Online Research Services, Scoped and Sourced

A folder of links is not research

Links are evidence inputs. Research turns them into a bounded answer by stating what was sought, how sources were selected, what each field means, which records were excluded, how coverage was assessed and what could not be established.

Without that method, a buyer cannot tell whether an empty field means “not published,” “not applicable,” “not checked” or “researcher missed it.” They cannot distinguish a company’s own claim from an official record, or a current page from an archived statement.

The failure is worse than an incomplete assistant lookup because a project deliverable is likely to enter a board paper, investor memo, pitch, operating plan or client recommendation. It carries the apparent authority of a finished study.

OVELITHUB has delivered 130+ projects across different business contexts. That experience shapes project control, but no project count proves a new dataset is complete or accurate. The method, sample, source log and limitations provide that evidence for the current engagement.

The project types we scope

• Structured list building: identify eligible organisations, locations, products, documents or events and verify required fields against approved sources.
• Landscape and vendor mapping: define a market or solution universe, classify participants and compare capabilities, geographies, business models or public proof.
• Regulatory and requirement gathering: locate responsible authorities, current source documents, forms, dates and stated obligations across agreed jurisdictions, with interpretation reserved for qualified advisers.
• Competitor and pricing compilation: capture public positioning, features, packaging, displayed prices, conditions and access dates under fixed definitions.
• Organisation and contact research at volume: compile purpose-limited business and role evidence where lawful, without turning the output into an outreach campaign.
• Literature and public-document sweep: search defined databases, repositories, authority sites, filings or publications against a documented query and eligibility rule.
• Recurring monitoring: establish a baseline universe and check it on a cadence, delivering a dated change log rather than rebuilding the full narrative.

A named person handling varied daily lookups fits an online research assistant. Market, competitor and pricing evidence for a continuing strategy role fits a market research assistant. Outreach-ready account intelligence belongs to prospect research services.

The research scope is written before collection

The scoping conversation establishes:

1. Decision. Which board, pitch, investment, planning or operational decision will the evidence support?
2. Question. What can the project answer, and which conclusion remains outside the research?
3. Unit of analysis. Is one row a company, site, product, document, regulation, person, transaction, study or event?
4. Eligibility. Which inclusion and exclusion rules can two researchers apply consistently?
5. Universe and coverage. Is the target a known complete register, a constructed list, a sample or an open-ended landscape?
6. Geography and period. Which markets, languages, publication dates, observation dates and historic windows apply?
7. Sources. Which primary, official, licensed, secondary and direct-contact sources are required, allowed or prohibited?
8. Fields and definitions. What is captured, in which format, and how are unknown, not applicable and conflicting values encoded?
9. Outputs. Which summary, dataset, source log, appendix, presentation or change file must be usable by whom?
10. Quality and deadline. Which fields receive second review, which thresholds pause scaling, and when is the decision?

Ambiguity here creates the largest rework. A client may think “UK software vendors” means UK-headquartered companies. A researcher may include any company selling into the UK. Both can work diligently and produce incompatible outputs.

Worked example: a vendor landscape with testable criteria

Suppose a consultant requests “a list of UK appointment-booking platforms for independent salons.” The project cannot begin at scale until those words become rules.

Include

• a currently accessible product site and evidence observed during the project window;
• business-facing software that publicly describes online appointment booking as a live capability;
• evidence that independent salon or beauty-service businesses are an intended customer group;
• service available to UK customers, with the evidence source recorded;
• a distinct product or company, with parent and brand relationship noted; and
• required fields supported by a company source or explicitly marked not found.

Exclude

• consumer-only booking directories with no business software offer;
• general calendar tools without salon-facing positioning or relevant product evidence;
• consultants, agencies and implementation partners that do not supply the platform;
• discontinued products, inaccessible sites or products lacking current evidence in the agreed window;
• platforms limited to a market outside the UK where no UK availability evidence appears; and
• duplicate brands or reseller pages representing the same underlying product.

The comparison fields could include legal or trading name, URL, headquarters as claimed or registered, UK availability evidence, target segment, booking channel, relevant public feature, displayed pricing status, integration claims, source, source date, access date and uncertainty. The project does not infer adoption, market share, quality or commercial success from website presence.

Edge cases enter a decision log. If a marketplace also sells business software, the project owner decides which evidence qualifies it. That decision then applies to later records and is available to the second reviewer.

Coverage is stated honestly

A known universe may come from an official register, supplied database or finite event list. The report can state how many eligible records were reviewed, excluded, inaccessible and included, along with reconciliation to the source total.

An open landscape has no provable complete denominator. The project instead states the search routes, queries, directories, associations, sources, snowball criteria, language, geography, stopping rule and date. It may call the result “identified eligible organisations under the stated method,” not “every provider.”

A sample states the target population, frame, selection method, size and limitations. Convenience, purposive, stratified, random and quota selections answer different questions. OVELITHUB does not describe a convenient online sample as representative without an appropriate design.

Excluded records remain useful evidence. The exclusion log records candidate, reason, supporting source and reviewer where required. This prevents the same ineligible item from being rediscovered and added later.

A coverage limit is a quality signal because it tells the reader where the evidence stops. Any provider promising complete coverage of a fragmented, changing market without a defined register is overselling.

Every finding has a source, and a second researcher checks the risk

Each material field or claim records source title, publisher, URL or locator, document or update date where available, access date, extracted evidence within permitted limits, source class and researcher note. Self-published company information is labelled as a company claim rather than independent validation.

Before scaling, a second researcher independently checks a pilot sample. They receive the same scope, definitions and source rules but verify against source evidence rather than accepting the first researcher’s judgement.

