IPIS Conflict Minerals: Intelligence Source Guide

IPIS maps artisanal mining sites in eastern DRC from field visits, recording minerals worked, workforce and which armed actors are present and taxing. The only systematic open dataset linking specific mine sites to specific armed groups – and a record of visits, not a live picture.

At a glance

Source	IPIS Conflict Minerals
Category	Conflict, Crime & Human Security › Environmental & Wildlife Crime
Homepage	https://ipisresearch.be/
Machine interface	`https://www.ipisresearch.be/mapping/webmapping/drc_mining/`
Format	`JSON`
Access	Open — no account required
Disciplines	Environmental Intelligence, Geospatial Intelligence
Mission domains	Mining & Resource Crime

Mapped mining sites & armed control. — as catalogued in the platform’s own source registry.

The International Peace Information Service is an independent research institute based in Antwerp that has, since around 2009, sent teams to physically visit artisanal and small-scale mining sites in the eastern Democratic Republic of the Congo and, in later phases, in the Central African Republic and other conflict-affected areas. The output is a geospatial dataset and an accompanying webmap. Each record is a mine site with coordinates, the minerals worked there – principally gold, cassiterite for tin, coltan for tantalum, and wolframite for tungsten – an estimate of the number of workers present, the presence and identity of armed actors, whether those actors were levying illegal taxes or controlling access, the presence of state security forces, indications of child or forced labour, the site's status under the Congolese qualification and validation process, the nearest settlement, and the date of the visit. Data collection is done with Congolese partners and, in many phases, alongside state mining service agents, under funding from donor governments and international organisations. The dataset is published openly, downloadable and viewable on an interactive map, and accompanied by analytical reports covering the mineral trade, armed group financing, roadblocks and taxation on transport axes. It is a survey product: what it contains is what enumerators observed on the ground, on a date, at a place they were able to reach.

Due diligence regimes for conflict minerals require a company to know whether the material in its supply chain financed armed conflict. Nothing in a company's own paperwork can answer that, because the paperwork begins at the point of purchase and the question is about what happened upstream of it. This dataset answers the geographic half of the question directly: it tells you which specific sites, at specific coordinates, were under or subject to armed actor control at a specific time, and which armed actors those were. That is the only open, systematic, site-level empirical basis available for the OECD due diligence framework as it applies to the Great Lakes region, and it is why the dataset appears in company reporting, regulatory submissions, UN panel of experts work and academic literature alike. The second job it does is to make the informal economy legible at all. Artisanal mining is by definition outside the formal licensing and payment structures that transparency regimes such as EITI capture; a country can have a fully reconciled extractive revenue report while the majority of its gold leaves through channels that report contains no trace of. This dataset is the counterweight – it observes the sector that formal disclosure cannot see. Third, because visits recur over years, it supports change analysis: which groups have expanded, which sites have shifted from one actor to another, where militarisation has increased despite formal validation. That trajectory information is what turns a static map into an intelligence product.

Who publishes it, and why that matters

IPIS is a small independent research institute with an academic and peace-research orientation, funded by project grants from donor governments, international organisations and foundations. That funding model is the single most important thing to understand about the dataset. Field mapping in eastern DRC is expensive and dangerous, and the extent of any given year's coverage is determined by which project was funded, for which provinces, over which period. Gaps in the data therefore reflect grant cycles as much as they reflect anything on the ground, and a province that appears well mapped for one period and sparse for another has usually just moved in and out of a funded programme. The institute's incentives are academic and policy-oriented rather than commercial, which shows in the quality of the methodological documentation – the survey instrument, the limitations and the collection constraints are described openly, which is rarer than it should be. Working alongside Congolese state mining services gives access and local legitimacy, and simultaneously introduces the possibility that what enumerators are shown at a site is curated. The institute publishes its analytical conclusions as reports, and those reports are the place where the caveats live; the downloadable dataset carries the observations without the interpretive frame, and using it without reading the reports is the standard way people misuse it.

Provenance is the first question to ask of any dataset and the one most often skipped. Who collects it, what their incentive is, whether they publish a methodology, and whether they correct the record when they get something wrong all bear directly on how much weight a finding drawn from it can carry.

What a record actually contains

The fields you will be working with, what each one means, and whether it is something you can pivot on. Read the meanings carefully — more analysis is wrecked by misreading a field than by failing to find one, and a field that looks like an observation is often an inference.

