Blank and zero are not the same thing
Only empty or whitespace-only cells are treated as missing. An existing 0 stays untouched. If your data uses zero as a missing-value code, clean that convention before using this tool.
Free web tool
Fill missing CSV values using a constant, mean, median, or forward fill.
A column has a few blanks. Sometimes zero is the right fill; sometimes the mean, median, or previous value makes more sense. This tool fills only blank cells in one selected column, using one of those four rules.
Only empty or whitespace-only cells are treated as missing. An existing 0 stays untouched. If your data uses zero as a missing-value code, clean that convention before using this tool.
Mean and median remove blank cells first, then use only values that actually parse as numbers. That keeps “missing” separate from a real numeric zero, and values such as N/A stay out of the calculation.
Mean is easy to interpret when the distribution is fairly regular. Median is often more stable when a few extreme values are present. Neither one “restores the original truth”—both are imputation choices.
Forward fill copies the most recent non-blank value above the missing cell. That is useful for ordered or time-based data. In an arbitrary row order, it can simply copy whichever value happened to come first.
Leading blanks remain blank until the tool has seen a previous non-blank value. Those untouched cells are not counted as filled, so the filled-cell metric tells you whether the rule could actually populate every missing position.
Imputation creates new data. For analysis copies that is often exactly what you want, but keep the original and record which column and rule you used so the transformation stays auditable.
Mean- or median-filled values may never have existed in the source data. If downstream users need to know which values were imputed, add that provenance in your workflow; this tool does not add a separate imputation flag column.