How recommendations work
SourcrLab is not trying to be the largest recruitment-tech catalogue. The aim is to make buying decisions more defensible. We do not maintain one universal ranking of recruitment tools: recommendations are built around fit, evidence is separated from judgement, and unknowns are allowed to stay unknown.
1. Evidence has a visible status
On editorially reviewed pages, material claims are given provenance so a reader can tell what we checked, what a vendor stated, what SourcrLab concluded, what a practitioner reported, and what we could not establish. The five fact-level labels below are a closed set.
Checked against the vendor's own public page or documentation on a known date.
Stated by the vendor; not independently verified by us.
Our editorial judgment, labeled as judgment and never dressed up as measurement.
Reported by practitioners; useful evidence, but still anecdotal unless corroborated.
We do not know or could not establish it. We say so instead of guessing.
Tool pages also carry a page-level evidence status: Listed, Verified, Researched or Hands-on. That status describes how deep the editorial work behind the page goes; it does not turn every statement on the page into an independently verified fact.
2. Decisions are tested on fit, not one score
The Advisor and new decision content use the Decision Stack below as a framework. A tool can be strong on several dimensions and still be the wrong purchase because it fails on the one that matters in your situation. Older comparison and guide pages are being brought into line with this approach rather than treated as proof that the whole site already follows it perfectly.
Problem fit
Does it solve the actual bottleneck?
Workflow fit
Does it fit how recruiters actually work?
Stack fit
Does it integrate, or duplicate what you already own?
Economic fit
Does expected value justify total cost?
Risk fit
Are privacy, security and compliance acceptable?
Implementation fit
Can your team realistically deploy and adopt it?
We removed the single site-wide composite score on purpose. Where a number still appears, such as an Advisor fit score, it is scoped to the answers in that decision and explained by the factors behind it. It is not presented as a universal verdict on the product.
3. Fact and opinion stay separated
On reviewed pages, material claims about pricing, integrations and product capabilities are tied to a source, attributed to the vendor, or left explicitly unverified. Statements such as 'best for a small agency' are editorial judgement, not measurements. External review data stays attributed to its original platform and is never re-branded as SourcrLab research.
4. Reviewed pricing needs a source, date and real buying context
On editorially reviewed tool pages, public pricing claims are checked against a primary source and dated. We keep the currency and billing basis we actually found, and include material limits such as seats, credits, add-ons or annual billing when they change the buying decision. If the vendor does not publish a price, we say so rather than estimate it.
Pages that have not yet received an editorial pass may still show catalogue data imported from public listings or vendor material. Those pages explicitly say that the information is a starting point and has not yet been verified by SourcrLab.
5. Hands-on experience is scoped, not implied
A small set of tools carries a personal note from Pieter Henderyckx. Those are tools he genuinely uses or has used in real hiring work. Research, vendor material, practitioner reports and AI-assisted drafting never become first-hand experience by repetition; if there is no genuine use, SourcrLab does not imply that there was.
6. Commercial relationships never buy the verdict
SourcrLab is editorially independent. SourcrLab currently does not earn affiliate commissions when buyers choose a tool. A vendor can claim a profile and suggest factual corrections without paying, and that route stays free.
One paid service is available to vendors: the Founding Vendor programme, an annual membership described on the vendor page. It buys operational help. We structure the factual product information and sources a vendor supplies into the fields buyers compare on, and support them in keeping those facts current. It never buys inclusion, rankings, scores, positive editorial conclusions, Best-list placement, comparison outcomes, removal of criticism or silence, and it never changes the evidence level or the provenance a page carries.
Evidence can change our assessment. Money cannot. A vendor who shows us something material we got wrong or did not know can have the assessment reconsidered, and that route costs nothing.
Vendors own their facts. Recruiters own their experience. SourcrLab owns the analysis.
7. Accuracy and corrections
Information can come from public documentation, vendor-supplied corrections, practitioner reports and genuine hands-on notes, and it can change after we checked it. For a final purchase, verify material contract, pricing, security and compliance details with the vendor. Found something wrong? Report it, material corrections are applied to the page itself.
8. AI assists the work; it does not manufacture certainty
AI tooling is used extensively for research support, extraction, checking, drafting and maintaining the catalogue. In the current reviewed-brief workflow, editorial conclusions are reviewed before publication, and hands-on claims are never generated by AI. If the available evidence does not support a conclusion, the intended output is Unknown rather than a plausible-sounding guess.
Coverage is still uneven because the catalogue is larger than the set of fully researched tool briefs. Depth therefore follows evidence: a short, accurate page is better than a complete-looking page built on assumptions.
The reasoning behind this approach is documented in The Tech Noise Problem About the person behind it: About the editor.