Field	Type	What it means	Pivot value
`site_id`	string	Identifier for the mining site in the IPIS dataset. It is a dataset key, not a national cadastre reference, and it does not resolve against Congolese licensing records without manual work.	Repeat visits to the same site across survey rounds; joining to the webmap record.
`visit_date`	timestamp	When enumerators were physically at the site. This is the single most important field in the dataset: every other value is true as of this date and only as of this date.	Change analysis across rounds; correlation with conflict event timelines.
`latitude / longitude`	array	Site coordinates recorded in the field, typically by handheld GPS. Precision is generally good; what varies is what the point represents, since a large artisanal area can be recorded as a single point.	Concession overlay, protected area overlay, forest change alerts, road and river access, distance to trading centre.
`minerals`	array	Minerals worked at the site – gold, cassiterite, coltan, wolframite, and others. Sites can work more than one, and the mix can change seasonally with water levels and prices.	Supply chain analysis by mineral; downstream smelter and refiner mapping; price and export data.
`workers`	int	Estimated number of people working at the site at the time of the visit. Artisanal workforces fluctuate enormously with season, price and security, so this is a snapshot rather than a capacity figure.	Scale assessment; comparison with declared production; labour and humanitarian caseload estimation.
`armed_actor_present`	enum	Whether an armed actor was present at the site during the visit. Presence is the observable; control is an inference the analytical reports make, and the two should not be conflated.	Armed group profiles, conflict event data, sanctions designations of group leadership.
`armed_actor_name`	string	The identified group or force – a named non-state armed group, or state forces including the national army and police. Naming is based on what enumerators could establish locally and groups fragment and rename frequently.	Group order of battle, UN group of experts reporting, leadership sanctions listings.
`interference_type`	enum	How the armed actor was involved: illegal taxation, control of access, direct extraction, forced labour, or presence without evident interference. This distinction determines the due diligence consequence.	OECD risk categorisation; supply chain red flag assessment.
`state_service_present`	enum	Presence of state mining service agents, police or army. State presence is not automatically benign in this context; the reports document illegal taxation by state forces as a recurring finding.	Governance assessment; comparison with formal validation status.
`validation_status`	enum	Status under the Congolese site qualification process, commonly expressed as green, yellow or red. Validation is periodic, lags reality, and a green site visited later under armed presence is exactly the discrepancy worth investigating.	Official validation records; supply chain scheme eligibility; certification claims by downstream buyers.
`child_labour_indicator`	enum	Observed presence of children at the site. Recorded as an observation under a survey protocol; it is a protection indicator requiring referral, not an investigative lead to be pursued independently.	Referral to child protection actors and national authorities; humanitarian programming.
`nearest_town`	string	Nearest settlement or trading centre, which is the practical link between a remote pit and the trading and transport network that moves the mineral.	Trading house and negociant networks; transport axes; roadblock data; export routes.
`province / territory`	string	Congolese administrative division containing the site. The basis for aggregation to a unit with an identifiable authority and for joining to conflict and humanitarian datasets.	Provincial mining authority, humanitarian cluster data, conflict event aggregation.
`survey_round`	string	Which field campaign produced the record. Rounds differ in geographic scope, instrument and funder, so cross-round comparison requires knowing whether the same questions were asked.	Methodological comparability assessment; identification of coverage expansion versus real change.

Coverage — and what is not in it

Coverage is concentrated in the eastern provinces of the Democratic Republic of the Congo – the Kivus, Maniema, Ituri, Tanganyika and adjacent areas – which is where the conflict minerals problem is located and where the funded programmes have operated. Later work extended to the Central African Republic and to thematic mapping of transport axes and roadblocks. The dataset now contains thousands of visited sites accumulated over more than a decade, which makes it substantial by the standards of field survey work and small by the standards of the sector it describes: the number of artisanal mining sites in eastern DRC is not known with any precision and is certainly larger than the number that have been visited. Coverage within the mapped provinces is a function of accessibility. Sites near roads, near towns and in areas where security permitted a visit are over-represented; sites deep in forest, in areas under active hostilities, or controlled by groups hostile to outsiders are under-represented, and those are frequently the sites where armed interference is most severe. There is a mineral bias too: the tin, tantalum and tungsten sector has attracted more sustained donor and industry attention than gold, while gold is by volume and value the greater problem and is easier to smuggle. Time coverage is uneven and round-based rather than continuous – a site visited in one campaign may not be revisited for years, and the interval between visits is the resolution limit for any change analysis. Update rhythm is publication of new survey rounds when a programme completes, not continuous refresh.

Known blind spots

Absence of evidence here is not evidence of absence. These are the conditions under which IPIS Conflict Minerals will not show you something that is nevertheless real:

Inaccessible sites are absent, and inaccessibility correlates directly with the severity of armed control, so the dataset systematically under-represents the worst situations it exists to document.
Every record is a visit-date snapshot. Armed actor presence in eastern DRC changes over weeks, and a record more than a year old describes a situation that may have reversed entirely.
Gold is under-covered relative to its economic importance, because gold sites are numerous, small, dispersed and less well served by the industry traceability schemes that fund much of this work.
What enumerators are shown at a site can be managed. A visit that is announced, escorted, or conducted alongside state agents may not reveal the actual pattern of taxation or the presence of children.
Coverage follows grant funding, so an apparent decline in sites or in armed presence in a province may reflect the end of a project rather than any change on the ground.
The dataset records presence and observed interference, not the trading chain. What happens to the mineral after it leaves the site – who buys it, how it is consolidated, where it is smuggled across a border – is largely outside these records.
Cross-border smuggling into neighbouring states, which is where a large share of eastern DRC gold enters the legitimate market, is not visible at all in a site-level survey.
Armed group naming is unstable. Groups split, merge, rename and rebrand, and a name recorded in one round may correspond to a different formation in the next.
Formal validation status is periodic and lags reality, so both the presence of a green designation and the absence of one carry limited information about conditions at the time of your query.

Write the blind spot into the product. A statement that something “was not observed in IPIS Conflict Minerals” is defensible; a statement that it “did not happen” is not, and the difference is what survives cross-examination.

Access, licensing and what you may do with it

Access model: Open — no account required

The mapped data is published openly through an interactive webmap and downloadable datasets, with no registration and no fee, and this is deliberate: the institute's purpose is that companies, regulators, researchers and Congolese actors can all use the same evidence base. The practical route is to take the full download rather than querying the map, because the analysis you want is almost always over the whole set rather than a single point. The essential companion is the report series. Each survey round is accompanied by analytical publications that explain the methodology, the geographic scope, the constraints encountered and what the aggregate patterns mean, and those documents contain the limitations that the raw records do not. Download both, and archive the dataset version with a date, because updates replace rather than append and there is no versioned history you can go back to. Where the institute publishes a webmap alongside a dataset, the map is the better tool for orientation and site-level inspection and the dataset is the better tool for anything systematic. For research use, contacting the institute is worthwhile: methodological questions about instrument changes between rounds are not answerable from the published files alone.

Licence

The data is published for open use with attribution, consistent with the institute's mission of making the evidence base available to industry, regulators and researchers alike. Confirm the specific licence terms on the download page before redistributing or building a commercial product, because the exact terms attached to different products and rounds have not always been stated identically. Attribution is expected in all cases and, for academic use, citation of the accompanying report rather than only the dataset is the norm. Two constraints sit outside the licence. Records containing protection indicators – child labour, forced labour – concern identifiable vulnerable populations at identifiable locations and their onward use carries obligations that no data licence addresses. And redistributing site-level armed actor attributions in a form that could reach the actors named creates risk for the enumerators and communities who supplied the observation, which is a reason to think carefully about republication format independently of whether it is permitted.

Rate limits and fair use

This is a small dataset from a small institute and the correct pattern is a single download, held locally, refreshed when a new round is published. There is no reason to poll and no service to poll against; the webmap is a viewer, not an API, and hitting it programmatically is both unnecessary and impolite. Check for new publications quarterly. If you need something the published files do not contain – a specific round, an instrument, a variable that was collected but not released – ask, because a research institute is far more likely to help a request than to notice a crawler favourably.

Licensing changes, and it changes without warning. A dataset that was free for research this year may not be free for commercial or evidential use next year. Confirm the current terms before you build a dependency on it, and record the terms you relied on alongside the data — the licence in force at the time of collection is part of the provenance.

Collecting it

How IPIS Conflict Minerals is actually pulled, in the order you would set it up. Prefer the bulk or export interface over per-item lookups wherever one exists: it is kinder to the publisher, faster for you, and gives a reproducible snapshot rather than a series of point-in-time answers you cannot reconstruct later.

Method	Format	Cadence	Notes
Full dataset download	CSV	per survey round; check quarterly	The primary route. Take the whole file, record the retrieval date and archive it, because updates replace prior versions and there is no rollback.
Webmap inspection	HTML	ad hoc	Site-level context and orientation. Use it to sanity-check anything derived programmatically and to understand what a cluster of points actually looks like.
Analytical report series	bulk	per publication	The methodology, scope and interpretation. These documents are not optional; the raw records are not safely usable without them.
Roadblock and transport axis data	CSV	per publication	Where published, this maps the taxation points between mine and market and is the missing link between site-level observation and supply chain analysis.
Geospatial layer export	JSON	per round	Sites as point geometry for overlay against concessions, protected areas, forest change alerts and conflict event data in your own GIS.

Ingesting it into the platform

Every step below is idempotent and cursor-based: interrupt one and it resumes from where it stopped rather than duplicating rows or losing progress. Collection is recorded per source, so a feed that quietly stops publishing shows up as a stale timestamp instead of silently thinning your coverage.

Register the source as a dated survey — Add IPIS in sources.php with an explicit note that records are field observations valid as of a visit date, not a current state. Every record must carry the visit date prominently, because an analyst who reads this data as current will reach dangerous conclusions.
Schedule discovery rather than collection — Configure collect.php to check quarterly for new rounds under cron.php. There is nothing to stream. The failure mode to guard against is missing a publication for a year, not missing an update for an hour.
Model the site as an entity with observation history — In ingest.php, create a facility entity for each site keyed on identifier and coordinates, and attach each visit as a dated observation rather than overwriting attributes. The change between visits is the analytical product, and an overwrite destroys it.
Normalise armed actor names carefully — Resolve group names to actor entities with alias handling, because groups fragment and rename between rounds. Do this under review rather than automatically; a naive string match will merge two distinct formations or split one across a rename.
Overlay geography at ingest — Use enrich.php to stamp each site with province, territory, nearest town, protected area, mining concession where boundaries exist, and distance to road and border. Location context is what turns a coordinate into an assessment.
Correlate with conflict and observation data — Run correlate.php against conflict event datasets, forest change alerts and, where relevant, extractive licence disclosures. The high-value correlations are a validated site with subsequent armed presence, and a site whose surrounding land conversion is inconsistent with artisanal-scale activity.
Route protection indicators separately — Child and forced labour indicators must not flow into general analytical views. Configure the pipeline so these records are handled under a restricted path with referral to appropriate child protection and labour authorities as the defined output, not as another attribute on a map.
Publish to mission areas with the date attached — Surface sites into organized-crime.php and the human rights views, and to link-analysis.php for the actor-site graph, with visit date rendered on every record and alerting configured for new rounds affecting watched areas.

Registered sources and their last-collected state are listed in sources.php, and the scheduled chain that keeps them current is in automation.php.

How it is wrong, and how to tell

Every dataset is wrong in characteristic ways. Knowing which ways is the difference between using a source and being used by one, and it is the part of source evaluation most often skipped because it is the part that takes work.

For what it claims to be, this is a high-quality dataset, and the reason is that it is a survey with a documented instrument rather than a compilation of secondary reports. Enumerators were physically present, the coordinates were taken in the field, and the observations are the observations of trained researchers rather than aggregated rumour. The methodological documentation is candid about constraints, which is itself a reliability signal. Where judgement is needed is at three points. First, sampling: the set of visited sites is not a random sample of sites and cannot support incidence estimates for the sector as a whole, so any statement of the form a given percentage of mining sites in eastern DRC are militarised needs to be phrased as a percentage of visited sites. Second, observation validity: what a team sees during a visit is a moment, potentially a managed moment, and the absence of an observed armed actor does not establish that the site is free of armed interference. Third, comparability: rounds differ in scope and instrument, so a change between rounds may be a change in the survey rather than in the world. The strongest way to use the data is triangulated – a site record cross-checked against conflict event data for the same period, against UN expert reporting on the named group, and against the analytical report's own regional narrative. Where those agree, the finding is about as solid as anything available for this environment. Where they disagree, the disagreement is usually informative.

Characteristic false positives

The visit-date problem dominates everything: a record read as current will attribute an armed group presence that ended, or miss one that began, and this produces confident statements about a supply chain that are simply out of date.
Absence of an observed armed actor is read as absence of armed interference. Enumerators see a moment, visits can be anticipated, and taxation frequently happens at trading points and roadblocks rather than at the pit.
Site-level statistics are generalised to the sector. The visited sites are an accessibility-biased sample, and percentages computed over them are routinely quoted as though they described all artisanal mining in the region.
Armed group names are matched naively across rounds, merging formations that split or splitting one that renamed, which corrupts any longitudinal analysis of a specific group's expansion.
A single coordinate is treated as the extent of the site, when large artisanal areas are recorded as one point, producing overlay errors against concession and protected area boundaries.
Validation status is treated as certification of clean supply. It is a periodic administrative designation that lags conditions, and the mismatch between it and observed presence is precisely what the data is useful for exposing.
Coverage change is mistaken for real change. A province that shows fewer militarised sites in a later round may simply have had less of it surveyed under a different grant.
Worker counts are used as production estimates. They are headcounts at a moment in an activity with extreme seasonal and price-driven fluctuation, and they do not convert to volumes.

None of these make the source unusable. They make it a source that requires corroboration before an assertion built on it goes into a product, which is true of every source and admitted by few.

Ageing

This source ages faster than almost anything else in this catalogue, because the variable of greatest interest – which armed actor is present – is the variable that changes fastest in eastern DRC. Groups advance, retreat, split, integrate into the national army and re-defect, sometimes within a single season. A record two years old still tells you that the site exists, roughly how large it was, and what mineral is worked there, all of which are relatively stable. It tells you very little that is reliable about who controls it today. Validation status ages on an administrative cycle and is frequently stale even at the moment of publication. Worker counts age with the season and the price. Coordinates and mineral type are the durable fields. A stale record looks entirely credible – a named site, a named group, precise coordinates – and is wrong about the only thing you needed it for. The practical mitigations are to render the visit date on every display of the data, to refuse to make any control-related assertion from a record older than a stated threshold without corroboration, and to pair every site query with a check of conflict event data and UN expert reporting for the same area over the period since the visit. For supply chain due diligence in particular, an old record is a reason to ask a current question, not an answer to it.

What this source feeds

A source is only worth what it lets you conclude. These are the disciplines that collect through it, the mission domains it serves and the data points it yields — every one is a tag, so you can follow any thread from here into the rest of the library.

Collected by these intelligence disciplines

Serves these mission domains

Mining & Resource Crime

Yields these data points

How each sector uses IPIS Conflict Minerals

The same dataset is worked very differently depending on who you are, what authority you hold, and what you are ultimately producing. A military analyst is supporting a commander’s decision; a journalist is meeting a publication standard; an NGO caseworker is protecting a person. The records are shared — the constraints, thresholds and outputs are not.

