How Search Engine Marketing Works (And How to Win with AI)

Key takeaways

• Expensive keywords aren't a cost to minimize — they're a map. The price buyers force each other to pay for a click tells you exactly which problems are worth writing about, whether you ever run the ad or not.
• Google's search-ad business is still growing (+17% YoY, Q2 2026) even as AI Overviews spread — right now the two channels add up, they don't replace each other.
• Getting your brand cited in ChatGPT, Gemini, Perplexity, and AI Overviews is the unpaid version of winning that same auction.
• Do this month: price out what a lead is worth to you, check whether you show up in 20 buying-intent AI prompts, and fix the one page that actually converts.

Search engine marketing (SEM) means paying to show up in search results. You bid on a keyword, your ad appears above the regular listings when someone searches it, and you pay per click, not per placement. That's the line most sources now draw between SEM and SEO: SEM is bought, SEO is earned — though the two terms have blurred together enough over the years that plenty of guides still use “SEM” to cover both. Either way, the mechanics of the paid side are not in dispute: an auction, a bid, a budget, a result you can measure to the cent.

How much money does Google make from ads?

Google's advertising business earned $81.6 billion in the second quarter of 2026 (April to June), about 14.5% more than a year earlier. Alphabet reported the figures on July 22, 2026.

Search is the engine. It brings in more than three quarters of the total and grew about 17%, while the Network business stayed flat. The first quarter of 2026 added $77.3 billion, so the first half of the year came to roughly $159 billion.

For the full year 2025, Google's ad revenue was about $294.7 billion. At the Q2 pace, that is heading toward $326 billion a year.

How does the Google Ads auction work?

Google sells clicks in an auction that runs every time someone searches, and the price of a click is set by what competing advertisers are willing to pay for it. Advertisers choose the keywords they want to appear for and the most they will pay per click. You pay only when someone actually clicks your ad.

Bid size is not the only factor. Google also scores how relevant your ad is to the search and how good the page behind it is. An advertiser with a better ad and a better landing page can win the top spot while paying less than a competitor with a bigger bid.

There's no fixed price list. A click costs whatever the most motivated advertisers can afford, and what they can afford depends on what a customer is worth to them.

Why are some keywords so expensive?

Keywords get expensive when one click can turn into a large sale. Take a mesothelioma lawyer. A firm that signs one such client can earn a fee in the tens or hundreds of thousands of dollars. If the firm turns even 1 in 20 clicks into a signed client, paying $500 per click, or about $10,000 per client, is still profitable. So the firm can bid $500 and come out well ahead. The same logic applies to a business loan, a commercial insurance policy, or a compliance contract.

Four things push the price up further:

• Many bidders want the same searches. Personal injury firms in every big US city compete for “truck accident lawyer.” When ten firms can each afford $300, the winning price climbs toward that level.
• The searcher is ready to buy. Someone typing “truck accident lawyer” has just had an accident and is choosing a lawyer now. That click is worth far more than one from a person reading about trucks.
• Quality scores level the field. In crowded categories every advertiser has a strong ad, so bid size becomes the deciding factor.
• The searches are rare. Only so many people each month need this exact service, so demand for those clicks stays high.

Cheap keywords work the opposite way. A click on “how to write a LinkedIn post” is worth almost nothing to any advertiser, because it rarely leads to a sale.

What are the most expensive Google Ads keywords right now?

Legal keywords top every list, with estimated costs of $250 to $900 or more per click for the most expensive terms. The table shows estimated ranges from a July 2026 analysis.

Business software and services cost less but are still high — SaaS and marketing terms run $30 to $70 per click, with EMR software for clinics reaching $100 or more. Broad single words such as “insurance” or “loans” cost less than specific ones, at around $45 to $55 per click in an older list.

Treat every number here as an estimate. Google does not publish keyword prices, and each list comes from a different tool or sample, so the ranges vary from source to source.

What can a small business learn from expensive keywords?

The main lesson is to work out what a lead is worth before deciding how much to spend on one. The lawyers pay $500 per click because they know what a signed client is worth. Most small businesses have never done that calculation. Take your average customer's lifetime profit and multiply it by your close rate. If a customer is worth 2,000 euros in profit and you close 1 in 5 leads, a lead is worth about 400 euros, and a click is worth a fraction of that.

Four more lessons follow from the same logic:

• Do not compete where you cannot win. A small business with a modest customer value will lose the auction for broad terms like “insurance” or “business loan.” Target narrower searches with a specific city, problem or customer type. “SOC 2 compliance for a 30-person SaaS startup” costs far less than “SOC 2 compliance” and is often a better lead.
• What happens after the click matters more than the click price. A $50 click that becomes a customer beats a $5 click that never does. Improving how many visitors become leads is usually cheaper than paying less per click.
• Build channels that do not charge per click. Content that answers your buyers' specific questions, comparison pages, LinkedIn, email and referrals all reach the same people over time.
• Read high prices as a map of what is profitable. Expensive terms in your niche show which problems buyers pay to solve. Those are the topics worth writing about, even if you never buy the ad.

What should a SaaS company owner do differently?

A SaaS owner should bid according to what a customer is worth over the whole subscription, not the first payment. A customer paying 200 euros a month who stays for three years is worth over 7,000 euros in revenue. Many SaaS companies bid as if they were selling a one-off product, undervalue their own customers, and lose auctions they could easily afford to win.

