> ## Documentation Index > Fetch the complete documentation index at: https://docs.getvector.me/llms.txt > Use this file to discover all available pages before exploring further. # Duplicates and merging > Find features and candidates that describe the same capability, and merge them so each capability is counted once. The same capability reaches Vector from many directions — a competitor's docs, another competitor's changelog, a customer's words in a call — and each source names it differently. Left alone, one real capability ends up as two or three entries, and every count built on them is off: how many competitors have it, how much you know about it, where your gaps are. Merging duplicates keeps each capability counted once. ## Find duplicates Open **Features** → **Possible duplicates** in the left menu. Vector compares names and meaning: a pair can match because the names are the same — marked **same name** — or because they describe the same thing in different words, shown as a similarity scale of dots. Nothing is merged automatically; every match is a suggestion for you to review. Pick the tab for the job you came to do: * **Candidates** — raw findings that aren't features yet, compared with each other and with your features. Clearing these keeps the [candidates](/features/candidates) inbox short. * **Features** — established features, compared with each other and with candidates. Clearing these keeps the [catalog](/features/features-in-domain) clean. Each card shows one entry and the entries that look like it, with their cluster and where they came from. To check a single feature, open its [dossier](/features/feature-dossier) and go to **🧩 Possible duplicates**: it lists the matches for that feature alone. ## Merge a duplicate Every button on the page does the same thing: **Merge** moves the content to the record you keep, and the other one closes with a link to it. What that means depends on the pair: | Pair | What Merge does | | ----------------------- | --------------------------------------------------------------------------------------------------------------- | | Candidate and candidate | The candidate in the row is merged into the one the card is about: its card closes, and its mentions come along | | Candidate and feature | The candidate is attached to the feature at once: its mentions become evidence on it | | Feature and feature | Opens the merge dialog, where you decide what survives | ## Merge two features Merging features can't be undone, so the dialog asks you to decide what survives: The feature you keep takes in the other. You can search for more features to merge into it at the same time. Under **Content to merge**, tick what moves across: **Captured insights**, **Attachments**, **Linked delivery work items** and **Connections**. Pick whose description to keep, and whose field values. Where the two features disagree about a competitor's coverage, pick which to keep. Click **Merge into** the feature you keep. The merged-away feature isn't deleted. It's marked as merged and points to the one you kept, so old links to it still open the right page. You can also start a merge from any feature's page, with **Merge…** next to its name. ## Keep two entries apart If two entries only look alike — similar names for different things — click **Not a duplicate**. The pair won't be suggested again. Changed your mind? Click **Undo** in the confirmation that appears. ## Where to go next Where most duplicates start. What your customers say about your features.