When to fight, influence, or generate demand

AI Overviews reshaped the search results page, yet most paid search playbooks still run as if nothing happened.

You still hear the same default rule: bid up where performance looks strong, bid down where it looks weak. That assumes every query is the same kind of opportunity, a small auction for the same kind of click. In B2B and industrial accounts, where sales cycles run for months, conversion data is thin, and much of the demand is informational, that assumption falls apart.

When Google answers a question directly on the results page, the value of a paid click varies across your keywords.

• Some queries still lead to a real commercial decision, and winning the top spot above the answer is worth paying for.
• Others get resolved inside the AI answer itself, so the goal shifts from buying a click to influencing the answer and building a brand strong enough to be included.
• A third group has moved so far toward zero-click that paid search on its own can no longer rebuild the journey.

Those are three different problems, and a single bid lever can’t solve all three. Trying to is how accounts spend budget on the wrong surface. I separate them into three buckets: fight, influence, and generate demand.

The framework: Fight, influence, and generate demand

Each bucket describes a different mechanism on the results page, so each one calls for a different approach.

Bucket 1: Fight

Fight is for bottom-of-funnel queries where intent is clearly commercial, and a click still turns into direct leads and sales: product + modifier, supplier shortlists, “buy,” “quote,” “distributor,” and brand terms combined with buying intent. Here, the job hasn’t changed much. You want to win the auction that sits above the answer, and you want the ad itself to answer the buying question.

This is where paid competition concentrates, and the data show it. The likelihood of an ad appearing rose steadily with cost per click, according to a July SE Ranking study of commercial queries in Google’s U.S. AI Mode:

• About 54% at $10 or more.
• Around 24% for keywords under $2.
• 32% between $2 and $10.

AI Mode isn’t the same surface as AI Overviews, but what they share is the pattern: paid inventory follows commercial weight. Where the market already pays more for a click, AI-era ad surfaces appear more often. You rarely compete alone, since about seven in 10 answers with an ad in that study showed two advertisers in the block.

Fight is expensive for a reason, so treat it that way.

• Reserve the premium for terms where winning absolute top still beats your incremental cost per acquisition. Use bid simulators to estimate the cost of that jump in position first.
• Watch Absolute Top impression share and impression share lost to rank, because that’s how you see the AI block pushing you down the page.

What you shouldn’t do is pour Fight budget into informational queries the Overview has already answered.

Bucket 2: Influence

Influence is for queries where the Overview, or a longer conversational answer, does the educating. Very often, the user won’t click.

The goal here isn’t to gain a cheap visit anymore. It’s to be part of how the answer gets built: cited as a source or presented as one of the brands named inside the answer.

As Shashi Thakur put it at Google Marketing Live EMEA 2026, the best ads must be answers. In an Influence query, your ad brief starts from the question the user is really asking, rather than from a keyword list, and it means accepting that SEO and PPC address the same challenge, not competing against each other.

In B2B and industrial accounts, the queries that trigger AI answer blocks are still mostly informational. In one of our accounts, the intent split of the terms surfacing these blocks looked like this:

Intent	Share
Informational	83.3%
Learn and solve	10.0%
Commercial	6.7%

For “learn and solve” and some “informational” terms, plan for share of the answer instead of maximum CPC.

There is a budget consequence worth stating plainly. On influence queries, we work very closely with SEO, and we’re usually not interested in paying to appear where organic is already winning. Being visible twice, in paid and in the answer, barely happens anyway.

When the data pushes us to move money into the Fight and Generate demand buckets, we deliberately avoid being aggressive on terms where SEO already earns the visibility. That keeps paid spend pointed at the gaps we’re interested in covering.

Bucket 3: Generate demand

Generate demand is for queries where zero-click has already won and paid search alone won’t rebuild the top of the funnel. Informational demand that used to create assisted journeys now often ends on Google, with no visit to anyone.

You have two options. You can:

• Keep bidding while your CPCs climb and volume falls.
• Fund the work that happens before the search: Demand Gen, YouTube, first-party content, and the review and community presence that shows up later when the same person searches with commercial intent.

This work is about recognizing that some keywords have stopped being capture channels and have become a signal of a demand gap. The capture still happens later, in Fight, once the person is ready to act.

Generate demand fills the pool that Fight then fishes from. Measured only on last-click Search ROAS, this work always looks weak because its payoff appears as brand searches and better-qualified pipeline weeks later.

