Broad match vs exact match for shopping search terms
Start wide and tighten with evidence. Shopping listings are matched from your product data rather than from keywords you choose, so the practical control is the negative list plus how narrowly you fence each campaign. Fence everything from day one and you will never discover the phrasing customers actually use. Leave it open forever and you pay for queries that were never going to buy.
The rule for tightening is specific. Read the search term report weekly, group the terms into themes, and act only where a theme has taken meaningful spend with no orders. Those become negatives. Where a theme is producing orders, give it a home of its own so it can be bid on properly.
For the text search campaigns that sit alongside the same catalogue, the same logic applies: wide matching finds the language, tight matching protects the terms you already know are profitable.
Choose
Wide matching
when
- The campaign is new and nobody knows how customers describe the product
- You sell something people name in many different ways, such as parts or materials
- You have the time to review search terms every week and act on them
- A new product line has no query history at all
- Automated bidding needs volume before it can find patterns
Choose
Tight matching
when
- A small set of queries reliably produces most of the orders
- Spend is leaking to research phrases, job searches or the wrong product category
- Similar brand names or model numbers pull in traffic that never converts
- Budget is small enough that discovery spend would use all of it
- You need one campaign fenced so another can be tested cleanly
| Criterion | Wide matching | Tight matching |
|---|---|---|
| Cost | Some spend goes on discovery that will not convert | Spend concentrates on known terms, at a higher price per click |
| Setup effort | Low to start, but weekly search term review is required | Higher up front: negatives, fenced structure, and rules for what belongs where |
| Time to signal | Two to three weeks to gather enough terms to judge themes | Faster, since traffic is already qualified |
| Control | Limited, you steer after the fact with negatives | Direct control of which queries can trigger which products |
| Best for | Discovery, new catalogues, unusual product vocabulary | Mature accounts, tight budgets, protecting proven terms |
| What can go wrong | Nobody reads the report, and irrelevant themes quietly take the budget | The account stops finding new demand and volume slowly declines |
| How we measure it | Share of spend on terms with no orders, week by week | Cost per order and impression share on the fenced query set |
When both apply
The structure most accounts settle on runs both at once. One campaign is fenced to the queries you know convert and holds the higher bids, and a second stays open to everything else with the proven terms excluded from it.
That second campaign is the discovery budget, and it should be small and reviewed on a schedule. Each week, promote the terms that produced orders into the fenced campaign and add the dead themes as negatives. Over a few months the fenced side carries most of the revenue while the open side keeps finding language you had not thought of.
Questions
Related
Your plan is free.
A person reads it. You keep the plan. No logins, no call first.