of every market participant: it is an organised set of candidates found by a stated method in stated sources.

Consider the effect on the plan. If most additional entries are sellers, that may call for channel analysis rather than prove there are more users. If profiles are outdated, current activity needs checking. One discrepancy can lead to different tasks. Name them before handing a database to sales as a finished result.

Prepare a short resolution record

Include the question, two values or observations, definitions, sources, dates and checks. State what was explained, what remains unknown and the effect on the decision. Add an owner for further work and the event that would justify an update. Not every dispute needs a long report; every important conclusion needs a reproducible basis.

A sample entry: source A covers only manufacturers; B also includes trading firms. For planned conversations with users we use the verified part of A, while analysing trading firms separately as a possible channel. This is our research decision, so label it as such rather than attributing it to the source's author. Keep the reason in a project decision log (in Polish) for future evidence.

Does the difference change the action plan?

Consider two candidate segments: users replacing an existing solution and companies designing a new process. The first question may require the number of installations needing service; the second, the number of projects starting. If the reports use different units, you need not choose one winner for all applications. In Asian B2B segmentation (in Polish), assign each source the question it can actually answer. Useful material stays in the analysis with its scope visible.

Test both scenarios. Suppose in turn that the narrower and wider description is closer to the needed scope. In both cases, can you support an initial trial, reach the audience and identify product assessment conditions? If not, uncertainty about market size may not be the most urgent task. Keep it in the report without letting it replace a feasibility check.

The position changes if a decision depends on crossing a specific threshold. Explain where that threshold came from and whether another method can check it more precisely. You may need a list of entities meeting particular conditions, not another general report. Choose an additional source to fill a gap in the reasoning. More open tabs do not guarantee less uncertainty; sometimes they hide the one you needed first.

Ask a source's author an answerable question

If the method is unclear, identify the document, table, unit and precise uncertainty. Ask whether the scope includes a named subgroup or whether a value replaces an earlier publication. A general request to certify an entire report is harder to answer. Do not send confidential project assumptions when a table number and product definition suffice.

Keep the author's answer as a separate source with date and scope. Silence confirms neither side. If clarification does not arrive before the decision, record the limitation and working scenario. This allows a later correction without reconstructing the analyst's intentions.

Accept the result by reproducing the reasoning

Give the record to someone who did not conduct the analysis. They should understand the source of the difference and the reason for the chosen next step. “Reliable source” alone omits the scope comparison. Two numbers and an unexplained average omit the reason for the transformation. Test meaning, not just whether a spreadsheet works. If an AI model reads a chart, also check its axes, scale, legend and missing points (in Polish). Comparing sources cannot repair an error introduced when transferring data from a graphic.

In an Aeneia Oriens research project, comparison of conflicting material can be discussed as a defined scope: definitions, source tracing and implications for a decision. Bring the two disputed documents and the question the analysis must answer. The result may identify two valid scopes and explain which one is needed for the next step.

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Contact Aeneia

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear