Research method
How the research classes and QA work
Relevance comes first. A company must meet the written targeting criteria; a project signal is an additional evidence requirement for one class, not the definition of every lead.
Our method keeps company fit, observed evidence, interpretation and uncertainty separate. It does not infer private buying intent.
Start with a research unit and a written profile.
- Define the brief. Agree the industries, territory, company criteria, exclusions and evidence requirements. Specify a company account or a facility record; do not count several sites as separate companies.
- Check source use. Identify sources that may be reviewed and what may be included in customer-facing output. Public access does not settle reuse rights.
- Resolve the entity. Check which company and site the evidence concerns; review parent/subsidiary relationships, duplicates and conflicts.
- Record the fit. Support each material criterion with a source reference and relevant dates. State why the account is included and what remains unresolved.
- Apply the class check. A project-signal record needs evidence of a relevant project or development as well as account fit.
- Review before release. Check the criteria, source posture and claims. A research pass is not, by itself, authority to deliver data or contact anyone.
Two classes, with project research status kept separate.
Industry-matched leads (Look Alike)
A rules-based match to the agreed profile. No current or upcoming project is required. Where project research was not part of the brief, say not assessed—not required. Where an agreed bounded search found none, say no qualifying signal identified in the bounded review. Do not turn either statement into “no project exists”.
Opportunity-signal leads
A profile-matched account with a qualifying project or development signal supported by dated evidence. State the actual stage and the unknowns; do not infer an open tender, unawarded package or purchasing contact.
Both classes require source-supported fit. Our 90% industry-matched / 10% opportunity-signal mix is a target, not a guarantee. Counts and class treatment are agreed in writing. Neither class is automatically a marketing-qualified lead.
Read the record in three layers.
- Observed fact
- What the source actually supports, attributed to that source and its date. An announcement of a proposed expansion is not evidence that construction or procurement has begun.
- Interpretation
- Why the fact may be relevant to the supplier’s range. An application hypothesis is not a verified technical match.
- Unknown
- Information not established: for example specifications, timing, procurement route or package availability. State the next useful check rather than filling the gap.
Keep publication dates distinct from retrieval or review dates. A recent review does not make an old source current. Reconcile conflicting sources before relying on a material claim.
Several references can still lead back to one claim
When a research record lists several references, ask which statements they actually support—and where those statements originated. Different websites do not necessarily provide independent evidence.
This wholly fictional exercise concerns invented company PROV-C01 and site PROV-S01 during one stipulated period. Six invented documents describe a proposed packaging line, permission for an extension, equipment delivery or operation. There are no real source dates or URLs. The origins below are stipulated teaching inputs, not independently verified real-world facts.
Count each claim separately. A reference appearance is one document's assertion of that claim. A stated origin is the explicitly identified firsthand statement reached by tracing its references. A missing origin stays unresolved.
| Claim | Reference appearances | Distinct stated origins | Trace or remaining gap |
|---|---|---|---|
| Proposed-line announcement | 4 | 1 | D01/A01; D02/A02 → A01; D03/A03 → A02 → A01; D05/A06 → A01 |
| Extension permission | 1 | 1 | D04/A04, its own decision |
| Skid delivery | 1 | 1 | D05/A05, its own delivery statement |
| New line operating | 1 | 0 | D06/A07; origin unresolved |
The document and assertion labels above abbreviate the original PROV-prefixed identifiers. Seven claim appearances occur in six documents. The three stated origins concern different claims; they are not three confirmations that the line is operating. Neither count is a quality score, probability, mandatory threshold or vote.
Trace repeated statements. D01 is the operator's announcement. Trade digest D02 attributes its report to D01; industry roundup D03 identifies D02 as its sole source. Contractor document D05 also repeats the announcement, attributing that sentence to D01. All four appearances lead back to A01—not four independent observations of a completed line.
Read mixed documents claim by claim. D05 separately reports the contractor's own delivery of a labelled skid. That delivery statement has its own stated origin, A05; the announcement sentence does not. Delivery does not establish installation, acceptance, commissioning or operation. D04 supports the stipulated permission decision only: conditions and their fulfilment are outside this exercise, so permission does not establish compliance. Evidence about an adjacent project stage does not establish the operational claim.
Do not fill a missing origin. Directory D06 says the line is operational but names no underlying statement, visit or evidence. Retain the attribution with origin unresolved; operational claim unverified. No cited parent is not proof of firsthand knowledge. It is also not proof that the claim is false.
A buyer can ask: “For the claim that the new packaging line at PROV-S01 is operating during the exercise period, please identify the underlying attributable, dated, scope-matched evidence and its permitted review route.” Keep that conclusion unverified pending the answer without discarding useful announcement, permission and delivery context.
Change one fact. Suppose D03 instead supplies its own directly documented observation of operating machinery at this site during the exercise period. The machinery-operation observation now has a distinct stated origin. It does not establish that this is the new packaging line until that identity is resolved. In a real case, check source authenticity, documentary adequacy, company/site and equipment identity, observation and publication dates, fitness for the intended use and permitted review. D06's lineage remains unresolved. No generic two-source rule or automatic acceptance follows.
Use the maintenance guide for actual contradictory claims; missing provenance alone is not a contradiction. For a written evidence and handover scope, see the service page. Start with a non-sensitive description of the research decision, site scope and period—not an evidence upload.
A score cannot replace the fit decision.
The written criteria determine whether a record is accepted, held for review or rejected. Keep the decision reason visible. A score cannot rescue an excluded account or supply missing project evidence.
The detailed fictional opportunity brief includes an illustrative prioritisation score. It is not a buyer-intent score, a probability of sale or a quality ranking between the two lead classes.
Sources and data have boundaries.
Company-level research is the default described on this site. Named-contact fields and outreach require a separate agreed scope and applicable source, privacy, suppression and security controls.
Research may use approved public sources and, only where contracted and permitted, licensed sources. Evidence retained internally is not automatically available for customer copying. Unknown source or customer-use rights are a hold, not a permission.
No confidential client information belongs in a public example. The published samples are fictional and do not demonstrate live-data clearance, a software platform or client results.
Review, correct and learn within the agreed scope.
Review dates and triggers are agreed for the research, rather than presented as continuous monitoring. If a material fact is challenged, identify the record and claim, inspect the evidence, and record the correction or unresolved question.
Acceptance, replacement windows and any replacement limits are agreed in writing. A missing opportunity signal must not be silently replaced with an industry-matched record to make a batch look complete.
Commercial outcomes can inform later work only when suitable client feedback is available. No meeting, pipeline or conversion improvement is claimed here.