Paid Search Campaigns

Introduction and Context

Paid search is the practice of purchasing advertising placements on search-engine results pages (SERPs) when users search for particular terms. Unlike organic search engine optimisation, which seeks unpaid visibility through relevance, authority and technical quality, paid search provides immediate but budget-dependent access to potential customers. It is commonly delivered through auction-based platforms such as Google Ads and Microsoft Advertising.

A paid search campaign is not simply a collection of advertisements. It is an integrated system that connects business objectives, search intent, keywords, ad messages, landing pages, audiences, bids, budgets and conversion measurement. Poor performance may result from any weak link in this system. For example, a highly relevant keyword can still generate poor results if its landing page is slow, its conversion tracking is incorrect, or its bid is inappropriate for the commercial value of the traffic.

Paid search therefore requires both operational competence and marketing judgement. Practitioners must understand how auctions work, interpret performance data, protect users from misleading claims, and allocate limited budgets to activities that support organisational objectives.

Learning Objectives

By the end of this chapter, you should be able to:

• design a logical paid-search account and campaign structure;
• distinguish keyword match types and apply negative keywords;
• create relevant ads and use appropriate ad assets;
• select audience-targeting and bidding approaches;
• calculate and interpret core paid-search metrics;
• configure conversion tracking and evaluate landing pages;
• use remarketing and experiments responsibly; and
• identify ethical, legal and practical limitations of paid search.

1. The Paid Search Auction

When a user submits a search, an advertising platform identifies eligible advertisements and conducts an auction. Eligibility may depend on factors such as keyword or query relevance, geographic location, device, audience, budget status, policy compliance and targeting settings. The advertiser does not usually buy a permanent position. Instead, the advertiser participates in repeated auctions, often with a different competitive environment for every query.

A simplified account of ad ranking is:

This is a conceptual model rather than a complete public formula. Platforms use proprietary systems and may incorporate expected click-through rate, ad relevance, landing-page experience, auction competition, device, location and other contextual signals. A larger bid does not guarantee the highest position, and a high-quality advertisement may obtain strong visibility without being the highest bidder.

Advertisers commonly pay per click (CPC), although some campaigns use automated bidding toward conversions, conversion value, impressions or other outcomes. The amount charged can differ from the maximum bid because auction pricing depends on competitors and platform rules.

The marketing role of paid search

Paid search is especially useful when:

• users actively express a need through a query;
• the organisation can respond with a relevant product, service or piece of information;
• results can be measured through a defined conversion; and
• the expected value of a conversion justifies media and operating costs.

It is less suitable when the audience is unaware of a problem and does not search for it, when the product requires extensive education before demand exists, or when tracking and fulfilment are unreliable.

2. Account, Campaign and Ad-Group Structure

A platform account contains campaigns, which contain ad groups, keywords or targeting criteria, ads and assets. A sound structure reflects meaningful differences in budget, geography, language, objectives, product category or bidding strategy.

A typical structure might be:

Campaigns should not be divided merely to create a large number of labels. Excessive fragmentation can reduce data available to automated bidding and make management difficult. Conversely, combining unrelated products or intentions can produce weak ad relevance, inappropriate budgets and misleading reporting.

A practical principle is one ad group, one closely related intent. Keywords such as “buy hiking boots” and “hiking boots repair” may belong in separate groups because their users require different messages and landing pages. Brand, non-brand, competitor and product campaigns are often separated because they have different levels of intent, costs and strategic roles.

3. Search Intent, Keywords and Match Types

Keywords are not exact descriptions of what every user will type. They are targeting signals interpreted by the advertising platform. The central planning task is to connect a keyword theme with search intent.

Useful intent categories include:

• Informational: seeking knowledge, such as “how to choose a tent”.
• Commercial investigation: comparing solutions, such as “best waterproof tents”.
• Transactional: showing readiness to act, such as “buy waterproof tent”.
• Navigational or brand: seeking a known organisation or product.