🎖 Military and defence

For forces operating in or planning around the Great Lakes, mining sites are terrain features with economic and political weight: they are revenue nodes for armed groups, gathering points for large civilian populations, and frequent flashpoints. The dataset gives a coordinate-level layer of where those nodes are, which minerals fund which actors, and where state forces are themselves implicated in illegal taxation – a fact with direct implications for partnering and for the credibility of security sector engagement. It supports pattern-of-life and revenue-denial assessment better than any other open source for the region. The limits are strict: records are dated observations, the sample is accessibility-biased, and the data has no protective marking, meaning it is equally available to the actors it describes. It informs assessment and planning. It is not a targeting product and should not be treated as one, not least because a mine site is a place where thousands of civilians work.

🕵 National intelligence

This is the primary open collection route for the intersection of ENVINT, ECONINT and armed group financing in Central Africa. Its distinctive contribution is granularity: not a general assessment that a group taxes mining, but the specific sites, minerals and interference types, at coordinates, with dates. That supports network analysis linking armed formations to revenue streams, to trading centres, to transport axes and eventually to export and cross-border flows. It also supports a denial-and-deception line: comparing observed conditions with formal validation status and with official production figures exposes where the paper record and the ground diverge systematically. Corroborate against UN group of experts reporting, which covers the same terrain from a different collection base, and against conflict event data. The tradecraft requirement is date discipline – assessments built on stale control attributions are the characteristic failure with this source and they fail in the direction of confidence.

👮 Law enforcement

For investigators working mineral smuggling, sanctions evasion and conflict financing, this dataset establishes the upstream geography that customs, trading and export records cannot. It identifies sites, minerals and armed actors, which supports both the predicate for a supply chain investigation and the questions to put to importers, refiners and smelters about their upstream due diligence. Where a company claims a clean chain from a specific area, the record of conditions at sites in that area is a direct test of the claim. Evidentially, this is survey research: it will not prove that a particular consignment financed a particular group, and it should be used to direct enquiry toward documentary and testimonial evidence obtained through process. Two further points. Child and forced labour indicators are protection matters requiring referral rather than investigation in the ordinary sense. And any operation informed by this data should assume that the communities at these sites are exposed to reprisal if the source of information is inferred.

🔍 Private investigation and corporate security

For corporate due diligence, this is the reference dataset behind conflict minerals compliance in the Great Lakes, and clients in the electronics, automotive, aerospace and jewellery sectors have direct regulatory exposure to the questions it answers. The standard task is to test a supplier's stated upstream origin against observed conditions at sites in that area, and to identify where a smelter or refiner's declared sourcing region includes sites with documented armed interference. The output is a risk finding under the applicable due diligence framework, not an accusation, and framing it that way is both more accurate and more useful to the client. The professional care points are the ones that recur: date the record, do not generalise from visited sites to a region, do not treat validation status as assurance, and state explicitly that the absence of an observation is not evidence of clean sourcing. Pair with smelter-level scheme data and with the analytical reports rather than working from the coordinates alone.

📰 Journalism and OSINT media

For reporting on conflict minerals, supply chains and the gap between corporate claims and upstream reality, this dataset provides something rare: specific, mapped, independently collected evidence that can be checked. The durable story forms are the branded product traced back toward a militarised site, the officially validated mine found under armed control, the state force taxing the mine it is deployed to protect, and the smuggling route across a border into a neighbouring state's export figures. The craft requirements are date discipline, precision about what the data does and does not establish, and care in the field. Reporting from these sites carries real risk for local fixers, interviewees and mine workers, who remain behind after publication; the security practices used by experienced correspondents in the region exist for good reason. Never publish material that identifies children observed at a site, and treat any labour finding as a protection issue with a referral pathway rather than as colour.

🌍 NGO, humanitarian and human rights

Human rights, child protection, labour rights and peacebuilding organisations use this data for programme targeting, advocacy and monitoring, and for many of them it is the only source that puts their operational geography and the conflict economy on the same map. Concretely: identifying where child labour indicators cluster for protection programming, identifying communities under armed taxation for humanitarian access planning, monitoring whether formalisation and validation efforts are producing changes on the ground, and holding downstream companies to account against a documented upstream record. Two obligations follow. First, protection indicators concern identifiable vulnerable populations at identifiable locations and their handling should follow child safeguarding and do-no-harm standards rather than open data instincts. Second, community and enumerator security must shape publication: a report that names a site and an armed actor can be read by that actor, and the people who supplied the observation live there. Coordinate disclosure with local partners.

🎓 University and research

The dataset has supported a substantial literature on conflict financing, artisanal mining livelihoods, the effectiveness of due diligence regulation, and the political economy of eastern DRC, and it is unusually well suited to it because the collection methodology is documented and the institute is responsive to methodological enquiry. The research uses that hold up are those that respect its nature as a purposive, accessibility-constrained survey: case-based analysis, within-sample comparison, change analysis over repeat-visited sites, and validation studies against independent data. The uses that do not hold up are incidence and prevalence estimates for the sector as a whole, and cross-round comparisons that ignore changes in geographic scope and instrument. Ethics review is warranted for any work engaging with the protection indicators or with community-level identification. Where quantitative work is planned, contact the institute about instrument comparability between rounds first; that information is not fully recoverable from the published files and it determines whether your panel is a panel at all.

