HOW TO ADD KEYWORDS TO A GOOGLE SHOPPING CAMPAIGN

The question of how to add keywords to a Google Shopping campaign comes up in almost every onboarding call I have with an e-commerce advertiser, and the honest first answer disappoints them: there's no keyword box to type into. Shopping doesn't take positive keywords, match types or keyword-level bids. But that's not the end of it, because there are real keyword-shaped inputs you can add, they just aren't called keywords and they don't behave like a Search keyword list. This article is the version I wish someone had given me early: what you actually add, and when.

I'll be concrete, because vague answers here waste money. The inputs that do keyword-adjacent work are search themes in Performance Max, audience signals that steer who you reach, and negative keywords that exclude searches you don't want. Each sits in a different place in the campaign and each has a right moment to add it. Getting the timing wrong, adding everything on day one before you have data, is the mistake I see most, and it usually just handcuffs the algorithm before it's learned anything.

So this is less a click-path tutorial and more a decision guide drawn from running these campaigns on live accounts. If you came looking for the keyword field, you'll leave understanding why it's absent and, more usefully, which levers actually move your query eligibility and the exact point at which I add each one.

How to add keywords to a Google Shopping campaign: what you actually add

Let's set the frame precisely. Google states in its own product documentation that Shopping ads use your Merchant Center product data, not keywords, to decide how and where to show: Google's About Shopping ads documentation. That single fact reframes the whole task: you're not adding keywords, you're adding inputs that sit around Google's product-to-query matching. The feed is the keyword layer, and everything below is a lever on top of it.

There are three real levers. First, negative keywords in Standard Shopping, the only place a literal keyword field exists, and it's subtractive: you add terms to exclude, never to target. Second, search themes in Performance Max, phrases you feed the algorithm to help it find converting searches faster. Third, audience signals, which tell Google which people are likely buyers rather than which searches to chase. None of these is a positive keyword in the Search sense, but together they're the closest thing you have, and knowing which one to reach for is the actual skill. If the absence still feels strange, the structural reason sits in why Google Shopping keywords don't exist as a targeting input at all.

The reason all three matter is that they do different jobs. Negatives clean the pool you've already matched into. Search themes accelerate discovery of pockets you know convert. Audience signals steer toward the right people. Confusing them, treating a search theme like an exact-match keyword, or expecting audience signals to exclude traffic, is where most of the wasted effort goes. So before you add anything, be clear on which of these three problems you're actually solving.

Search themes: the closest thing to adding a keyword

If any input feels like "adding a keyword" to a Shopping-style campaign, it's search themes in Performance Max. You give Google a handful of phrases you already know convert, and it uses them to find matching searches faster than it would by learning alone. In accounts where I've migrated Standard Shopping or Search intent into Performance Max, search themes are how I hand the algorithm the query pockets I've already proven, instead of waiting weeks for it to rediscover them.

The critical distinction is that search themes are hints, not exact-match keywords. Google treats them as one signal among many, weighed alongside your feed and audience signals, and it can still serve well outside them. That's a feature, not a limitation: they bias discovery without boxing the campaign in. I've watched a well-chosen set of themes pull a Performance Max campaign toward high-intent commercial queries within days, but I've also seen over-stuffed theme lists add noise, so I keep them tight and grounded in real converting terms rather than aspirational volume.

Where do those phrases come from? For me, always the data, not a brainstorm. I pull the terms that already convert in Search or that surface in the Performance Max search terms insight, and I add those as themes. This is also the natural bridge to feed work: the language you'd research as keywords still applies, but you deploy it inside your product titles and search themes rather than as a bid-per-keyword list. The research doesn't disappear in Shopping; it changes where it lands.

Audience signals: who you reach, not which searches

Audience signals are the second input people mistake for keywords, and the confusion costs performance. A signal doesn't tell Google which searches to appear for; it tells Google which people are likely to convert, your customer lists, in-market segments, custom segments built from URLs and search behaviour. In Performance Max they steer the campaign toward the right users and shorten the learning period, which on a fresh campaign is worth a lot.

The key word is signal. These suggest, they don't restrict. Google can and does serve outside the audience you provide when it finds conversions elsewhere, which frustrates advertisers who expect audience-list rigidity from their Search days. I lean into that flexibility: I seed the campaign with my strongest first-party data, converters, high-value customer lists, and let the algorithm expand from there rather than fighting it. Strong signals mainly accelerate and orient learning; they don't cap it.

In practice I treat audience signals as the "who" and search themes as the "what," and I add them for different reasons. Signals go in when I have quality first-party data worth handing over; themes go in when I have proven query language worth accelerating. If you want the deeper mechanics of how these steer a campaign, the way audience signals work in Performance Max is worth reading alongside this, because the steering behaviour is what makes them complementary to the feed rather than a replacement for keyword targeting.

Negative keywords: the one literal keyword field

Here's where an actual keyword field lives. In Standard Shopping you can add negative keywords directly, and they're the most reliable keyword-like control in the whole Shopping toolkit. They're subtractive, you exclude searches that are irrelevant, wrong-intent, or historically unprofitable, and they refine the pool Google already matched your products into rather than choosing that pool. That distinction is exactly why they can coexist with a keyword-less targeting model.

On the accounts I manage, disciplined negatives are consistently the fastest way to cut wasted Shopping spend. You pull the search terms report, find the queries burning budget with no return, and exclude them. Performance Max gives you far less of this, typically only account-level negatives via support, plus brand exclusions, so if granular query exclusion matters to you, that's a genuine argument for keeping a Standard Shopping campaign in the mix. The full operational detail sits in the negative keyword workflow for Shopping campaigns, which is where I'd point anyone who wants the step-by-step.

Treat negatives as ongoing hygiene, never a one-off. Search behaviour drifts, new irrelevant queries appear, and a list left untouched slowly loses its edge. A short weekly pass through the search terms report keeps the excluded set current, and it matters more in Shopping than Search precisely because you can't proactively pick the queries in the first place, so trimming the wrong ones is a bigger share of your control.

When to add keywords in Google Shopping: timing over inputs

The most common mistake isn't choosing the wrong input, it's adding it too early. The answer to when to add keywords in Google Shopping is: when the data tells you to, not at setup. On a new campaign I resist the urge to pre-load search themes, audience signals and negatives on day one, because before the campaign has query data those additions are guesses that constrain the algorithm during exactly the period it needs room to learn.

My sequence is deliberate. I launch with a clean feed and strong titles first, because the feed is what drives eligibility. I add audience signals early only if I have quality first-party data, since those genuinely speed up learning. Then I wait. Once the search terms report shows a pattern, I act on it three ways: recurring wasted queries become negatives, proven converting language that Google isn't finding becomes a search theme, and known buyer segments get reinforced as signals. That report is the trigger for every addition. It tells me exactly how my feed is being matched and where each lever should go, which is the same feedback loop at the heart of optimising Google Shopping campaigns in modern accounts.

If I had to compress how to add keywords to a Google Shopping campaign into one line: you don't add keywords, you add search themes, audience signals and negatives, and you add them when the search terms report earns them, not before. Launch on the feed, layer signals when you have real data, and let the report drive every subsequent edit. Do that and you get the query control Search users chase with keyword lists, earned through timing and data rather than a targeting field that was never there.