Lecture 4: SEM Account Structure: Campaigns, Ad Groups, and Keywords

Short answer: Account structure is the skeleton that everything else in SEM hangs on. How you group keywords into ad groups, ad groups into campaigns, and campaigns into an account determines what budgets you can set, what bid strategies will actually work, how clean your reporting is, and how much manual labor you'll do every week for the life of the account. Get it right early and the account scales with almost no restructuring. Get it wrong and you'll spend months untangling overlapping keywords, diluted conversion data, and budgets that can't be isolated by product line or funnel stage. This lecture gives you a repeatable framework for structuring any account, from a five-keyword local business to a thousand-SKU e-commerce catalog.

What You'll Learn in This Lecture

• Why account structure is a performance lever, not just an organizational nicety
• Which campaign-level settings actually matter and why they can't be set at the ad group level
• The three dominant ad group philosophies — SKAG, themed groups, and STAG — and which one fits which situation
• How to build naming conventions that make Excel pivot tables and scripts trivial instead of painful
• How to structure by product line, funnel stage, and geography without exploding campaign count
• Why Smart Bidding needs conversion volume concentrated per campaign, and how consolidation fixes underperforming automation
• Where every modern campaign type — Search, Shopping, Performance Max, Display, Video, Demand Gen — sits in the structure
• A step-by-step process for restructuring a messy legacy account without losing years of historical performance data
• A full worked example: an account blueprint for a mid-size e-commerce business, campaign by campaign
• The ten most common structural mistakes that quietly waste budget
• How structure decisions made today constrain or enable what you can do in Lecture 5 onward

Why Account Structure Determines Performance and Manageability

Every setting in Google Ads lives at a specific level of the account hierarchy: account, campaign, ad group, or keyword/ad. Budgets, networks, locations, languages, and bid strategies are campaign-level settings — you cannot give two products in the same campaign different daily budgets or different geographic targeting. Ad copy relevance and Quality Score are ad-group-level concerns — the tighter the keyword theme in an ad group, the more relevant your ads can be, and the higher your Quality Score tends to run, which lowers your effective cost per click as covered in Lecture 3. Conversion data, meanwhile, aggregates at the campaign level for bid strategy purposes, which means how you split campaigns directly determines whether Smart Bidding algorithms have enough signal to work with.

Structure is therefore not administrative housekeeping — it is a set of decisions that caps or unlocks what you can do downstream. A poorly structured account cannot isolate budget by profitability, cannot run different bid strategies for different intents, cannot report cleanly to stakeholders, and cannot give automated bidding the conversion density it needs. A well-structured account, by contrast, lets you increase budget for high-margin categories in seconds, pause a whole funnel stage during a promotion, and read performance reports that map directly onto how the business actually thinks about its products.

Example: An outdoor gear retailer originally built one campaign called "All Products" with 40 ad groups mixing tents, boots, and jackets. When a supplier ran out of stock on jackets mid-quarter, the team could not pause jacket spend without also touching the shared campaign budget field, and the Search Term reports were a wall of unrelated queries. After splitting into three campaigns — Tents, Boots, Jackets — each with its own budget and location targeting, the marketing team could reallocate spend in minutes and Smart Bidding converged twice as fast because conversions were no longer split three ways within one campaign.

Campaign-Level Settings That Matter

Because these settings can only be set once per campaign, every keyword grouped underneath inherits the same values. Getting the campaign boundary wrong forces compromises on all of the following:

Budget. Daily budget is a campaign-level field. If two product lines with very different margins share a campaign, you cannot give the higher-margin line more room to spend without also inflating the lower-margin line. Campaigns are the smallest unit at which you can say "spend more here, less there."

Network. Search Network, Search Partners, and (for other campaign types) Display Network are toggled per campaign. Mixing intents that perform differently on partner networks into one campaign means you either accept diluted performance or forfeit the ability to test partner inclusion separately.

Location and language targeting. If you sell in the US and Canada with different pricing or promotions, or in English and French, these need separate campaigns — location and language are campaign settings, not ad group settings. A single campaign targeting "United States, Canada" cannot show Canadian pricing only to Canadian searchers via targeting alone.

Bid strategy. Target CPA, Target ROAS, Maximize Conversions, and Manual CPC are all chosen per campaign. This is the single biggest reason structure and automation are linked: a campaign mixing a $20 product and a $2,000 product under one Target ROAS strategy will fight itself, because the algorithm is trying to hit one blended target across wildly different conversion values.

Ad schedule and device adjustments (in older structures) and campaign-level negative keywords. These also live at the campaign level and are worth planning for even though modern Smart Bidding reduces the need for manual dayparting.

Ad Group Organization Approaches

Within a campaign, ad groups are where keyword-to-ad relevance is won or lost. There are three schools of thought, and which one is right depends heavily on account size, bid strategy, and how much conversion volume you have to spare.

