textRay · the read · likes you over text

green flag explainer · warm reads need receipts too

how to tell if someone likes you over text. the green flags, cited.

good news deserves the same evidence standard as bad news. here are the signals that actually hold up, and the friendly noise that doesn't.

how can you tell if someone likes you over text?

the reliable signs are behavioural, not decorative: they match your energy, they start conversations too, their questions follow up on things you actually said, and their plans come with days and times attached. interest shows up as effort that repeats. one flirty tuesday proves nothing in either direction.

if the pattern holds for weeks, that's not you being delulu. that's a trend.

which green flags actually hold up?

the evidence criteria
  • matched energy. you send a paragraph, something paragraph-shaped comes back. effort, warmth, and initiative flow both ways, and neither of you is performing a one-person show.
  • they initiate too. check who spoke first each day for a fortnight. if the split is anywhere near even, the conversation has two engines. if it's 13 to 1, it has a passenger.
  • questions with a memory. not "wyd". things like "how did the second interview go" and "did your sister survive the hen do". follow-ups mean they filed your life somewhere and went back to check on it.
  • plans that get days and times. "we should go" becomes "thursday, 7, the place with the questionable jukebox". words match actions. the calendar agrees with the chat.
  • effort survives a boring week. anyone can sparkle during a good week. interest that persists through a week of nothing happening, admin, rain, a documentary about canals, is the real signal.
themhow was the dentist. did you play it cool or apologise to the hygienist for existing again
youapologised twice. once on the way in. she hadn't done anything yet

questions about the other person's day: them 7 · you 6 · past fortnight

what looks like interest but might not be?

the honest hedge, because wishful reading is still reading badly:

  • some people are just friendly texters. warm to everyone, emoji for all, replies at speed. friendliness at scale is a personality, not a signal. the tell is specificity: friendly texters are warm, interested ones remember.
  • one great night of texting means nothing. a single sparkling conversation is a mood, possibly a wine. patterns need repetition before they count, in the good direction too.
  • fast replies are not feelings. some people answer everything instantly, including their dentist and their broadband provider. speed without questions or plans is just a phone habit.
  • the receipts cut both ways. the point of checking the evidence is that it can come back "yes, this is mutual" or "this is friendly and flat". both answers are useful. only one of them was going to come from the group chat.

sometimes the boring explanation is true here as well. sometimes it's the nice one. verdict: pending. evidence: available.

how do you check it against the evidence?

gently, and without printing anything out:

  1. count initiations over two weeks. two engines or one?
  2. count questions that reference something you said earlier. specificity is the tell.
  3. list the plans mentioned. how many acquired a day and a time?

textRay reads both sides of the chat and reports what's there, including the good news. healthy conversations get a proper report too: matched energy, words matching actions, genuine check-ins, each finding cited to the messages with counterevidence one tap away. paste the chat, share screenshots, or add a WhatsApp export, and get the read either way.

healthy? we'll say that too

if it's mutual, the receipts will show it.

download on the App Store

iOS · 16+ · no subscription · credits from £2.99

The plain version.

textRay describes patterns in a conversation, positive ones included. It cannot confirm how a person feels, only what the messages show, and it never claims more than that. Consent is asked before any AI processing, both sides are always read, and reports stay until you delete them. An initial scan is 1 credit, a full scan is 3, no subscription.