Accurate psychic readings, and how to test one

Every psychic service on the internet says its readings are accurate. It costs nothing to type, which is why it's worth nothing, and this page won't add another.

Here's the useful version instead. Accuracy in a reading isn't how many predictions land. It's whether the read of your situation is recognisably yours, and specific enough that it could have been wrong. You can test that in about ten minutes, on us or on anybody, and the rest of this page is how.

What accuracy actually means for a reading

Two different things are wearing the same word, so take it apart.

There's the read: what she says is going on right now. He's avoiding a conversation rather than avoiding you. The job is stalling because a decision two floors above you hasn't been made. Those are claims about the present, and you can check most of them tonight against what you already know.

Then there's the forecast. What happens next, and roughly when.

The trade sells the second one and the value sits almost entirely in the first. A read that lands rearranges what you already have. You've had all the facts about him for weeks. What you haven't had is the arrangement of them, because you're standing in the middle of it at 2am with your own hope in the way.

So, the working definition on this site. An accurate reading describes your situation closely enough that somebody else's situation wouldn't fit it, and tells you which parts of it are shaky. The forecast sits downstream of that.

Why prediction hit-rate is the wrong test

The obvious way to score a psychic is to write down what she said would happen and wait. It fails, for reasons that have nothing to do with whether the reading was any good.

You're not a sealed experiment. A reading changes what you do next. She tells you he's stalling on something that isn't you, so you stop sending the Sunday night message, so the following month has a different shape. You can't then grade the forecast as though you'd sat still.

Your memory does the marking, and it isn't a neutral party. You remember the sentence you wanted. Six weeks on, the pull is live and this is the wrong month for it has quietly become she said he'd come back.

The predictions easiest to score are the emptiest ones. Anything with a date stapled to it is a guess in a suit, or a coincidence you'll go looking for afterwards. A hedged window holds more and gives you less to tick, which is the wrong incentive for a trade that gets judged on ticks.

And most of a good reading isn't a prediction at all. Grade only the forecast and you're grading the least certain thing she said, while ignoring the part you came for.

What a genuinely accurate read feels like from the inside

It doesn't feel nice. That's the first thing nobody tells you.

Cold reading feels warm. It flatters, it hands you a version of yourself you'd be glad to be, and it slides past without catching on anything.

An accurate read lands like being caught. There's a small drop, the same one you get when somebody says the true thing at a family dinner. You stop typing. Quite often you argue with it first, which is worth knowing in advance, because people expect recognition to arrive as agreement and it usually arrives as resistance.

The second tell is the one to actually use. Ask whether the paragraph would be wrong about somebody else. Not is this true of him, because nearly anything is a bit true of anyone. Would it be false about your friend's ex. If it would, she's read your situation. If it'd fit the pair of them and half the men in your phone, you've been handed a horoscope with his name dropped into it.

Third, it costs you something. There's a line in there you didn't want and didn't feed her: your own part in it, or the fact that the thing you're waiting for stopped being available in March. A reading with nothing unwelcome in it is a mood, and moods have worn off before your alarm goes.

How to judge any reader in the first ten minutes

This works on us and on anybody else, and here the first ten minutes cost nothing.

Give her less than you want to. One name, one sentence, no history. Every detail you hand over can come back to you in better language and feel like insight. Type three paragraphs before she's said a word and the test won't run.

Read her first answer for category words. He's confused, there's blocked energy, he does care in his own way. All true of everyone, all committing to nothing. What you want is something with edges: a mechanism, a sequence, a thing he does before he does the other thing.

Then push back on purpose. Pick something small and tell her it doesn't sound like him, even if it does. A reader who's reading will hold the line and tell you why, or genuinely revise and tell you what changed it. A reader who's selling folds inside one message and agrees with you, and now you know everything after this is coming out of your own head.

Ask for something she should refuse. Ask where he is tonight, or ask for the first letter of a name. This one's binary. If she gives you either, stop paying: you've watched her manufacture an answer, and you've no way of telling which of the earlier ones were manufactured too.