Single Keyword Ad Groups (SKAG) — Still Relevant?

SKAG structure puts exactly one keyword (in one or two match types) into its own ad group, with ad copy written to mirror that keyword as closely as possible. In the exact-match-heavy, manual-bidding era, this maximized Quality Score and click-through rate because the headline could echo the search term almost verbatim. The tradeoff was always account bloat: a 200-keyword account became 200+ ad groups, each needing its own ads, and each splitting conversion volume into a sliver too small for any bidding algorithm to learn from.

Today, SKAG is largely obsolete for accounts running Smart Bidding, for two reasons. First, Google's close-variant matching and broad match expansion mean an ad group built around one exact keyword still serves on a wide variety of query variants, undermining the tight-relevance premise SKAG was built on. Second, and more importantly, fragmenting conversions across hundreds of tiny ad groups starves Target CPA and Target ROAS of the volume they need per campaign to exit the learning phase reliably. SKAG still has a narrow use case: a handful of extremely high-value, high-volume "money" keywords (often branded terms or a flagship product) where you want a dedicated ad and landing page and where conversion volume per keyword is already large enough to stand alone. As a default account-wide structure, though, it is no longer best practice.

Themed / Tightly-Grouped Ad Groups in the Smart Bidding Era

The current default is a themed ad group: 5-20 closely related keywords that share a search intent and could plausibly be served by the same one or two ads. "Waterproof hiking boots," "waterproof hiking boots men," "best waterproof hiking boots" belong together; "hiking boots" and "hiking socks" do not, even though both are hiking-related. The test is whether a single ad headline would feel natural as a response to every keyword in the group — if you'd need a materially different headline for some of them, split the group.

Themed grouping keeps relevance and Quality Score high while consolidating enough conversion volume per ad group and per campaign that Smart Bidding has a real signal to optimize against. This is the right default for the large majority of Search campaigns in 2026.

Single Theme Ad Groups (STAG)

STAG is themed grouping taken one level further: instead of grouping by loose topical similarity, each ad group represents exactly one theme or one product variant, with all its reasonable keyword variations (plurals, synonyms, modifiers like "buy," "price," "near me") inside it, and nothing else. The distinction from a generic "themed" group is discipline — a STAG has a written definition of what belongs and what doesn't, which prevents theme creep as new keywords get added over months. In practice, well-run themed ad groups and STAG converge on the same output; STAG is simply the more rigorous version of the same idea, useful when multiple team members add keywords to the same account and need a clear rule to follow.

Naming Conventions for Campaigns and Ad Groups

At five campaigns, naming barely matters. At fifty, an inconsistent naming scheme makes filtering, pivot tables, scripts, and automated rules unusable. Build the convention before the account grows, using a fixed, pipe- or underscore-delimited field order so every campaign name sorts and filters the same way.

A durable pattern: [Network]_[ProductLine]_[FunnelStage]_[Geo]_[Match/BidType]. For example: Search_Tents_NonBrand_US_Exact, Search_Tents_Brand_US_Exact, PMax_Boots_US, Search_Jackets_Competitor_CA. Ad group names follow the same discipline at their level, naming the specific theme: 4-Season-Tents, Backpacking-Tents-Ultralight. The goal is that anyone on the team can read a campaign or ad group name and know its network, product, funnel stage, and geography without opening it — and that a spreadsheet export can be split into columns on the delimiter to build pivot tables instantly.

Avoid free-text names like "New Campaign - Spring Sale Test 2" that carry no structured information; if you need to note a test or date, append it as a final, separate field rather than replacing the structured prefix.

Structuring Accounts for Reporting Clarity

Reporting clarity comes from choosing one primary axis to split campaigns on and layering secondary axes underneath rather than trying to encode everything at the campaign level. The three most common primary axes are:

By product line. Each product category or line of business gets its own campaign or campaign set (Tents, Boots, Jackets). This is the natural choice for e-commerce and multi-product B2B accounts because budget and margin decisions are usually made at the product-line level by stakeholders.

By funnel stage. Separate campaigns for Brand terms, Non-Brand/generic terms, Competitor terms, and Remarketing/retargeting audiences. This split matters because these four groups have wildly different cost-per-click, conversion rate, and strategic purpose — blending them hides whether generic prospecting is actually profitable versus riding on cheap brand conversions.

By geography. Needed whenever pricing, promotions, shipping, language, or business hours differ by region. Combine with product-line splitting only where the business genuinely requires it — an unnecessary geo split (e.g., separating US-East and US-West with identical everything) just fragments data for no reporting benefit.

The practical approach for most accounts is to combine product line as the primary axis and funnel stage as the secondary axis, layering geography in only where it changes settings that matter (targeting, budget, language). This keeps the campaign count proportional to genuine business differences rather than growing arbitrarily.

