Search Marketing FAQ

Closely related. AI Overviews are powered by Gemini and pull from Google's organic ranking + knowledge graph. Optimizing for Gemini and optimizing for AI Overviews are largely the same exercise.

in Gemini

Google-Extended is the user-agent token Google uses to control whether your site is included in Gemini training, separate from Googlebot for Search. Allowing it lets Gemini learn from your content; blocking it removes you from training but does not affect Google Search ranking. Default-allow for commercial sites.

in Gemini

Two layers. (1) Training: Google-Extended ingests web content for model training. (2) Retrieval: Gemini taps Google's live search index for current information, similar to how AI Overviews work. Strong organic Google rankings translate directly to Gemini visibility.

in Gemini

No, but strong Google ranking is the largest single input. Gemini also weights extractability - direct-answer leads, schema, structural clarity - on top of organic position. A page that ranks #5 with clean extraction can appear in Gemini answers above a page that ranks #2 without it.

in Gemini

Gemini in Workspace surfaces your content when users in Gmail, Docs, or Slides ask Gemini for research or drafting help. For B2B brands targeting users who do research inside their work tools, that's a meaningful new surface - and it's optimized via the same fundamentals as standalone Gemini.

in Gemini

Hard to quantify exactly because Gemini powers multiple surfaces (standalone app, AI Overviews in Google Search, Workspace), but the integration into Google Search alone makes it the most-served LLM by query volume - billions of Gemini-powered impressions per day via AI Overviews.

in Gemini

Copilot is Microsoft's LLM, integrated across Microsoft 365 (Word, Excel, PowerPoint, Outlook, Teams), Bing Search, Edge browser, and the Copilot standalone app. It's available to 400 million Microsoft 365 commercial users plus consumer Bing users - heavily enterprise-skewed.

in Copilot

Copilot uses a mix of OpenAI's GPT models and Microsoft's own Prometheus orchestration layer plus retrieval over Bing's index. From a GEO perspective, Copilot draws on the same retrieval foundation as ChatGPT Search - Bing rankings feed both.

in Copilot

Yes, directly. Copilot's live retrieval pulls from Bing, so Bing SEO is the foundation of Copilot GEO. Optimizing for Bing produces Copilot visibility as a bundled outcome.

in Copilot

Heavily enterprise - Microsoft 365 commercial users, Windows users, Edge browser users. Demographics skew older, higher-income, business-context. For B2B and enterprise-targeted brands, Copilot is often a higher-priority GEO target than ChatGPT despite lower aggregate volume.

in Copilot

Copilot in Microsoft 365 surfaces content (yours and others') when users research inside their work tools. A user drafting a vendor evaluation in Word can ask Copilot for sources; if your brand is well-represented in Bing's index and structured for extraction, you appear. It's a quiet but increasingly important B2B surface.

in Copilot

Copilot uses Bingbot for retrieval - there's no separate Copilot crawler. Bing's robots.txt directives control Copilot indexability the same way they control Bing Search indexability.

in Copilot

Query-panel methodology, same as the other LLMs - representative queries run regularly with citation, presence, and competitive context tracked. Copilot's enterprise user base often warrants a different query panel than the consumer-leaning ChatGPT or Perplexity panels.

in Copilot

Hierarchy: campaign to ad group to ad. Campaign sets budget, network, and bidding strategy. Ad group sets keywords (in Search) or targeting (in other networks) and contains the ads. Ad is the actual creative shown to the user. Account architecture is one of the highest-leverage decisions because misstructured accounts can never optimize past their structural ceiling.

in Google Ads

Wildly category-dependent: legal can run $30-$300+, e-commerce $0.30-$3, B2B SaaS $5-$40. Benchmarks matter less than the cost-per-conversion at your target volume. We size budgets to the addressable demand and the unit economics, not to a generic CPC benchmark.

in Google Ads

Search Network: text ads on Google search results pages - intent-driven. Display Network: image and rich media ads across millions of partner sites - audience and context-driven. Shopping: product-image-and-price ads driven by your product feed - high-intent commercial. Each has distinct optimization disciplines and shouldn't be combined into one campaign by default.

in Google Ads

Responsive Search Ads (RSAs) are Google's modern ad format - you provide up to 15 headlines and 4 descriptions, and Google's algorithm assembles combinations to match each query. They outperform old expanded text ads on average because the algorithm finds matches you wouldn't have hand-built. They require disciplined headline diversity to perform well.

in Google Ads

Performance Max is an automated campaign type spanning all Google networks, optimized by machine learning. It works when (a) conversion tracking is clean, (b) you have rich asset libraries, and (c) it runs alongside branded search to prevent cannibalization. Without those, it underperforms. We deploy it surgically.

in Google Ads

Smart Bidding (Maximize Conversions, Target CPA, Target ROAS, Maximize Conversion Value) lets Google's algorithm bid on each auction. It outperforms manual bidding once it has 30+ days of clean conversion data. The trust isn't in the algorithm - it's in your conversion data. Bad data in, bad bids out.

in Google Ads

Impression share is the percentage of available impressions your ads received vs. the total available. Lost impression share splits into lost to budget (you ran out of money) and lost to rank (your bid or Quality Score wasn't competitive). It's the diagnostic for why a campaign isn't scaling - different lost-IS reasons require different fixes.

in Google Ads

A keyword is what you bid on in your campaign. A search term is the actual query a user typed that triggered your ad. The search-terms report shows you what you actually showed for - and is the source of negative keyword candidates and new keyword expansion ideas. Reviewing it weekly is a non-negotiable for healthy accounts.

in Google Ads

The trimming phase is where we identify and pause underperforming campaigns, ad groups, keywords, and search-term variants - typically the long tail consuming 30-50% of spend with sub-target return. The freed budget is redirected to the keywords and audiences that are converting. It's the highest-ROI move in the first 60 days of most engagements.

in Google Ads

Three main types. Sponsored Products: keyword- or product-targeted ads on search results and product pages - the workhorse. Sponsored Brands: banner-style ads with logo, headline, and product set - top of search results. Sponsored Display: retargeting and audience ads on and off Amazon. We layer all three with distinct goals: defense, discovery, and remarketing.

in Amazon Ads

Auction-based, cost-per-click - you bid against other sellers for keyword-triggered ad placement. Bids depend on category competition and your product's conversion rate (Amazon's auction favors high-converting products at lower bids, similar to Google's Quality Score logic).

in Amazon Ads