Zero-evidence identification: reading a ghost when the journal gives you nothing
On Insane difficulty and in many challenge modes, the game strips away your evidences. What remains is behavior. This guide builds a repeatable method around it, so one suggestive moment never becomes a confident wrong answer.
Why zero-evidence runs punish knowledge and reward discipline
No-evidence contracts are where most misidentifications happen, and the failure pattern is remarkably consistent. A player sees one suggestive behavior early, anchors on a single ghost, and then unconsciously stops testing. Every ambiguous event afterward gets interpreted as support for the theory they already hold. Ten minutes later the journal gets locked in on a Revenant that was actually a Moroi that happened to be chasing someone with line of sight.
The fix is not more knowledge. Plenty of players who can recite every ghost's traits still misidentify constantly. The fix is discipline about how much weight each observation actually deserves, and a habit of naming the test you would need to run before you allow yourself to commit.
The core distinction: strong tells versus weak tells
Every behavioral observation falls into one of two buckets, and keeping them separate is the entire method.
Strong tells
A strong tell is a behavior that only one ghost can produce. When you witness one clearly, you are allowed to commit. Examples the community relies on:
- A six-fingered handprint under UV points uniquely at the Obake.
- A hunt that ends the instant you step into an official hiding spot, while the ghost can see you, is the Aswang's signature weakness.
- A ghost that walks straight past you during a hunt while you stand perfectly still and silent, with nothing between you, behaves like the blind, hearing-based Kormos.
- A ghost model that visibly changes appearance mid-event or mid-hunt suggests the Obake's shapeshift or a Mimic copying, which then demands a follow-up test to separate the two.
Weak tells
A weak tell is a behavior many ghosts can produce, or one that depends on conditions you cannot fully see. Speed impressions are the classic trap. A ghost that "feels fast" might be a genuinely fast entity, or a normal-speed ghost with line-of-sight acceleration, or a conditional speedster whose condition happens to be met. Activity level, interaction frequency, early hunts, and single odd events all live in this bucket. Weak tells should narrow your list. They should never close it.
A practical weighting rule: one weak tell moves a ghost up your list. Two independent weak tells make it your working theory. Only a strong tell, or a deliberate test you designed and repeated, should make it your answer. If you cannot name the test you ran, you guessed.
Designing deliberate tests instead of waiting for luck
The difference between a veteran and a guesser on a no-evidence run is that the veteran manufactures information. Instead of wandering and hoping the ghost reveals itself, you set up situations where different ghosts must behave differently, then watch which branch happens.
Test 1: The sanity check-in
Track average sanity deliberately, not vaguely. Most ghosts begin hunting around the fifty percent average sanity mark. A hunt that arrives while your crew is still comfortably high is itself information: it narrows toward early aggressors like the Demon, condition-based hunters such as a Mare in a dark room or a Yokai reacting to voices, or ability-driven hunts. A ghost that refuses to hunt even at rock-bottom sanity, in a game where you are doing everything wrong, points toward a Shade with people nearby or an unusually passive draw.
Test 2: The speed lane
Set up a long, safe sight line during a hunt, ideally near a hiding spot or loop you trust, and listen to footstep cadence with and without line of sight. You are separating three profiles: consistently slow, consistently fast, and accelerating when it sees you. One pass is an impression. Two or three consistent passes across separate hunts is a real observation you can weight.
Test 3: Electronics and voice bait
Some ghosts key off what you carry and what you say. Standing in the ghost's territory with active electronics, or speaking deliberately near it, changes hunt ranges and aggression for specific entities. If aggression clearly spikes when the crew talks and fades when everyone goes quiet, that is a repeatable, testable pattern and worth far more than a speed impression.
Test 4: The incense timer
When you smudge a ghost during a hunt, different entities stay disoriented for meaningfully different lengths of time, and the interval before the next hunt can differ too. Counting seconds out loud feels silly and works. A dramatically shortened calm period after smudging is one of the community's favorite separators for the Demon.
Running the elimination out loud
On a team, the biggest upgrade is narrating weights instead of conclusions. "It felt fast, so it's a Revenant" invites anchoring. "Fast with line of sight, normal without: that keeps Jinn and Moroi alive and argues against Revenant, because Revenant should stay fast while chasing" is an elimination the whole crew can check. You will be wrong sometimes. Being wrong out loud is how the crew catches it before the journal gets locked.
Solo, the same principle applies in miniature: before committing, say what you observed, what it rules out, and what test confirmed it. If the sentence ends at "it just felt like one," go run one more test.
How the Smart Scanner supports this method
The scanner on this site is built around exactly this weighting idea. Evidence toggles do the hard filtering, behavior observations act as filters rather than answers, and the reasoning notes state why each candidate survived or fell. On zero-evidence runs, work from the observations panel alone and read the notes critically: the tool is deliberately conservative, and if it keeps a ghost alive that you have mentally eliminated, that disagreement usually means you are leaning on one shaky observation instead of converging evidence. Treat the shortlist as a checklist of tests still worth running, not as an answer machine.
A worked example
Say you are on a small map, no evidence available. Early on, the breaker keeps tripping and rooms go cold fast, a weak tell that nudges toward electricity-hungry or cold-associated entities. First hunt arrives with average sanity still clearly high: a second, independent weak tell toward early hunters. Working theory forms, but it is not an answer yet. During the hunt you stand still and silent in the open, and the ghost pushes straight toward you anyway: it is not hearing-based, and your "stand still" test just eliminated a branch. You smudge it, count the calm, and the next hunt arrives strikingly early. That short cycle after incense is your strong separator. Now the pieces converge from three independent directions, and locking in is a decision rather than a hope.
Every figure referenced here, from sanity thresholds to smudge timings, is a community estimate and can shift with any patch. Phasmophobia's 2026 rework season makes re-testing especially important. Confirm behavior in-game, and see our corrections log for what changed and when.