Structuring for Automated Bidding

Smart Bidding strategies — Target CPA, Target ROAS, Maximize Conversions — learn from the conversion data recorded within a campaign (and, for portfolio strategies, across a shared bid strategy). Google's own guidance is that a Target CPA or Target ROAS campaign generally needs on the order of 30-50 conversions in the trailing 30 days to exit the learning phase and bid reliably; below that, the algorithm has too little signal and bids erratically, often overspending or underspending unpredictably.

This is where structure and automation directly collide. An account split into twenty small campaigns to satisfy a granular product-line reporting preference, each getting only 5-10 conversions a month, will underperform an account consolidated into five campaigns each getting 40-80 conversions, purely because of bidding-algorithm data density — even though the granular version looks tidier in a spreadsheet. The fix is consolidation: merge low-volume campaigns that share a bid strategy and target audience into a single campaign, and rely on ad group-level and asset-level reporting to preserve the product-line detail you actually need for analysis, rather than encoding it in the campaign boundary.

Where genuine business reasons force a split (different budgets, different margins requiring different ROAS targets), and volume is too low to support it, consider a shared portfolio bid strategy across those campaigns so the algorithm pools conversion data while budgets remain separately controllable. This is frequently the right compromise for growing accounts that aren't yet large enough to support fully separate Smart Bidding per product line.

Campaign Types Overview and Where They Sit in the Structure

Modern Google Ads accounts rarely run Search alone. Each campaign type has a distinct role and should be planned as part of the same structural blueprint rather than bolted on afterward:

Search campaigns target explicit typed queries and remain the backbone for high-intent, bottom-of-funnel capture — this is where the ad group philosophies above apply most directly.

Shopping (via Performance Max or standalone Shopping where still available) surfaces product listings with images and price directly in search results, organized by product feed structure (via product groups) rather than keywords — structure here is largely inherited from your merchant feed's category and custom label fields, so feed organization becomes an extension of account structure.

Performance Max is a goal-based campaign type that spans Search, Display, YouTube, Discover, Gmail, and Maps inventory from a single campaign, organized internally by "asset groups" rather than ad groups. Because Performance Max pools budget and signal across channels, it sits at the same structural level as a Search campaign but should generally be split along the same product-line or funnel-stage axis you use elsewhere, and given its own separate budget so it doesn't compete unpredictably against standalone Search campaigns for the same queries.

Display campaigns (standalone, increasingly folded into Performance Max) handle awareness and remarketing against a visual ad network and are structured around audiences and placements rather than keywords.

Video campaigns on YouTube follow a similar audience/placement structure and typically sit as a distinct upper-funnel layer in the account, often mapped to the "awareness" funnel stage in your naming convention.

Demand Gen campaigns target visually-driven placements (YouTube Shorts, Discover, Gmail) using audience signals and lookalike-style targeting, generally positioned as a mid-funnel complement between Video/Display awareness and Search/Shopping conversion capture.

The structural principle that ties all six together: decide your product-line and funnel-stage taxonomy once, then apply it consistently as you add each campaign type, so a stakeholder can look at the full account and understand which campaigns, regardless of type, are serving the same business goal.

How to Restructure a Messy Legacy Account Without Losing Historical Data

Restructuring is common — most accounts inherited from a previous agency or added to organically for years need it — but done carelessly it can sever historical performance data and reset Smart Bidding learning. Follow this sequence:

1. Audit before touching anything. Export campaigns, ad groups, keywords, and 12-24 months of performance data. Identify which campaigns hold the account's conversion history and Quality Score equity — these are assets you don't want to discard casually.

2. Design the target structure on paper first. Map every existing keyword to its new campaign/ad group home using the product-line and funnel-stage taxonomy above before making any changes in the interface.

3. Prefer renaming over rebuilding wherever possible. Renaming an existing campaign preserves 100% of its historical data and bid strategy learning; deleting and recreating a campaign discards its history entirely and forces bidding into a cold-start learning phase again. If a campaign's existing scope already roughly matches its new intended scope, rename it rather than starting over.

4. Where genuine restructuring is needed, move rather than delete. Google Ads allows copying ad groups and keywords across campaigns. Copy the keyword and its accumulated Quality Score signal into the new ad group location, confirm performance is being recorded correctly, and only then pause (not delete) the old ad group — pausing preserves the historical record for trend reporting while stopping new spend.

5. Stage the cutover. Run old and new structures in parallel for a short window if budget allows, or cut over during a lower-traffic period, and expect a short re-learning phase for any bid strategy attached to a genuinely new campaign. Document the change with a dated note in the account so future analysts understand a discontinuity exists in the data.