Match types

Platform terminology changes over time, but the main concepts remain:

• Broad match: allows the platform greater freedom to match related searches. It can discover demand but requires strong monitoring, conversion data and negative-keyword controls.
• Phrase match: generally preserves the meaning or intent of a phrase while allowing variations before or after it. It provides more control than broad match but is not necessarily a literal word-order rule.
• Exact match: offers the greatest keyword-level control, although close variants and semantically similar searches may still be eligible under current platform rules.

Match types should be selected according to risk, data maturity and the advertiser’s ability to monitor search terms. A new account with a narrow budget may begin with tightly controlled high-intent themes. As reliable conversion data accumulates, broader matching can be tested to discover incremental demand.

Negative keywords

Negative keywords prevent advertisements from serving for unsuitable searches. Examples include “free”, “jobs”, “course”, “repair” or a competitor’s name when those searches do not fit the offer. Negative keywords should be based on business logic and evidence rather than assumptions.

A search-term review should examine:

1. whether the query matches the product or service;
2. whether its intent is commercially appropriate;
3. whether it produces conversions or valuable engagement; and
4. whether it should be excluded at campaign or ad-group level.

Campaign-level negatives are appropriate for terms irrelevant to all ad groups in that campaign. Ad-group negatives can prevent overlap between closely related groups.

4. Ad Creation and Assets

A strong search advertisement makes a clear promise that is supported by the landing page. It should identify the offer, communicate a meaningful benefit and provide a specific next step. Relevance is more valuable than exaggerated language.

Important message components include:

• the product or service and its category;
• a differentiating benefit, such as delivery speed or warranty;
• evidence, where permitted, such as ratings or accreditation;
• a call to action; and
• a qualification that prevents unsuitable clicks, such as “business plans” or “UK delivery”.

Ad assets—formerly called extensions in some platforms—can add structured information or alternative routes to action. Common examples include sitelinks, callouts, structured snippets, image assets, call assets, location assets, price assets and promotion assets. Assets do not guarantee a particular position, but they can improve usefulness and visibility when relevant.

Advertising claims must be accurate, current and supportable. Claims involving price, scarcity, health, finance, environmental performance or comparative superiority require particular care. Platform policy and national advertising law may prohibit misleading omissions, discriminatory targeting or deceptive urgency.

5. Audiences and Targeting

Search campaigns may use audience signals such as interests, purchase intent, customer lists, website visitors, demographic characteristics or first-party behavioural data. Audience settings can be used for observation—collecting performance data without restricting reach—or targeting—restricting delivery to selected users.

Audience targeting should complement, not replace, query relevance. A user who belongs to a target audience but searches for an unrelated topic should not automatically receive an irrelevant advertisement. Similarly, customer-list targeting requires a lawful basis, appropriate consent or another valid legal ground, secure data handling and compliance with platform requirements.

Remarketing reaches people who have previously interacted with a website, application or brand. It can support follow-up messages, product reminders or exclusions for existing customers. Frequency controls and membership-duration settings help reduce annoyance. Sensitive categories, including health, finance and children’s services, require especially cautious treatment because inferred personal circumstances may create privacy or discrimination risks.

6. Bidding, Budgets and Financial Metrics

A budget is a control on expected expenditure, not a guarantee of traffic or conversions. Daily budgets may be adjusted by platforms according to delivery rules, so practitioners should evaluate expenditure over the platform’s stated billing period rather than assuming every day is identical.

Common bidding approaches include:

• Manual CPC: the advertiser directly controls bids, subject to platform rules.
• Maximise clicks: seeks traffic within a budget, but traffic volume is not the same as business value.
• Maximise conversions: uses automation to seek conversions while spending within budget.
• Target CPA: seeks conversions around a target cost per acquisition.
• Maximise conversion value: prioritises reported value rather than simply counting conversions.
• Target ROAS: seeks a specified return on advertising spend where reliable conversion values are available.

Automated bidding requires sufficient, accurate and timely data. If conversion actions are duplicated, delayed, incorrectly attributed or too rare, automation may optimise toward misleading signals.

Core metrics

Let:

• I = impressions;
• C = clicks;
• S = advertising spend;
• V = recorded conversions; and
• R = conversion revenue or value.