Bucket	The job	The paid response
Fight	Clear purchase intent, where the click still feeds the sales pipeline directly	Pay for absolute top when your brand isn’t already winning the answer box
Influence	Top and mid-funnel, where the answer shapes consideration	Win share of the answer through citation and SEO coordination, not raw CPC
Generate demand	Zero-click is winning and Search alone can’t rebuild the funnel	Build demand before the search, through Demand Gen, YouTube, and awareness

Dig deeper: What happens when AI Overviews contradict paid search ads?

Why one bid rule fails after AI Overviews

AI Overviews sit at the top of the page, sharing space that used to belong to ads and organic results. The answer often satisfies the user before they scroll, changing the economics of paid search in three ways.

1. Click-through falls on informational and mid-funnel queries

When users get a usable answer without leaving Google, both organic and paid lose clicks. The effect is stronger when your brand isn’t mentioned in the AI Overview.

The clicks that remain tend to be better qualified, but the drop in volume usually exceeds the lift in quality, so total conversions decline.

2. Cost per click goes up

There are fewer premium positions at the top. More advertisers are shifting their budgets to paid search to offset lost visibility.

The pressure is strongest in B2B and services, and on comparative queries such as “best X” or “X reviews.” Pure transactional queries hold up better because users still need to click through to buy or request a quote.

3. Position gets harder to read

There are often only a few useful slots above the AI Overview or AI Mode answer, and falling below the Overview can sharply reduce click-through, especially on mobile. Absolute Top impression share and impression share lost to rank become especially useful for understanding that pressure.

Underneath these changes is a shift in behavior. People can research and decide within a single answer experience, shortening the window to capture and persuade them. Intent still lives in your account, but the economics attached to each type of intent have changed. That’s why you should classify a query before you touch the bid.

Dig deeper: 4 strategic paid search pivots to survive Google’s AI Overviews

---

How to classify your terms in practice

The framework only pays off if your account can distinguish between valuable and low-value traffic, so two foundations come first.

One is semantic coherence: keyword, ad, and landing page as a single unit of meaning, with each ad group built around one clear concept. Mixing concepts teaches the system the wrong matches, and that gets worse the moment you hand control to broad match, Performance Max, or AI Max.

The other is conversion quality: primary conversions that reflect real business value, lead scoring, and offline or CRM signals wherever you can pass them back. Smart Bidding and the AI surfaces amplify whatever you give them, so thin data makes every bucket noisier.

With those in place, classification becomes a routine you can repeat:

• Pull your search terms and map intent, separating clear commercial intent from research and from purely informational queries.
• Use SEO data, Search Console, AI Overview Reports, plus manual SERP checks as a proxy for where Overviews appear and whether your brand is cited.
• Sort each term into Fight, Influence, or Generate demand, and revisit the sorting regularly because the SERP keeps changing.

Step	Action	Data Sources	Output
1. Map intent	Pull search terms and classify by intent type.	Google Ads keywords and search terms (+ Search Console).	Terms grouped by: commercial, research, informational.
2. Identify AI presence	Check where AI Overviews appear and whether your brand is cited.	AI Overview reports, manual SERP checks.	Map of AI Overview presence by term.
3. Sort into buckets	Assign each term to a strategic bucket.	Intent mapping + AI presence data.	Fight, Influence, or Generate demand list.
4. Revisit Regularly	Re-sort terms as SERP landscape changes.	Ongoing SERP monitoring, AI Overview reports.	Update bucket assignments and review budget assigned to each.

On Fight terms, accept a position premium only where absolute top still wins your incremental cost per acquisition.

On Influence terms, align ad copy with the question the Overview answers, coordinate with SEO on content and citation, and pull back paid pressure where organic already wins.

On Generate demand terms, move budget into YouTube, Demand Gen or content when the Search clicks are gone but the query still matters and measure through brand search and assisted pipeline.

AI Max and broad match sit on top of this classification. They don’t replace it. Google needs broad match or keywordless campaigns for your ads to appear around AI answer surfaces, which expands reach while reducing control over what you show.

Use them where you already know the bucket and your tracking is clean, and introduce them through small, gradual tests.

Dig deeper: How to get your Google Ads seen in AI Overviews

Put the budget where the opportunity is

Accounts that apply one rule to every query will keep funding the wrong mechanism, seeing it as rising costs they can’t explain. Accounts that re-segment and move money between buckets and into channels outside the auction spend less time arguing with a results page that has already answered the user.

In B2B and industrial search especially, shifting from optimizing bids to reallocating budget across the three buckets is the difference between defending last year’s setup and building for how people search now.