Playbook: working IPIS Conflict Minerals end to end

A repeatable sequence from first pull to finished product. Each phase states what you are trying to establish, not merely what to click — the objective is a defensible chain of reasoning, not a completed checklist.

Phase 1 — Read the accompanying report before opening the dataset

The methodology, geographic scope and constraints for each round live in the analytical publications, not in the file. Without them you cannot tell whether a sparse province was peaceful, unfunded or unsafe, and that distinction governs every conclusion you might draw.

Phase 2 — Establish the temporal frame of your question

Decide whether you are asking about conditions at a past date or about conditions now. This source answers the first directly and the second only as a starting hypothesis. Conflating them is the single most common and most consequential error with this data.

Phase 3 — Build the site inventory for your area of interest

Extract all sites within your geography with their full visit history, minerals and worker estimates. You are establishing the physical universe: how many sites, how large, working what, how recently anyone was there.

Phase 4 — Separate the durable attributes from the volatile ones

Coordinates, mineral type and rough scale persist. Armed actor presence, interference type, workforce and validation status do not. Structure your analysis so that conclusions rest on the durable fields and the volatile ones are treated as dated hypotheses requiring corroboration.

Phase 5 — Reconstruct the trajectory for repeat-visited sites

Where a site has been visited more than once, lay the visits on a timeline. A site that moved from no armed presence to taxation by a named group, or from red to green validation while presence continued, is a specific, checkable finding of a kind the single-visit records cannot produce.

Phase 6 — Overlay the formal picture

Place sites against mining concessions, protected areas and administrative boundaries. Artisanal activity inside an industrial concession, inside a national park, or inside a nominally validated zone each raise different questions with different responsible parties.

Phase 7 — Connect sites to the trading and transport network

Use nearest town, road access and, where published, roadblock and transport axis data to trace how material moves from the site toward a trading centre and a border. Taxation frequently occurs on the route rather than at the pit, and the route is where the chain becomes traceable.

Phase 8 — Corroborate armed actor attributions independently

For every group named at a site you intend to rely on, check UN expert reporting, conflict event data and specialist regional analysis for the same period and area. Group names are unstable and control changes fast, and this step is what prevents an out-of-date attribution reaching a client or a report.

Phase 9 — Test against the downstream claim

If the question originated in a supply chain, compare what the company says about its sourcing region with what was observed at sites in that region. The output is a documented risk finding under the applicable due diligence framework, phrased as risk rather than as accusation.

Phase 10 — Handle protection indicators through a separate pathway

Child and forced labour observations are not analytical attributes. Route them to the appropriate child protection and labour authorities and to organisations with a protection mandate, under safeguarding rules, and keep them out of general-purpose maps and reports.

Phase 11 — Assess the harm your output could cause on the ground

Before publishing, consider who is identifiable from your product: the communities at a named site, the local researchers who visited it, the partners who facilitated access. Adjust specificity accordingly and, where possible, decide it with people who have local knowledge rather than for them.

Phase 12 — Archive the version and set a re-check trigger

Store the dataset version and retrieval date with your analysis, register the sites and actors of interest in watchlist.php, and set a review trigger for the next publication round and for significant conflict events in the area. Findings from this source have a shelf life and it should be an explicit one.

The platform ships this as a step-checked workflow in playbooks.php, so progress is recorded against a case rather than held in someone’s head.

What to pair it with

No single source carries a finding. These are the datasets that corroborate, extend or contradict this one — and a source that contradicts is worth more than one that agrees, because it is the only thing that will tell you when you are wrong.

Source	Relationship	What it adds
UN Group of Experts reporting on the DRC	corroborates	Independent investigative reporting on armed groups, mineral smuggling and sanctions violations, with a different collection base and formal standing.
ACLED	extends	Conflict event data for the same areas, essential for establishing what happened between a site visit and today and for corroborating actor presence.
Global Witness	corroborates	Investigative depth on the trading houses, exporters and downstream companies that connect these sites to international markets.
EITI	contradicts	The formal-sector revenue picture. The contrast between reconciled industrial payments and the observed artisanal economy is itself a finding about what formal disclosure misses.
Global Forest Watch	extends	Satellite observation of land disturbance around mapped sites, useful for detecting expansion between visits and for distinguishing artisanal from mechanised scale.
OECD due diligence guidance	prerequisite	The framework that defines what constitutes a red flag and what a company is expected to do about it, and therefore what your finding means in compliance terms.
OpenSanctions	extends	Designations against armed group leadership and associated entities, connecting site-level presence to formal restrictive measures.
UN Comtrade and national export statistics	corroborates	Mirror-trade analysis of tin, tantalum, tungsten and gold flows, the standard method for detecting smuggling through neighbouring states.
OCHA and humanitarian data services	extends	Population, displacement and access data for the same territories, needed to understand the civilian context around any mining site.