The principal measures are:

CTR and conversion rate are percentages. Average CPC and CPA use the account’s currency. ROAS is a ratio; it may also be expressed as a percentage by multiplying by 100. These formulas assume that the underlying counts and values use the same reporting period, attribution settings and scope. They describe observed performance, not causal proof.

Worked example

An online retailer records 20,000 impressions, 800 clicks, advertising spend of £960, 32 purchases and £4,800 in attributed revenue.

CTR = (800 ÷ 20,000) × 100 = 4%.

Average CPC = £960 ÷ 800 = £1.20.

Conversion rate = (32 ÷ 800) × 100 = 4%.

CPA = £960 ÷ 32 = £30.

ROAS = £4,800 ÷ £960 = 5.0, or 500% when expressed as a percentage.

The retailer should not conclude that the campaign is profitable from ROAS alone. Product cost, fulfilment, tax, refunds and overheads must also be considered. If gross profit before advertising is £2,000, the campaign’s contribution after advertising is £1,040 before other costs.

7. Conversion Tracking and Attribution

A conversion is a valuable action defined by the organisation, such as a completed purchase, qualified lead, phone call, application submission or download. Tracking may use platform tags, analytics systems, server-side events, imported CRM outcomes or enhanced conversion mechanisms.

A robust implementation should specify:

1. the conversion definition and business owner;
2. the event or transaction trigger;
3. the value and currency, where relevant;
4. whether the action is primary for optimisation or secondary for observation;
5. deduplication rules; and
6. testing and quality-assurance procedures.

Tracking errors include counting page views as purchases, firing tags on refresh, losing transactions across domains, ignoring consent choices, or importing duplicate conversions. Test conversions should be clearly distinguished from genuine business outcomes.

Attribution assigns credit to marketing interactions. Last-click attribution is simple but may undervalue earlier interactions; data-driven models may be more informative but require sufficient data and remain dependent on modelling assumptions. Reported conversions are therefore estimates shaped by measurement design, privacy restrictions, browser behaviour and platform methodology. They should be compared with CRM, finance and controlled-experiment evidence where possible.

8. Landing Pages and Quality

A landing page should continue the promise made by the advertisement. Users should immediately understand what is offered, who it is for, what it costs or requires, and how to proceed. Relevant content, mobile usability, accessibility, clear navigation, trust signals and technical performance are essential.

Quality-related platform measures often incorporate expected engagement, ad relevance and landing-page experience. These measures are diagnostic rather than universal scores of marketing success. A high quality indicator cannot compensate for an unattractive offer, while a lower indicator may sometimes be acceptable when a highly valuable niche query has limited competition.

Landing-page evaluation should consider:

• message match between query, ad and page;
• page speed and stability;
• prominent and accessible calls to action;
• transparent price, delivery, privacy and cancellation information;
• form length and error handling;
• evidence of credibility; and
• accessibility for users with different abilities and devices.

9. Experiments and Optimisation

Optimisation is a disciplined process of changing one or more controllable factors, observing outcomes and deciding whether evidence supports adoption. Possible tests include landing-page layouts, calls to action, audience exclusions, match-type strategies, bidding targets and budget allocations.

A useful experiment states a hypothesis, success metric, decision rule, duration and eligible traffic. For example: “Replacing a long lead form with a shorter form will increase qualified leads without increasing cost per qualified lead.” The primary metric should be selected before viewing results. Short tests with small samples can produce unstable results, and simultaneous changes make causal interpretation difficult.

A practical optimisation cycle is:

1. verify tracking;
2. diagnose performance by query, device, location, audience and landing page;
3. identify a specific problem;
4. implement a controlled change;
5. monitor spend, volume and quality;
6. compare results using consistent attribution; and
7. document the decision and remaining uncertainty.

Illustrative Case: Northstar Skills

Background

Northstar Skills, a fictional training provider, offers evening digital-marketing certificates. Its initial search campaign used one ad group containing broad terms such as “marketing course”, “SEO training” and “free marketing lessons”. The campaign generated many clicks but few enrolments.