Three rules follow:

• Work out your numbers first. Know your customer lifetime value, how much you can spend to win one customer (a common rule of thumb is a third or less of lifetime value), and how long it takes to earn that money back. Work backwards to the most you can pay per lead and per click.
• Bid on searches close to a purchase. In SaaS these are terms like “[category] software for [specific industry]” and “[competitor] alternative.” They can cost $30 to $100 or more per click, but the visitor is close to buying. Broad terms like “software” attract people who are only researching.
• Fix the page before raising the budget. If your landing page and demo request convert twice as well, you can pay twice as much per click as a competitor and still pay the same per customer. That advantage lasts, and any small SaaS company can build one.

Keep building channels that do not charge per click, such as comparison pages and content that answers buyers' questions. Paid ads then work as a testing ground: they show which topics and offers convert, and you can put the winners into content that keeps working after the ad budget stops. [INTERNAL LINK: point this at your conversion-optimization / funnel-audit page — none is linked yet]

Is AI Search Replacing Search Engine Marketing?

Not yet: Google's search ad revenue grew about 17% in the second quarter of 2026, even as AI answers spread through its results. AI is changing where buyers look, but the paid auction is still growing.

On the July earnings call, Google's CEO said AI Mode has passed 1 billion monthly active users since it went global in October. He also said AI Overviews and AI Mode are “driving an incremental increase in Search queries overall” and that Google is “sending billions of clicks to websites every week” through AI features. He gave no details on how ads perform inside those AI features.

For marketers this means two channels to work on at once. Buyers still search and click ads, and they also ask AI tools questions like “best [software] for [industry]” or “[competitor] alternative.” Being named in those answers is the organic version of winning an expensive auction, without paying per click.

How do you get your brand mentioned in AI answers?

Start with the prompts that matter, then earn mentions on the pages and platforms AI tools draw from. The same value logic applies as in paid search: focus on the questions where a buyer is close to choosing. Nine steps:

• Pick the prompts using the keyword logic. Turn your high-value, high-cost searches into the questions buyers ask AI: “best [category] for [specific customer],” “[competitor] alternative,” “[category] pricing,” “[A] vs [B].” Aim for 20 to 30 prompts, weighted toward buying intent.
• Take a baseline before changing anything. Run each prompt in ChatGPT, Google AI Mode and Perplexity. Record whether you are mentioned, who else is, which URLs are cited, and whether the AI says anything wrong about you. Repeat weekly, because answers change.
• Write pages that answer these prompts directly. Put the answer in the first lines, then the detail: comparison pages, alternative pages, pricing explainers and “for [segment]” pages. Say clearly who the product suits and who it does not.
• Build presence outside your own site. AI engines cite third parties heavily. In a study of 680 million citations (August 2024 to June 2025), Wikipedia led for ChatGPT, while Reddit led for Google AI Overviews and Perplexity. YouTube, LinkedIn, G2 and Gartner also appeared. That points to real customer reviews, useful contributions in Reddit and LinkedIn discussions, video, and inclusion in independent “best of” lists.
• Treat each engine separately. One analysis found only about 11% of cited domains appear in both ChatGPT and Perplexity. Winning in one does not carry over to the others.
• Keep your facts consistent everywhere. Use the same product description, pricing logic and customer types on your site, review profiles and social pages. AI tools merge sources, and mismatches produce wrong or missing mentions. Fix any errors from step 2 at the source.
• Refresh key pages regularly. Update comparison and pricing pages with current dates and figures. The same analysis calls Perplexity the most freshness-sensitive engine.
• Make the click count. Visitors from AI answers have often already narrowed their options, so the landing page should confirm they are in the right place and offer a clear next step, such as a demo or a specific comparison.
• Keep paid ads for terms you cannot yet win organically. Use them to test which offers and messages convert, then write content around the winners.

What are the limits of this data, and what should you do first?

The advice here is sound, but the numbers behind it have limits. Keyword prices are third-party estimates, since Google does not publish them. The AI citation studies end in mid-2025, and AI platforms change how they choose sources often, so your own weekly prompt tracking is a better guide than any benchmark. The 2025 full-year ad revenue figure comes from Statista, not from Alphabet's own filing. I also found no reliable data on how AI-driven traffic converts compared with paid clicks, so measure that yourself, for example by asking new leads how they found you.

If you do only three things this month:

• Calculate what one lead is worth to your business.
• List 20 buying-intent questions a customer might ask AI, and check whether your brand appears in the answers.
• Improve the one page that turns visitors into leads.

Sources

Alphabet Q2 2026 results (SEC exhibit) — sec.gov

Google Q1 2026 Earnings, AdStatus — adstatus.app

Google advertising revenue 2025, Statista — statista.com

Sundar Pichai, Alphabet Q2 2026 remarks — blog.google

Most expensive Google Ads keywords, ALM Corp (July 2026) — almcorp.com

Most expensive Google keywords in 2026, ClickPatrol — clickpatrol.com

The 20 Most Expensive Keywords in Google Ads, WordStream — wordstream.com

AI Platform Citation Patterns, Profound — tryprofound.com

AI Citation Behavior Index, GEO Software Rankings — geosoftwarerankings.com