How Klivka Finds the Person You Meant

You remember José, but you type jose. On another day you search for Jonathan and swap two letters. Sometimes all you remember is js for John Smith. A personal relationship notebook should be forgiving about those small gaps in memory.

Klivka searches each user’s own collection of people. Its search handles several common ways a typed name can differ from a saved one. It removes differences in capital letters, extra spaces, and accents before comparing anything. That is why jose can find José.

Put the strongest clue first

From there, the search follows a fixed order. It looks for the name exactly as typed, then moves through clues that carry less certainty.

People search ranking flow. Amber: direct matches. Gray: spelling fallback. Green: ranking and result selection.

The first five checks use letters that appear directly in the saved name. Those direct matches stay ahead of spelling guesses.

If none of those checks finds a convincing answer, Klivka allows for a misspelling. It measures how similar the typed name is to each saved name. Short searches are treated more cautiously: one letter is too little for a useful guess, and two or three letters must be very close to the saved name. Longer input gives the search more evidence, so it can forgive a little more.

The order matters because a fuzzy guess can look plausible. Someone who types john should see John before Johnny, Johnathan, or a more distant name that happens to share several letters. Klivka gives every direct kind of match its own place in line, with misspellings behind all of them.

Let sorting settle the ties

Klivka lets users sort people by name, by when they were added, or by when they were last updated. During a search, relevance takes priority over that dashboard preference. The chosen sort order settles ties between equally good matches, so a recently updated partial match cannot move above an exact name.

Klivka shows at most 50 results for a search. That limit is applied after every name has been considered. If Klivka selected 50 records before ranking them, an exact match could be omitted while weaker matches remained. It compares the full collection first and keeps the 50 strongest results.

When the collection grows

There is a cost to that choice. Each search reads and compares every name in the current person’s collection. Klivka is meant for a small, intentional list of relationships, where that work stays modest. A directory with thousands or millions of names would need an index to narrow the candidates before ranking them. Until the collection grows that large, the search rules remain close to the human problem they solve: remembering enough of a name to find the right person again.