Five systems.
One yardstick.
We put the same three open roles through four competing sourcing tools and through Atlas, and asked one question: how much of what the others find do you find too?
Full circle = everything the other four systems found.
What we tested.
Three real roles that are hard to fill, in three markets. Every system that takes free text got the word-for-word identical prompt. For LinkedIn Sales Navigator the same criteria were translated into filters.
Senior Backend Engineer
Senior Product Manager
Process Integration Engineer
The same free-text prompt went into every system that accepts free text. LinkedIn Sales Navigator has no free-text search, so the same criteria were translated into its filters.
For each role, the reference set is the union of everything the other systems returned, deduplicated. A system is never measured against its own output.
Hits were matched on first and last name, accent-free, so that spelling variants of the same person count as one.
How much of the others' finds does each system see?
Every system covers its own hits completely, so those are taken out of the count. What is left is the fair question: of the people the others found, how many did you find too?
AtlasSprad
Juicebox
Metaview
LinkedInColumns A, B and C are the three roles above. Read one row like this: of the people the other systems found for role A, Atlas also found 30 percent. A dash means the system returned no usable result for that role. Atlas covers 38 percent overall, the best of the other four covers 9 percent.
The majority of all profiles is reached by no system in this field.
That is the honest reading of this chart. Measured against everything the other tools together put on the table, even the strongest result leaves most people unfound. The gap between the systems is real, but so is the gap to complete coverage.
Every tool shows you a different slice of the market.
The four competing systems barely overlap with each other. On role A, 76 people were found in total. 72 of them were found by exactly one single system.
Atlas and JuiceboxRole C, EindhovenAtlas and MetaviewRole B, LondonAtlas and LinkedInRole A, BerlinMetaview and JuiceboxHighest value between two other systemsTwo of the other systems never share more than a fifth of their hits. Atlas reaches 40, 47 and 52 percent against the individual systems. It sits in the middle of the field while the others fish at the edges.
One employer, or the ecosystem around it.
LinkedIn found 18 of its 19 hits at a single employer, ASML. Atlas and Juicebox return the supplier ecosystem around it. Both answers are defensible. They are simply different slices of the same labour market, which is why the two systems share only 7 percent of their hits on this role.
How many strong candidates does one run put on the table?
Coverage measures overlap. It says nothing about volume. So for role A we counted something else: how many people each system returned that actually meet the must-have criteria.
AtlasDeep research mode · extrapolated from the 120-profile sampleMetaviewCounted exactlyJuiceboxCounted exactlyLinkedInCounted exactlyThis answers a different question than the chart above. Coverage asks how much two systems share. This asks how many suitable people one run produces. Both numbers come from the same study, and they do not replace each other.
The Atlas run behind this number was carried out in deep research mode with 24 queries. The four competitor counts are exact: every hit those systems returned was reviewed. The Atlas figure of roughly 147 is an extrapolation from a sample of 120 profiles out of 1,608, checked rule-based against the same criteria rather than judged by a model. Treat it as an order of magnitude, not a precise count.
And the rest of the run.
More volume also means more noise. This is how the 120 checked profiles were distributed.
60 of the 120 sampled profiles fail on one thing only: the search result does not state their skills. They are not necessarily unsuitable, they are unproven at that point in the process. We count them as weak anyway, because a claim we cannot back up is not a shortlist.
What this study can and cannot say.
One benchmark, one day, three roles. Here is what that does not cover.
The reference set is what the other four systems found, not everyone who exists. A person no system found does not appear anywhere in these numbers.
It comes from 120 rule-checked profiles out of 1,608. The four competitor counts next to it are exact.
Measured on 15 August 2026. Search indexes and ranking change, so a rerun will not reproduce the values to the percentage point.
Backend engineering in Berlin, payments product in London, semiconductor process engineering in Eindhoven. Other role families may behave differently.
LinkedIn Sales Navigator has no free-text search. Its criteria were translated into filters, which is a faithful translation but not the identical input.
All 218 reviewed profiles are documented with source and rating per person, deduplicated, alongside the per-system hit lists for each role.
See it on your own hardest role.
A benchmark is three roles on one day. Your open position is the only test that counts.