A label that has meant two different things
Search engine marketing is the work of acquiring visibility on search results pages, by paid and organic means together. The term arrived as an umbrella: everything a company did to appear in a search engine. Through the 2000s, as pay-per-click advertising grew into an industry with its own platforms and job titles, SEM drifted in trade usage toward meaning paid search alone, with SEO cast as a separate sibling. Both usages are still in circulation, which is why the question of what search engine marketing means has two common answers depending on who is asked.
The wider definition is worth adopting for a practical reason rather than a historical one. A results page is one piece of real estate. Someone typing a query sees advertisements, organic listings and, increasingly, a generated summary, and does not experience these as separate channels. Organisations that split them into two teams with two keyword lists end up bidding against their own rankings and paying twice for the same research.
Why the split costs more in business markets
In consumer categories the split is merely inefficient. In B2B it is expensive, because the audience is small enough that the same people see both of your appearances. If the qualified market is a few thousand people worldwide, the population that could see your ad and the population that could see your organic listing are effectively identical, encountered repeatedly across a process Gartner describes as a loop rather than a sequence.
Gartner's May 2025 survey of 632 B2B buyers, fielded in August and September 2024, describes buying teams of five to 16 people across as many as four functions. One deal therefore involves several people running many searches over months. Paid search reaches those searching on the days you are bidding; organic reaches everyone at every other moment, including the engineer reading at nine in the evening before an internal review. Deciding which half deserves credit for the resulting opportunity is a bookkeeping argument. Deciding jointly which queries the company must be present for is the actual work, and it is done once.
The data each half owes the other
The exchange runs both ways.
- Paid gives organic query truth. The search terms report shows phrases people actually typed with conversions attached, at a granularity Search Console cannot match, since it withholds low-volume and anonymised queries. In a thin market it is often the only validated commercial vocabulary available.
- Paid gives organic a demand test. Before commissioning a page for a phrase, buy the phrase for a month. If nobody converts on a landing page you fully control, an organic listing will not rescue it.
- Organic tells paid where not to spend. Terms where you hold a strong position with a healthy click-through rate are candidates for lower bids — tested, not assumed.
- Organic supplies better landing pages. The detailed technical page that ranks usually converts a specifier better than a stripped-down campaign page.
One mechanic makes shared vocabulary unavoidable. Google documents that "By default, all keyword match types are eligible to match to close variants. There's no way to opt out." Exact match means same meaning, not same string. Negative keywords and regular search-term review are the only controls, and in B2B that review is where you find a consultant term collecting job seekers.
Bidding on your own name: both cases honestly
The case for. Competitors can bid on your name, and an absent brand ad hands the top of the page to them at the moment a shortlist is forming. An ad gives you control of the message and the destination, which an organic listing does not. Brand terms are usually inexpensive relative to category terms because relevance is high. And during an active procurement, a rival's advertisement above your own listing is a poor look in front of a committee.
The case against. Much of that click may be one you would have had for nothing. Someone searching your company name is already looking for you; if you hold the first organic position and no competitor is bidding, the ad buys traffic you already had.
Resolve it with a test rather than an opinion. Pause brand campaigns in a matched set of regions, or on a matched schedule, and compare total brand-driven conversions — paid and organic combined — against the periods when the ads ran. Check auction insights and the live results page for competitor ads. Bid when someone is taking the position from you, or when the page you want promoted is not the page that ranks.
What the generated summaries change, and what they do not
Google documents two AI features in Search. AI Overviews help users "get to the gist of a complicated topic or question more quickly" and act as "a jumping off point to explore links to learn more." AI Mode is for "queries where further exploration, reasoning, or complex comparisons are needed." Both are grounded in the same core ranking systems as ordinary results, retrieving from the Search index and using query fan-out — "issuing multiple related searches across subtopics and data sources."
Three consequences follow, and nothing beyond them should be claimed. Occupying a results page now includes being retrievable for questions adjacent to the one typed, which rewards depth over single-phrase targeting. There is no separate optimisation to buy: Google states that "There are no additional requirements to appear in AI Overviews or AI Mode," and that no new files, markup or Markdown are needed. The controls are the existing ones — nosnippet, data-nosnippet, max-snippet, noindex, robots.txt — and using nosnippet removes your ordinary snippet too.