Problem and analysis

Search-term analysis showed that a large proportion of queries concerned free resources, university degrees and employment vacancies. The advertisement promised “Learn Marketing Fast”, while the landing page described a paid evening certificate. Tracking counted brochure downloads but not completed applications, so the bidding system optimised toward low-value actions.

Approach

Northstar separated campaigns for high-intent certificate searches and informational content. It added negatives such as “free”, “jobs” and “degree” where appropriate, created ad groups for SEO, paid search and social media, and wrote ads that stated the evening schedule, qualification level and price range. The landing page was revised to match those messages and include an application call to action.

The organisation configured separate primary conversions for completed applications and secondary conversions for brochure downloads. It began with controlled bidding while validating data, then tested automated conversion bidding after sufficient reliable application data accumulated. A remarketing audience received reminders about application deadlines, with frequency limits and privacy-compliant consent management.

Outcome

After the changes, traffic volume declined, but the proportion of relevant visitors and completed applications increased. Northstar judged the campaign using cost per completed application and enrolment quality rather than clicks alone. The case illustrates that reducing irrelevant traffic can improve business performance even when headline traffic metrics fall.

Learning points

The main lessons are to align structure with intent, measure the actual business outcome, use negative keywords systematically, and treat landing-page and tracking quality as part of campaign management rather than as separate technical tasks.

10. Governance, Ethics and Contemporary Practice

Paid search is subject to platform policies, consumer-protection law, privacy regulation, accessibility expectations and sector-specific rules. Organisations should maintain approval procedures for claims, document targeting logic, respect consent and deletion requests, and limit access to customer data.

Practitioners should avoid exploiting vulnerable users, creating discriminatory exclusions, concealing material conditions, or using sensitive inferences to personalise advertising. Automated systems can amplify historical bias or optimise for cheap but low-quality leads. Human review remains necessary, particularly in regulated sectors.

AI increasingly supports keyword expansion, copy generation, audience prediction, bidding and performance summaries. It can improve speed, but generated content may contain unsupported claims, duplicated messages or policy violations. Human specialists should verify factual accuracy, brand suitability, privacy implications, accessibility and performance objectives. Automation should be governed by clear conversion definitions, spending limits, change histories and escalation rules.

Best-Practice Checklist

• Begin with a measurable business objective.
• Separate campaigns only when budget, geography, product or strategy genuinely differs.
• Group keywords by close intent and provide a matching landing page.
• Review search terms and maintain negative keywords.
• Write specific, truthful ads and use relevant assets.
• Test conversion tracking before changing bids.
• Select bidding methods that match data quality and business goals.
• Evaluate CPA, conversion quality and profit—not clicks alone.
• Improve mobile experience, accessibility and message match.
• Run documented experiments and retain learning.
• Apply privacy, platform-policy and consumer-protection controls.

Chapter Summary

Paid search is an auction-based form of performance marketing in which advertisers seek visibility for relevant user searches. Effective campaigns connect account structure, search intent, match types, negative keywords, ad assets, audiences, bids, budgets, landing pages and conversion measurement. Match types influence control and discovery, while negative keywords reduce waste. Quality-related signals affect eligibility and cost but do not replace a valuable offer. Metrics such as CTR, CPC, conversion rate, CPA and ROAS are useful only when their definitions, time periods and attribution settings are consistent. Conversion tracking must measure genuine business outcomes, and automated bidding depends on reliable data. Landing-page relevance, experiments, privacy governance and ethical judgement are central to professional practice.

Key Terms

Ad asset: Additional information or functionality shown with an advertisement, such as a sitelink or call asset.

Ad group: A collection of closely related targeting criteria and advertisements.

Ad Rank: A platform’s auction-related assessment used to determine eligibility and position; its exact formula is proprietary.

Attribution: The assignment of credit for a conversion to one or more marketing interactions.

Broad match: A match approach allowing relatively wide semantic interpretation of a keyword.

Campaign: A management unit containing settings such as budget, location, network and bidding approach.

Conversion: A defined action that represents value to an organisation.

Conversion rate: Conversions divided by clicks, expressed as a percentage.

CPA: Advertising spend divided by recorded conversions.

CPC: Cost paid for a click, commonly measured as average spend divided by clicks.

