Experiments (under the Campaigns menu) let you test a change on a slice of traffic instead of betting the whole campaign: a copy variant, a bidding change, a landing page. With Google splitting traffic and reporting the comparison.
At small-business scale, the honest caveat is data volume: an experiment needs enough conversions per arm to mean anything, and a $30/day campaign may take months to get there. For most owners, bold sequential changes judged over full weeks beat formal experiments.
Where experiments genuinely earn their keep even at modest scale: risky changes you'd hate to apply blind. Like trying a smart-bidding strategy on 30% of traffic while manual bidding keeps the lights on.