A search that mentions your product is not the same as a search from someone who wants to buy it. The single biggest difference between an account that prints money and one that drains it is usually not budget or bidding. It is which slice of the query universe the account decided to pay for. Most accounts pay for far too wide a slice, because the keyword tools reward volume and the interface makes it easy to say yes.
Every search that touches your market falls into roughly one of four buckets.
Ready to transact. The person has decided what they want and is now choosing who to buy it from. Service plus city, brand plus buy, emergency phrasing, price and quote requests, near me searches for local services. This is where your money should go and where it should stay when budgets get tight.
Comparing options. Best, versus, alternatives, reviews, top ten. These people will buy, but not today, and they are shopping you against three competitors. Worth paying for if your offer is genuinely strong and your page is built to win a comparison. Not worth paying for if your landing page is a generic services page.
Researching or learning. How to, what is, why does, DIY. Some of these convert eventually. Most do not, and in service businesses a large share of them are people who intend to solve the problem themselves. This traffic belongs to SEO and content, not to paid search.
Wrong person entirely. Job seekers, students, suppliers, competitors, people looking for free versions, and people in the wrong country. Pure waste, and it is a maintenance job to keep it out.
Across the accounts I have worked on, these patterns waste money with impressive consistency regardless of industry.
Do this monthly. It takes about an hour once the habit is set.
This work matters more now, not less. Broad match plus Smart Bidding was already loosening the connection between the keyword you typed and the query you paid for. AI Max, which sits as a setting on top of a standard Search campaign and expands both query matching and creative, pushes that further. Google is increasingly deciding what your keyword means.
That is not automatically bad. In accounts with clean conversion data and real volume, the expansion often finds queries a human would never have written down. In accounts with weak conversion signals, it finds expensive nonsense faster than you can read the report. The deciding factor is whether the system knows what a good outcome looks like. If your conversion tracking counts form loads or newsletter signups, do not turn on aggressive matching. You are handing a bigger budget to a system with a broken definition of success.
Research traffic is worth buying when you sell something with a long, considered cycle and you have a remarketing path that actually works. Medical, education, and high-ticket B2B all have cases where an early-stage query is the start of a six-month relationship. But you have to be honest that you are buying an audience, not a sale, and you need the tracking to prove the second step happens.
Competitor terms are worth buying when you have a real, provable advantage and a page that states it plainly. They are not worth buying to make a point.
Pull the search terms report, sort by cost, and read the top hundred lines with a pen in your hand. Tag each one buy, compare, research, or wrong. Whatever share of spend lands outside the buy and compare buckets is the size of the opportunity sitting in your account right now, before you spend another cent on anything else.