Legal, ethical and operational constraints

The data is published openly and its collection and analysis carry little legal risk in themselves. The constraints are on use and on consequence. Attributing control of a specific site to a named armed group, or implicating a named company in sourcing from it, is a defamation exposure and, where the group or its associates are sanctioned, potentially a matter with sanctions implications for anyone who acts commercially on the information. Findings should be framed as observations on a date with an explicit basis. Conflict minerals due diligence obligations differ by jurisdiction and by market: mandatory frameworks exist in several major economies, with different scoped minerals, different thresholds and different reporting requirements, and at least one of them has been the subject of litigation that changed what companies must say. Confirm the current obligations applicable to your client or organisation rather than working from a general recollection of the regime. Protection law is the harder constraint: records indicating child labour or forced labour concern identifiable vulnerable people at identifiable places, and in most jurisdictions and under most organisational codes the correct response is referral to a competent protection authority, not independent investigation. Finally, do-no-harm is a legal as well as an ethical standard in many donor and humanitarian frameworks, and publishing in a way that endangers communities or enumerators can breach it.

Operational security

Downloading a published dataset from a research institute is unremarkable and reveals almost nothing. The exposure created by this source runs in the other direction, toward the people on the ground. Site records name locations, sometimes armed actors, and implicitly identify the communities who spoke to enumerators. Any product you build that combines these records with additional specificity – a company name, a trading route, a named individual – increases the risk to people who live at the site and to the local researchers who will visit it next. Armed actors in this region monitor reporting about themselves and have acted on it. The practical rules are: do not add identifying detail that the published data did not carry; consult local partners before publishing anything that narrows attribution to a community; do not share unpublished field detail through insecure channels; and if your work will be operational in the area, assume that the interval between your product and a visible response on the ground can be short. Where an investigation is genuinely sensitive, work from an archived local copy so that no query pattern reveals which sites you are examining.

Two rules that hold regardless of jurisdiction. Collection that is lawful is not automatically proportionate, and a dataset assembled for one purpose does not carry consent for another. Where the records concern identifiable people, the question is not only whether you may hold the data but whether holding it serves the purpose you are accountable for.

Is it earning its place?

Sources accumulate. Feeds get added during an incident and are never reviewed again, and a decade later the pipeline is carrying dead weight that nobody dares remove. These are the measures that show whether IPIS Conflict Minerals is contributing anything, and they are worth baselining now so the answer is available later.

Median age of the site records underpinning your active assessments, which for a source with this decay rate is the most important quality metric you can track.
Proportion of armed actor attributions you have corroborated against an independent source for the period since the visit, rather than relying on the visit record alone.
Coverage ratio of your area of interest: how many sites in the geography have ever been visited, and how many in the last two rounds, which sets the ceiling on any claim you make about the area.
Number of repeat-visited sites in your analysis, since trajectory findings are the source's highest-value output and require more than one observation.
Rate at which site records successfully overlay onto a concession, protected area or administrative unit with a boundary of known vintage, which measures the health of your context layers.
Count of protection-indicator records routed through the referral pathway rather than into general analysis, which should be all of them and is an auditable control.
Number of due diligence or investigative findings where this source supplied the upstream geography, distinguishing contribution from citation.
Time from publication of a new survey round to its ingestion and re-assessment of affected findings, which determines how long your stale conclusions stay in circulation.

Beware of volume. Indicator counts rise easily and say almost nothing. Unique contribution — findings this source produced that no other source in your stack would have — is the measure that matters, and it is usually far lower than anyone expects.

Tradecraft notes

The distinctions that separate a competent analyst from a fast one:

Put the visit date on every rendering of this data, in every product, without exception. It is the field that determines whether the record means anything, and it is the field that gets dropped first.
Presence is not control and absence of an observed actor is not absence of interference. Taxation happens at roadblocks and trading points as much as at the pit, and a visit is a moment that can be prepared for.
Never generalise from visited sites to the sector. The sample is accessibility-constrained in exactly the direction that biases it against severity, and percentages computed over it describe the sample only.
Handle armed group names as unstable identifiers requiring alias management under human review. Automatic matching across rounds will silently merge or split formations and corrupt every longitudinal conclusion.
Treat validation status as an administrative designation with a lag, not as assurance. The gap between a green designation and an observed armed presence is one of the most useful findings the dataset produces.
Read the round's report before using the round's data. Scope, funding geography and instrument changes explain most apparent trends, and they are only documented in the narrative.
Route child and forced labour indicators to protection actors and keep them out of analytical products. This is a handling rule, not a judgement call, and it exists because the alternative endangers children.
A single coordinate can represent a large working area. Check what a site looks like on imagery before making any overlay claim that depends on a boundary within a few hundred metres.
Assume your output will be read by the actors it names and by the communities it locates. Decide specificity with people who know the local risk, and default to less.

Questions analysts actually ask

Can I use this to certify that a supply chain is conflict-free?

No. It can identify risk at sites in a sourcing area as of a visit date, which is a red flag input to a due diligence process. It cannot establish that any particular consignment is clean, and absence of an observation is not evidence of absence of armed interference.

How current is the data?

Each record is current as of its visit date and no later. Rounds are published when programmes complete, and intervals between visits to the same site can be years. For anything about current conditions, treat the record as a hypothesis and corroborate with conflict event data and expert reporting.

