Two surveys can concern the same issue while measuring different things. Rating the current economy, describing personal finances, and predicting conditions next year are three distinct questions.
Read the response choices
A percentage is meaningful only in relation to the question and the choices offered. Selected response categories should not be mistaken for a complete distribution. Multiple-choice concerns may overlap and need not add to 100 percent.
Compare like with like
For a time series, check whether the wording, population, mode, and geography remain comparable. A methodological change may need an explicit break in the chart. Keeping a visible gap is more informative than suggesting continuity the source does not support.