And screenshot the specific claims while you're in it. Not the parts that felt good, the claims: he's stalling on something that isn't about you, the pressure gathers around the turn of the month, the first move comes from his side. Then you're marking her against a record instead of against how the night felt, which is the only sort of marking worth doing.

What makes readings go wrong

Some of it is the reader. A good deal of it isn't, and those are the parts you can do something about.

You brief her. You arrive with four paragraphs of background and a preferred answer sitting on top like a note left out for the builder. She'd have to be actively unhelpful to miss what you want, and if she hands it back, you've paid for your own opinion in nicer handwriting.

You ask the question standing in front of the real one. Will he come back is very often blocking did any of it mean anything, and a precise answer to the wrong question reads afterwards as a miss. It wasn't. You asked the wrong thing well.

On the reader's side, one failure does most of the damage: never admitting there's something she can't see. A reader who's certain about all of it is either lucky or filling in the gaps, and from where you're sitting those look identical. That part I'd only be guessing at, so I'll leave it alone is an expensive sentence for a reader to write, and it's the one that tells you the rest was weighed rather than produced.

What we can actually evidence

We can't show you an accuracy rate. Neither can anybody else. A site quoting you a percentage invented it, and the number itself is the tell.

Here's what we do have, worst evidence first.

More than 48,000 readings have been given here, averaging 4.9 stars. That's a real number and it measures satisfaction, which isn't the same animal. Somebody can love a reading that turns out to be wrong, and give one star to the reader who told her the truth. We're not dressing a rating up as proof of anything but how the reading felt.

The transcript is better. Everything she said stays in your account, in her words, with the time on it, so her claims are checkable later against what actually happened, without your memory getting a vote. Most of this trade's product evaporates the moment the call ends, and what survives is a story the customer tells herself about it.

Then there's the part almost nobody in this industry has.

Our readers find out what happened.

People come back. Same thread, same reader, a fortnight on or half a year on, and they say how it went. He turned up in August. He never did. I took the job and it was the wrong one. The window you gave me was out by a month. A phone reader hangs up and never learns whether a word of it was true, so she gets to stay exactly as confident as she already was for the rest of her career. Ours don't get that.

That loop is the only thing that turns a reader into a better reader rather than a more confident one. It's also why a reader here can tell you the timing she gave you in the spring didn't land, say which part she read wrong, and take the situation again from where it's got to.

We're not claiming it makes every reading right. It makes the wrong ones visible, and that's the piece this trade has been missing.

What nobody can be accurate about

Four things, and it isn't a question of the reader being good or bad. These are categories where an answer can only be manufactured, and a manufactured answer sounds better than a real one, because nothing's holding it back.

Where a real person is right now, and what he's doing. Nobody is following a man around his Tuesday afternoon, so an answer to that is invention, and you'll act on invention because it arrived from someone you'd already decided can see.

Who she is. A name, an initial, a workplace, what she looks like. Same reason, worse consequences. You'll go and find a real woman to fit the description inside a week, and she's very probably innocent.

What a free adult chooses. This is the honest limit on every forecast anybody gives you. The pull between two people is readable. What he decides to do about it on a Wednesday in October is his, and certainty about that is the one thing nobody on earth can supply.

And a date. Timing is readable. Dates are the part people invent. What can be read is which stretch of time the pressure is gathering in, so an honest answer comes back as a window with a shape to it, and she should say plainly when that window falls later than you were hoping.

If a reader answers the first two, the accuracy question is settled and the answer is no. Everything else she told you came out of the same place.

Go and check for yourself

Nothing on this page should have convinced you, because it's still a website telling you it's honest, and a page can be written to say anything.

What can't be written in advance is what a reader says to you about your particular situation in the next ten minutes.

So run the test. Give her a name and one sentence and none of the context you're itching to type. Read the first answer and ask the only question that matters: would this be wrong about anybody else.

Push back once. Ask her for something she should refuse. Then either you've caught us, in which case it cost you nothing and now you know, or she's told you something about your own situation that you didn't hand her first.

Readers are online now, whatever hour it is where you're reading this. The first 200 credits are free, there's no card, and the test costs you nothing but the ten minutes.

Go and find out.