Why is gold under-represented relative to its importance?

Gold sites are numerous, small, dispersed and harder to reach, and the traceability schemes and donor programmes that fund much of this work have historically focused on tin, tantalum and tungsten. The economic significance of artisanal gold is greater than its share of the dataset suggests.

Does a green validation status mean the site is clean?

It means the site passed an administrative qualification process at some point. Validation is periodic and lags conditions, and the dataset documents sites with formal designations where armed presence was subsequently observed. The discrepancy is a finding, not an anomaly to be explained away.

How reliable are the armed actor identifications?

They reflect what trained enumerators could establish locally at the time, which is a reasonable standard for this environment and not a definitive one. Groups fragment and rename constantly. Corroborate any attribution you intend to rely on against UN expert reporting and conflict event data for the same period.

Can I compare numbers between survey rounds?

Only after establishing that the rounds covered the same geography with the same instrument, which they frequently did not. Apparent changes in militarisation between rounds are often changes in what was surveyed. The reports document scope; ask the institute if the reports do not resolve it.

What should I do with child labour indicators?

Route them to child protection authorities and to organisations with a protection mandate, following safeguarding procedures. They should not be published as analytical attributes, mapped in general-purpose products, or used to direct investigation of individuals. The referral pathway is the output.

Is this data admissible or usable in a regulatory filing?

It is commonly cited in due diligence reporting as a risk input, which is an appropriate use. As legal evidence it is survey research and would need to be introduced with its methodology and its limitations. Do not build a case on it; use it to identify where to obtain evidence through process.

Does the platform infer site conditions or actor presence with a model?

No. Sites, visits, minerals and actor observations are ingested exactly as published, with visit dates preserved as first-class data. The only language model involvement is the Summarise skill in copilot.php, which writes prose about existing records and originates no observation, relationship or attribution.

Standards, formats and interoperability

What this source speaks natively, and what it has to be translated into before a partner can consume it. Work that arrives in a recognised format is easier to defend, easier to hand over and easier to automate against:

OECD Due Diligence Guidance for Responsible Supply Chains of Minerals from Conflict-Affected and High-Risk Areas, the framework that defines the red flags this data supplies.
The ICGLR Regional Certification Mechanism, the regional certification scheme for minerals from the Great Lakes, whose designations interact with the site validation status recorded here.
National conflict minerals disclosure regimes in major markets, which impose reporting duties on downstream companies and have differing scoped minerals and thresholds; confirm current requirements rather than assuming.
Industry upstream traceability and assurance schemes for tin, tantalum, tungsten and gold, which operate at smelter and refiner level and connect to this data at the sourcing region.
UN Security Council sanctions regimes covering the region, including designations of armed group leadership relevant to actor attributions.
ILO conventions on child labour and forced labour, which set the standards against which protection indicators in this dataset are assessed and referred.
Common geospatial formats – GeoJSON, shapefile, CSV with coordinates – which is how the site data actually moves between systems.
STIX 2.1 and MISP as platform export formats, with sites mapped to location objects and armed actors to identity objects, always carrying the observation date.

References

Primary documentation and authoritative references for this source. Publishers revise and retire material, so treat the retrieval date as part of the citation and re-check before relying on any of it in a formal product.

IPIS Research — International Peace Information Service. The institute, its publications and its data. Read the report accompanying a survey round before using that round's data.
IPIS maps and data — IPIS Research. The webmaps and downloadable datasets of visited mining sites, the fastest route to site-level context and to checking what a cluster of points represents.
OECD — OECD. The Due Diligence Guidance for minerals from conflict-affected and high-risk areas, which defines the compliance meaning of everything in this dataset.
United Nations Security Council — United Nations. Group of experts reporting and sanctions committee material on the DRC, the principal independent corroboration for armed actor findings.
ACLED — Armed Conflict Location and Event Data Project. Conflict event data for establishing what has happened in an area since a site was last visited.
Global Witness — Global Witness. Investigations into the trading and export networks that connect these sites to international markets.
International Labour Organization — ILO. Standards and guidance on child labour and forced labour, and the correct frame for handling protection indicators in this data.
UN Comtrade — United Nations Statistics Division. Bilateral trade statistics for mirror analysis of mineral flows through neighbouring states.
UN OCHA — United Nations Office for the Coordination of Humanitarian Affairs. Humanitarian access, displacement and population data for the civilian context around mining areas.
Global Forest Watch — World Resources Institute. Satellite observation of land disturbance for detecting site expansion between field visits.

Link integrity: every reference above was verified with a live request when this page was generated. Where a publisher had moved or withdrawn a document, the link was repointed at a preserved copy in the Internet Archive and marked as archived. Anything with no reachable copy anywhere had its link removed rather than left to rot — the source is still credited, it simply cannot be linked.

Put it into practice

The Quantus Intel threat intelligence platform operationalises this source: IPIS mining sites are ingested as dated field observations with visit history preserved, overlaid against concessions, protected areas and conflict events, and linked to armed actor entities so that trajectory rather than snapshot drives the assessment.. Browse the full source catalogue, or follow any tag above into the rest of the library.