The review design may include:

• 100% checking of identity, eligibility or high-consequence fields;
• a random sample of ordinary records, selected outside the first researcher’s control;
• targeted checks of missing values, difficult jurisdictions, new researchers and unusual sources;
• duplicate and consistency tests across the full dataset; and
• re-review after a definition change or defect threshold is crossed.

The re-check share is agreed against consequence, budget, population and expected variability. A small sample cannot support a precise claim about the entire dataset without an appropriate statistical design. The quality report states inspected unit, sample method, count, fields checked, defects by severity, disagreements, corrective action and residual limitations.

Disagreement is useful. It can expose an ambiguous definition rather than a careless researcher. The project lead resolves the rule, updates the decision log and determines whether earlier records require rework.

The deliverable has four parts

Decision summary. States the question, method, coverage, material findings, confidence, contradictions, open questions and what the evidence can and cannot support.

Structured dataset or comparison. Uses stable field definitions, validation values, identifiers and filters in an agreed spreadsheet, CSV, database-ready format or presentation table.

Source log. Provides the evidence trail with dates, source types, links or locators, access limitations and permitted extracts. Paid or restricted sources follow licence limits.

Method and limitations note. Records inclusion, exclusions, search routes, sample, coverage, second review, definition changes, unknown values, unresolved disagreements and recommended refresh.

The summary must state what could not be established. A blank field is encoded separately as not found, not applicable, inaccessible, conflicting, pending or not checked. These states prevent downstream users from treating absence as a negative fact.

What we will not claim

Desk research cannot access non-public pricing, realised contract terms, confidential customer lists, proprietary market share, private strategy, undisclosed product plans or restricted records without authorised access. A company statement can establish that the company made the claim on the date observed; it cannot prove performance.

Conflicting sources remain visible. OVELITHUB does not average incompatible definitions, choose the newest page automatically or turn a secondary citation into a primary source. The client sees which evidence supports each wording.

Public statistics can come from responsible government portals such as national statistical offices, regulators or multilateral data producers. Commercial databases such as Statista can help discover comparable indicators. Every quoted figure still needs its own inline citation, definition, period and geography; a platform name is not a citation.

Specialist legal, clinical, financial, engineering, scientific and security conclusions remain with qualified professionals. The research team can gather approved sources and structure evidence for their review.

Volume and verification compete for the same budget

Automated tools and scraping can collect predictable fields from consistent permitted sources at lower marginal cost. They are often the better fit when the page structure, access rights, fields and update logic are stable. OVELITHUB can refer that requirement to web scraping services.

Automation becomes less reliable when sources fragment, labels change, identity resolution matters, inclusion requires judgement, dates are ambiguous or the evidence must be interpreted in context. Human verification can handle these cases, but the cost rises with field depth and source difficulty.

The client chooses knowingly between:

• more records with fewer verified fields;
• fewer records with deeper primary-source confirmation;
• automated collection plus human review of exceptions and a sample; or
• full human research for a smaller high-consequence universe.

A pilot sample is run before scaling. It measures eligible yield, minutes per record, source mix, inaccessible rate, exception types, inter-reviewer disagreement and observed defects. The full quote and delivery estimate are revised when the pilot contradicts the assumptions.

Accuracy does not automatically fall because a list is large. Risk rises when verification effort per item stays fixed while source variability and edge cases increase. The control is to measure the actual pilot and change scope, automation or review depth.

Legal and ethical boundaries apply to collection

OVELITHUB respects authorised access, website and platform terms, database licences, intellectual-property restrictions, technical controls and the client’s lawful purpose. The project does not evade access barriers, create false identities or misrepresent the researcher to obtain information.

When compiling records about identifiable individuals, “publicly available” does not mean unrestricted. For projects subject to UK data-protection law, the ICO’s current data-protection principles guidance centres lawfulness, fairness and transparency, purpose limitation, data minimisation, accuracy, storage limitation, security and accountability. The client determines its lawful basis, transparency, rights, retention, transfer and direct-marketing obligations with appropriate advice.

The scope limits personal fields to what is necessary for the stated purpose, records source and date, defines a correction route and avoids sensitive or special-category information unless the client establishes a lawful, necessary and controlled requirement.

Direct contact with an organisation uses an approved identity, purpose and script. Researchers do not pose as customers, applicants, investors, journalists or officials. No contact commitment is made without authority.

Verification depth changes price more than raw volume

Fixed price against written scope suits a defined question, unit, criteria, coverage, fields, source plan, quality method, outputs and deadline. Change control applies when the universe, definitions, markets or deliverable changes.

Price per verified eligible record can suit list work when eligibility and field completion are testable. The agreement still needs an approach for ineligible candidates, unavailable sources, minimum volume, duplicate records and records requiring extra effort.

Pilot plus scale is appropriate when yield and source difficulty are unknown. A paid pilot creates evidence for the final unit economics and timeline.

Price drivers include number of jurisdictions and languages, universe construction, verification fields, primary-source requirements, identity resolution, direct contact, paid access, personal data, second-review share, output complexity, update frequency and deadline compression. Ten fields from one stable register can be cheaper than two fields requiring separate manual sources.

The timeline shows scope approval, pilot, client decision, collection waves, quality gates, draft, review and final. A deadline attached to a board or pitch includes client response dates; unanswered scope questions consume delivery time.

Your brief returns a method before a quote

OVELITHUB restates the decision, question, unit, inclusion, exclusions, geography, period, target coverage, source hierarchy, data fields, quality plan, outputs, timeline, client dependencies and limitations. Unknowns become assumptions or a paid pilot.

You then receive a coverage estimate, fixed price or verified-record proposal, change-control rule and sample output shape. A buyer can inspect what “complete” and “verified” mean before comparing prices.

If the work is still a series of small questions rather than one deliverable, start with hourly online research support and let the recurring pattern reveal the eventual project.