6. Never rebuild for its own sake. If the existing structure already satisfies the budget, network, geo, and bid-strategy requirements above, a rename-and-reorganize-ad-groups pass is enough; full campaign recreation should be reserved for cases where campaign-level settings themselves need to change.

Example Account Blueprint for a Mid-Size E-commerce Business

Consider a mid-size outdoor retailer selling tents, boots, and jackets in the US and Canada, with roughly $30,000/month in SEM spend. A clean blueprint, following product line as the primary axis and funnel stage as the secondary axis:

Search_Brand_US-CA (branded terms, all product lines, Target Impression Share) → ad groups: Brand-Exact, Brand-Misspellings.
Search_Tents_NonBrand_US (Target ROAS) → ad groups: Backpacking-Tents, Family-Camping-Tents, 4-Season-Tents, Ultralight-Tents.
Search_Boots_NonBrand_US (Target ROAS) → ad groups: Hiking-Boots-Men, Hiking-Boots-Women, Waterproof-Boots, Trail-Running-Shoes.
Search_Jackets_NonBrand_US (Target ROAS) → ad groups: Rain-Jackets, Insulated-Jackets, Softshell-Jackets.
Search_AllProducts_NonBrand_CA (Target ROAS, consolidated because Canadian volume alone is too low to split by product line) → ad groups: Tents-CA, Boots-CA, Jackets-CA.
Search_Competitor_US (Manual CPC or Maximize Clicks, low budget, tightly capped) → ad groups grouped by competitor brand.
PMax_Tents_US-CA, PMax_Boots_US-CA, PMax_Jackets_US-CA (Shopping-led Performance Max, one per product line, separate budgets, asset groups mirroring the Search ad-group themes).
Display_Remarketing_US-CA (all products, past-site-visitor audiences, Target CPA).
Video_Awareness_US (brand awareness, non-converting-focused, Target CPM or Maximize Conversions with a loose target).

This blueprint gives finance a clean way to see spend by product line, gives the marketing team the ability to pause or boost any one line without touching another, gives every Target ROAS campaign enough conversion volume to bid well, and gives Canada its own settings without fragmenting an already-thin data set further than necessary.

Common Structural Mistakes That Waste Budget

• Mixing brand and non-brand keywords in one campaign. Brand traffic's high conversion rate masks poor non-brand performance and misleads budget decisions.
• Over-fragmenting campaigns below the conversion volume Smart Bidding needs. Looks organized, performs poorly, because the algorithm never gets enough signal per campaign to bid confidently.
• Broad, catch-all ad groups with 50+ loosely related keywords. Kills Quality Score and ad relevance because no single ad can speak to every keyword in the group.
• No naming convention, or one that changes over time. Makes reporting, filtering, and automated rules unreliable as the account grows past a handful of campaigns.
• Running Performance Max and Search for the same products with no budget priority or exclusion strategy. The two can cannibalize each other's traffic without clear signals on which is actually driving incremental results.
• Geo-splitting campaigns that don't actually need different settings. Fragments conversion data for no reporting or targeting benefit.
• Deleting campaigns instead of pausing during restructures. Permanently destroys historical trend data that's often needed for later analysis or seasonal comparisons.
• Leaving legacy SKAG structure in place under a Smart Bidding strategy. Starves the algorithm of concentrated conversion data it needs across hundreds of micro-ad-groups.
• Ignoring product feed structure when building Shopping/Performance Max campaigns. Since Shopping structure is feed-driven, a disorganized feed (poor product_type or custom_label values) undermines even a well-planned campaign structure.
• Building structure once and never revisiting it as the product catalog or business changes. Structure should be reviewed at least annually or whenever the business adds a new product line, market, or funnel strategy.

Account structure is the foundation the rest of this course builds on: keyword research (Lecture 5) populates the ad groups you design here, ad copywriting (later lectures) is written per ad group theme, and bid strategy performance is bounded by exactly how well campaigns are consolidated for conversion volume. Get this scaffolding right now, and every later optimization compounds instead of fighting the structure underneath it.

Related Lessons Across SEO, AEO, GEO, SEM, and PPC

Use these connected lessons to move through organic search, answer engines, generative AI visibility, paid search, and PPC without losing the bigger strategy.

• Lecture 1: What Is SEM? How Search Engine Marketing Works in 2026 (SEM) - return to the course foundation when you need the big picture.
• Lecture 1: What Is PPC? How Pay-Per-Click Advertising Works (PPC) - separate SEM strategy from PPC execution.
• Lecture 38: PPC Strategy Roadmap: Bringing Search, Social, Retail, and AI Together (PPC) - connect SEM with the full PPC channel roadmap.
• Lecture 4: Keyword Research Fundamentals (SEO) - connect organic keyword research with the same demand signals.
• Lecture - 7: Prompt Research: Finding What People Ask AI Tools About Your Topic (GEO) - turn search queries into AI prompt research.