CTR: Clicks divided by impressions, expressed as a percentage.

Exact match: A relatively controlled match approach that can still include permitted close variants.

Landing page: The webpage reached after an advertisement is clicked.

Negative keyword: A term used to exclude unsuitable searches.

Phrase match: A match approach that generally preserves the meaning of a keyword phrase while allowing variants.

Quality-related factors: Relevance and experience signals used by platforms in auction and diagnostic systems.

Remarketing: Advertising to people who previously interacted with a brand or digital property.

ROAS: Attributed conversion value divided by advertising spend.

Search intent: The user’s underlying purpose in submitting a query.

Review Questions

Short Answer Questions

1. What is the purpose of a negative keyword?
2. Distinguish a campaign from an ad group.
3. Define conversion rate and CPA.
4. Why is message match important on a landing page?
5. What is the difference between observation and targeting for an audience?

Long Answer Questions

1. Explain how account structure should reflect differences in product, geography, budget and user intent.
2. Discuss the advantages and limitations of broad, phrase and exact match approaches.
3. Explain why accurate conversion tracking is necessary for automated bidding.
4. Evaluate the role of landing-page experience in paid-search performance.
5. Discuss ethical and privacy risks associated with remarketing and audience targeting.

Analytical / Application Questions

1. A campaign receives 12,000 impressions, 600 clicks, £900 spend and 24 conversions worth £3,600. Calculate CTR, average CPC, conversion rate, CPA and ROAS. Interpret the results cautiously.
2. A retailer receives many clicks for searches containing “free”, “repair” and “jobs”. Design a search-term review and negative-keyword strategy, including possible campaign- or ad-group-level exclusions.
3. A lead-generation campaign has many brochure downloads but few qualified applications. Diagnose possible tracking, targeting, ad and landing-page problems, and propose a staged optimisation experiment.

Multiple Choice Questions with Answers

Question 1

Which account level normally controls a paid-search budget?

A. Keyword

B. Campaign

C. Headline

D. Search term

Correct Answer: B. Campaign

Question 2

What is the principal purpose of a negative keyword?

A. To increase every advertisement’s bid

B. To prevent ads showing for unsuitable searches

C. To create additional landing pages

D. To calculate conversion value

Correct Answer: B. To prevent ads showing for unsuitable searches

Question 3

A campaign records 500 clicks from 10,000 impressions. What is its CTR?

A. 0.5%

B. 2%

C. 5%

D. 20%

Correct Answer: C. 5%

Question 4

Which metric is calculated as advertising spend divided by conversions?

A. CTR

B. Average CPC

C. CPA

D. ROAS

Correct Answer: C. CPA

Question 5

Which bidding approach is designed to seek a specified return on advertising spend?

A. Target ROAS

B. Maximise clicks

C. Manual placement

D. Target impressions only

Correct Answer: A. Target ROAS

Question 6

Why should keywords in one ad group usually have closely related intent?

A. To ensure every click is free

B. To support relevant ads and landing pages

C. To eliminate the need for tracking

D. To guarantee first position

Correct Answer: B. To support relevant ads and landing pages

Question 7

Which conversion is generally the strongest primary optimisation signal for a training provider?

A. A page view

B. A browser refresh

C. A completed application

D. An impression

Correct Answer: C. A completed application

Question 8

What does ROAS measure?

A. Revenue or conversion value relative to advertising spend

B. Impressions relative to clicks

C. Clicks relative to conversions

D. Profit after every business cost

Correct Answer: A. Revenue or conversion value relative to advertising spend

Question 9

What is a key risk of poorly governed automated bidding?

A. It cannot use any historical data

B. It may optimise toward inaccurate or low-value conversions

C. It always stops advertisements immediately

D. It removes all audience controls

Correct Answer: B. It may optimise toward inaccurate or low-value conversions

Question 10

Which practice is most appropriate for a paid-search experiment?

A. Change tracking, bids and landing page simultaneously without documentation

B. Choose a hypothesis and primary metric before evaluating results

C. Judge success only by impressions

D. End the test whenever one day performs poorly

Correct Answer: B. Choose a hypothesis and primary metric before evaluating results