Trusted by 800+ organisations worldwide
Obscura digitises the classic projective inkblot test, turning a four-minute candidate interaction into a predictive, objective, science-backed profile. Standardised, defensible, delivered in minutes.
Based on customer-reported outcomes. Individual results may vary. Full methodology available on request.
Organisations using Obscura include
The problem
The average unstructured interview has a predictive validity coefficient of around r = 0.20. That means it explains roughly 4% of variance in job performance. Most organisations conduct five or six of them before making a decision.
Each interviewer brings a different frame of reference, asks different questions, and reaches different conclusions. Two interviewers assessing the same candidate against the same criteria will agree less than 60% of the time.
Obscura reduces human variability at the screening stage. Every candidate sees the same standardised visual sequence. Every response is processed by the same model. No conversation to like or dislike. No CV to anchor on. No interviewer who reminds you of yourself. Every deviation is retained.
The full Obscura process, from candidate link to Insight Report.
Candidates receive a link via email or ATS. They open it on any device and view ten standardised images, responding briefly to each. The process takes approximately four minutes. No preparation is required. There are no right or wrong answers. Responses are evaluated against validated population norms for your target role family.
Completion rate: 97.4% when described as part of the process.
Obscura AI analyses each response across 34 signal dimensions simultaneously: response content, semantic categorisation, temporal latency, device orientation, scroll behaviour, session environment, and construct-specific indicators. The full list of signal dimensions is documented in the Technical Appendix.
34 signal dimensions. 8 scored constructs. 1 unified profile.
You receive a 47-page Candidate Insight Report within minutes of assessment completion. The report is always 47 pages. The length does not vary by candidate. It includes construct-level scores, percentile rankings, role-fit recommendation, and a structured interview guide.
Most hiring managers begin with the Executive Summary. The remainder of the report is available.
Obscura measures the signals that structured interviews cannot access, that CVs cannot contain, and that candidates cannot anticipate.
Construct names in this table reflect client-facing labels. Internal construct documentation uses extended nomenclature. Both designations are provided in the Technical Appendix. Where the client-facing label and the internal designation differ in scope, the internal designation applies. The BDQ, MFRI, and ROPS constructs are documented in Section 7.
The science
Projective assessment methodology has been studied in clinical and occupational settings since the 1920s. The challenge has always been standardisation: achieving consistent, scalable, quantitatively rigorous scoring across large candidate populations.
Obscura solves that problem. Our models are trained on 2.3 million validated response sets from 34 countries, scored against one-year performance ratings, manager satisfaction scores, and retention outcomes.
Where a candidate provides no interpretable response to a stimulus, the absence is scored as a response class rather than treated as missing data. Absence is information. The scoring of non-response is documented in the Technical Appendix.
Figures based on internal validation data, n = 2.3M. Competitor figures based on published meta-analytic estimates. A peer-reviewed paper is in preparation. The study pre-registration summary is available to Enterprise clients on request.
Traditional hiring introduces bias at every stage. Obscura removes human variability from the equation.
| Human Interview | Obscura | |
|---|---|---|
| Consistent stimulus across all candidates | ? | ? |
| Standardised scoring criteria | ? | ? |
| Free from interviewer bias | ? | ? |
| Independent of CV or academic credentials | ? | ? |
| Validated methodology | ? | ? |
| Consistent performance across demographic groups | ? | ?* |
| Candidate awareness of scored constructs | ? | ? |
*Performance data across demographic groups is documented in Technical Appendix Section 7. Consistency is assessed at construct level and does not account for downstream hiring decision variance attributable to client workflow configuration. Annual bias audit conducted by Perceptual Analytics Group.
Obscura does not use protected characteristics as model inputs. Every candidate sees the same ten images, presented in the same order, with the same instructions. Every response is processed by the same algorithm. The algorithm does not change between candidates.
"We replaced first-round screening interviews with Obscura in Q1. Time-to-hire dropped by 58% within three months. Hiring manager satisfaction scores went up. The data made the decision for us."Sarah K., Head of Talent Acquisition, UK financial services firm. 3,200 employees.
"One candidate saw a bird in every image. Another saw nothing in any of them. Our Obscura consultant said this was meaningful. I took their word for it. That hire is now our top-performing SDR."Marcus T., VP People, Series C SaaS company. 280 employees.
"Our legal team had questions about several of the constructs. Obscura provided the construct validity documentation. Our legal team seemed satisfied with the response. They have since moved to a different organisation. We have continued using Obscura."Name withheld, CHRO, professional services firm. 7,000 employees.
"We no longer need to speak to candidates. The Obscura profile tells us everything we need to know before they enter the building. We�ve eliminated the interview entirely."Elena R., Director of Operations, Global Logistics. 12,000 employees.
The average cost of a mis-hire is estimated at between 1x and 3x the role's annual salary, once recruitment costs, onboarding time, lost productivity, and severance are factored in. For senior roles, the figure is higher.
Obscura customers report an average reduction in mis-hire rate of 113% in the first year of deployment.
�1.4M average annual saving reported by customers with 500+ annual hires.
Projected annual saving
Legal challenges from candidates
Based on 113% mis-hire reduction and industry-standard replacement cost estimates. As stipulated in Technical Appendix Section 7, Obscura accepts no liability for legal challenges arising from rejected candidates. Projected saving does not account for litigation costs.
Obscura connects natively to your ATS and HRIS in minutes. No IT project required.
Candidate response data is transferred to your ATS on assessment completion. Obscura retains a copy for 36 months, or for the duration of your data processing agreement where this period is shorter. Integration count includes generally available, beta, and pipeline connectors.
Recognised by
Start with the core construct suite and scale as your hiring grows.
per month, up to 100 assessments
per month, up to 500 assessments
unlimited assessments
*We recommend reviewing Section 7 before deploying Obscura at scale. We have found that most Enterprise clients review it after deploying Obscura at scale.
And the answers we're able to give.
Join 800+ organisations using Obscura to make faster, more consistent, more defensible hiring decisions. No credit card required for your first 10 assessments. Candidate data is retained for 90 days following trial expiry.