Measurement is thinner than the discussion implies. Google's Search Console reporting for generative AI features, launched in June 2026, is impressions-first, carries no click data, and does not separate AI Overviews from AI Mode. Any figure quoted for AI Overview traffic loss is a vendor estimate — Google publishes none.
Mistakes that survive because nobody checks
- Optimising Quality Score. Google documents it as "a diagnostic tool" that "is not an input in the ad auction," and states it "is not a key performance indicator and should not be optimized." A dash instead of a score means too few exactly matching searches, which is normal in B2B rather than a fault.
- Repeating that Performance Max is a black box. Since 2025 it has full search terms reporting, channel-level and asset-level reporting, campaign-level negative keywords up to 10,000, and up to 50 search themes per asset group. Whether that reporting is actionable enough is a fair debate; claiming it does not exist is wrong.
- Chasing impression share on category head terms. Broad category queries in B2B are dominated by researchers and rivals. Impression share is budget consumption dressed as a goal.
- Letting an agency own the search terms report. It is the company's best record of how its market talks, and it belongs beside the content plan.
- Running the halves on different reporting cycles. Paid reports monthly, organic quarterly, and nobody ever sees the results page whole.
The plumbing, one report, and when to skip half of it
Left alone, an SEM programme reports lead volume, which in B2B is close to useless: the campaigns producing the most leads routinely produce the fewest deals. Three pieces of configuration make it answerable to pipeline instead.
- Conversion data returning from the CRM. Enhanced conversions for leads matches hashed customer data back to the campaign and supplements rather than replaces the GCLID click identifier; Google recommends upgrading to it from plain offline conversion import.
- An upload path that still exists. From 15 June 2026, offline conversion import and enhanced conversions for leads uploads move to the Data Manager API and are blocked in the Google Ads API. Pipelines built on the old endpoint have stopped working, quietly.
- Consent handling that does not discard data silently. Consent mode has seven parameters, with ad_user_data and ad_personalization added in version 2. Basic mode blocks tags until a user interacts with the banner; advanced mode loads them with consent defaulted to denied.
Then report once, at query-group level, paid and organic side by side, with a blended cost per qualified opportunity. Skip the paid half when the category has almost no volume to buy, and delay the organic half while a migration is in flight.
Frequently Asked Questions
Is SEM the same as PPC?
Not quite, though the terms are often swapped. PPC — pay-per-click — names a pricing model for advertising and, by extension, the paid campaigns that use it. Search engine marketing, in its original and more useful sense, covers everything a company does to appear on a results page, paid and organic together.
The distinction matters at budget time. Define SEM as paid only and the organic side becomes someone else's line item, never compared on the same terms. Define it broadly and the argument over which team owns a conversion largely disappears.
Should we bid on competitor brand names?
Sometimes, with clear eyes about the economics. Competitor terms convert worse than your own brand terms and better than broad category terms, and they usually cost more per click because relevance is low. They work best when you have a specific, defensible comparison to make and a page that makes it honestly.
Two cautions. Trademark rules differ on what may appear in ad text as opposed to what may be bid on, so involve counsel before writing creative. And expect retaliation: a competitor who notices will bid on your name.
How should we split budget between paid and organic search?
No ratio is worth copying, but there is a sequence. Paid search buys immediate presence and, more valuably at the start, query data you cannot otherwise obtain. Organic work compounds and costs nothing per click once it lands, but takes quarters. Early programmes are therefore usually paid-heavy in order to learn which phrases convert, then shift investment toward pages for the phrases that proved themselves.
The useful decision unit is a query group, not a channel percentage: for these phrases, is presence cheaper to buy or to earn?
How do we measure traffic from AI tools like ChatGPT or Gemini?
GA4 has a default channel for it. The AI Assistants channel covers users arriving from sources like ChatGPT, Gemini, Deepseek, Copilot and Grok, defined by a medium of exactly ai-assistant and assigned automatically when the referrer matches Google's list of recognised assistants. That gives you first-party session and conversion data without custom configuration.
Treat absolute numbers carefully. Google does not publish the referrer list, and every industry figure for AI referral share comes from a vendor analysing its own client panel. Report your own trend, not someone else's benchmark.