Wordle strategy is mostly one skill: reading the colours as information instead of as a verdict on your guess. A guess does three jobs. It tests letters you have not tried, it moves the yellow letters to new positions, and it confirms the greens. Early guesses should do the first job hardest, which is why the opener matters and why this site ranked openers by simulation rather than by taste. The rest of this page is two demonstration games, played against answers fixed in advance so every colour can be checked, and the one rule that decides more games than any other: how duplicates are counted.
One thing this page does not do is rank opening words. That ranking was computed by simulation and lives on Best Wordle Starting Words; this page starts where the opener has already been played.
Worked game one: the answer is FORGE
| Guess | Colours | Answers left |
|---|---|---|
| RATES | RATES | 156 of 8,636 |
| PRONE | PRONE | 10 |
| FORGE | FORGE | Solved in 3 |
RATES draws two yellows, R and E, and three greys. Read it precisely: R is in the answer but not in position 1, E is in the answer but not in position 4, and A, T and S are out entirely. That reading leaves 156 of the 8,636 five-letter words in this site's list. The second guess, PRONE, is chosen to work. It moves R to position 2 and E to position 5, and spends its other three letters on fresh tests, P, O and N. It draws yellow, yellow, green: R is still homeless (not position 2 either), O is present but not in position 3, and E sits down in position 5. Ten words survive: FORCE, FORGE, FORME, GORGE, HORDE, MOIRE, OCHRE, OMBRE, OORIE and OURIE.
Now the endgame, and it is worth being honest about it. Those ten are near-twins, and no reasoning separates FORGE from FORCE before you play one of them. What you have is frequency: FORCE is a far commoner word than FORGE, so a player ordering by commonness tries FORCE first and solves in four, or FORGE first and solves in three, as this game does. Strategy got the game to a one-letter bet in two guesses. The bet itself is luck, and a good player just makes sure the bet is cheap. That is also the answer to the eliminator question in the FAQs below: with four guesses in hand and ten near-twins standing, there are lines that test several first letters at once. With two guesses in hand, there are not.
The duplicate-letter rule
Here is the rule, once, in full. The colours are awarded in two passes. Greens are marked first, using up those copies of each letter. Then the remaining copies are handed out as yellows, left to right, until they run out. Any guess letter left over is grey. So a grey letter beside a coloured copy of the same letter means the answer holds exactly the coloured number of copies and no more. Grey never means "none" when a green or yellow of the same letter exists. It means "no more than you have already found".
Three shapes cover nearly every trap. Green plus grey, as with the Es in SPEED against an answer ending in E: one copy, already placed. Yellow plus grey, the same letter tested twice before it is placed: one copy, position still unknown. Yellow twice, or green plus yellow: at least two copies, and words with only one are dead, however well they fit everything else. The second game is built on that third shape.
Worked game two: the answer is ABBEY
| Guess | Colours | Answers left |
|---|---|---|
| SLATE | SLATE | 277 of 8,636 |
| BABES | BABES | 1 |
| ABBEY | ABBEY | Solved in 3 |
SLATE draws two yellows: A, not in position 3, and E, not in position 5. Two hundred and seventy-seven words fit. BABES is played to test B, and it returns the most instructive pattern in the game: B yellow, A yellow, B green, E green, S grey. Count the Bs the way the rule says. One is green in position 3. The other is yellow, which means a second B exists somewhere else. The answer holds at least two Bs. That single count kills every one-B word that fits the rest, AMBER among them, a reasonable-looking word that matches every green and yellow on the board and is wrong only on the count.
The rest of the pattern closes the box. A has now been yellow in position 3 and in position 2, and the only position left for it is position 1. E is green in position 4. The shape is fixed: A, then two Bs in positions 2 and 3 with one of them already placed, then E, then one unknown letter that is not S. Exactly one word in the list fits all of it: ABBEY. Notice, too, what BABES wasted. Its last letter retested S, already grey from SLATE, and taught nothing. The guess still won the game on the strength of the double B, but on a tighter day that spent letter is the difference. A follow-up guess should be audited the same way before it is played: every letter either new, moved, or confirming.
The strategy, in one paragraph
Open with a strong, five-distinct-letter word and keep it. Spend the middle guesses on new common letters while honouring what the colours proved, and read every repeated letter as a count, never as a presence. When the field narrows to near-twins, order them by how common they are, and accept that the last step is a bet. If you want the field counted for you at any point, the Wordle Helper applies exactly the rules in this guide to your own clues.
Questions
What is the most common Wordle mistake?
Reading a grey letter as absent when the same letter is green or yellow somewhere else. Grey caps the count of that letter; it does not erase the copies the colours have already proved. The second most common mistake is quieter: spending a follow-up guess on letters that earlier greys have already ruled out, which tests nothing and hands the turn back.
Should I ever play a word that cannot be the answer?
In standard mode, sometimes. When three or four candidates remain and they differ in one letter each, one guess can only test one of them. A word that cannot be the answer but tests two or three of the differing letters at once can settle the family in a single turn. It feels wasteful and is often the fastest line. In hard mode it is not allowed, because every guess must use the clues already revealed.
Does hard mode need a different strategy?
The opener is the same, and so is the counting. What changes is the follow-up: hard mode forces every guess to honour the greens and yellows already shown, so eliminator guesses are illegal and one-letter-apart families must be played through in order, most frequent first. Expect hard mode endgames to take a turn longer. That is the price of the badge.
How does the Wordle Helper on this site apply all this?
It filters the same way this page teaches: green positions fixed, yellows with their excluded positions, and the counting rule for duplicates, so a grey beside a green caps the letter instead of deleting it. It then ranks what is left by letter position. What it will not do is suggest an eliminator word, because it only lists words that can still be the answer.
Sources
- ENABLE word list (public domain), the 8,636 five-letter words both demonstration games were filtered against: github.com/brokensandals/wordlists
- Word frequency counts from published English text (Google Books n-gram data), collected by Peter Norvig, for the commonness order used in the endgame discussion: norvig.com/ngrams/count_1w.txt
- Every colour pattern and candidate count in the two games was computed from the list above with the game's colouring rules on 28 October 2026, and can be reproduced in the Wordle Helper. The full checking method is on How We Verify.
Update log
28 October 2026: published, with both demonstration games computed from the site's word list on the same day.