Quality, Fraud, Brand Safety and Transparency

Introduction

Programmatic advertising creates an efficient market in which advertisers, agencies, publishers, exchanges, supply-side platforms (SSPs), demand-side platforms (DSPs), verification vendors and measurement providers interact through automated auctions. Efficiency, however, does not guarantee quality. An impression may be technically delivered but have little commercial value because it was generated by a bot, served in an invisible location, misrepresented as premium inventory, surrounded by excessive advertising, or displayed beside harmful content.

Media quality is therefore a strategic issue rather than merely an operational concern. Poor-quality inventory wastes budgets, distorts campaign measurement, exposes brands to reputational risk and weakens trust in the advertising ecosystem. Quality management requires the coordinated use of technical controls, independent verification, contractual requirements, supply-chain transparency and informed human judgement.

This chapter examines invalid traffic, non-human activity, domain spoofing, made-for-advertising (MFA) inventory, ad clutter, unsuitable content and the principal transparency standards used to manage them. It also considers supply-path optimisation (SPO), contractual governance, privacy, regulation and the limitations of automated detection.

Learning Objectives

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

1. Distinguish legitimate, invalid and fraudulent advertising traffic.
2. Explain how bots, domain spoofing, MFA sites and ad clutter reduce media quality.
3. Apply verification concepts such as viewability, invalid-traffic detection, contextual suitability and brand-safety measurement.
4. Explain the functions of ads.txt, app-ads.txt, sellers.json and SupplyChain Object information.
5. Evaluate supply paths using transparency, cost, performance and risk criteria.
6. Design contractual and operational controls for safer programmatic buying.
7. Critically assess the ethical, regulatory and methodological limits of quality controls.

1. Media Quality as a Multi-Dimensional Construct

The term quality does not have one universal meaning. It is a bundle of attributes evaluated in relation to campaign objectives. A direct-response advertiser may prioritise measurable conversions and fraud resistance, whereas a brand advertiser may prioritise attention, suitability, reach and premium editorial environments.

A useful quality framework contains six dimensions:

These dimensions are related but not interchangeable. A viewable impression can still be fraudulent; a genuine publisher can still provide unsuitable content; and a brand-safe placement can still be commercially ineffective. Consequently, quality should be assessed as a portfolio of risks rather than reduced to one score.

2. Invalid Traffic and Ad Fraud

2.1 Definitions

Invalid traffic (IVT) is traffic that does not meet the requirements for legitimate advertising measurement or delivery. It includes both accidental and deliberately manipulated activity. Industry measurement systems commonly distinguish:

• General invalid traffic (GIVT): traffic identifiable through routine checks, such as known data-centre addresses, automated crawlers that identify themselves, malformed user agents or obvious irregularities.
• Sophisticated invalid traffic (SIVT): traffic requiring advanced analysis, corroboration or human investigation. Examples include coordinated botnets, hijacked devices, hidden automation and traffic designed to mimic human behaviour.

The boundary between fraud and invalid traffic is not always identical. Invalid traffic is a measurement and eligibility category; fraud is a broader description of deceptive or criminal conduct. Some invalid traffic may result from non-malicious testing, accessibility tools or legitimate crawlers, while some fraud produces traffic that initially appears valid.

2.2 Common mechanisms

Bots may request pages, load ads, generate clicks or imitate browsing sessions. Common mechanisms include:

• data-centre bots and headless browsers;
• malware-infected personal devices;
• click farms and low-paid human activity;
• domain or app impersonation;
• cookie stuffing and attribution manipulation;
• ad stacking, in which several ads occupy one location;
• pixel stuffing, in which an ad is rendered at an effectively invisible size;
• auto-refresh and forced refresh practices;
• injected ads placed into environments without the publisher’s authorisation.

Fraudsters increasingly use residential proxies, automated browsing frameworks and machine-learning systems to imitate normal session patterns. Detection therefore requires multiple signals rather than reliance on a single IP-address list.

2.3 Measuring invalid traffic

A simple campaign-level invalid-traffic rate is:

Here, invalid impressions are impressions classified as invalid under the applicable measurement methodology, and measured impressions are the impressions eligible for that analysis. The formula assumes that both quantities use the same scope, time period, platform and counting rules. It should not be compared across vendors unless their definitions and measurement coverage are comparable.

For example, if 80,000 of 2,000,000 measured impressions are classified as invalid, the IVT rate is 4%. This does not mean that exactly 4% of the advertiser’s total media exposure was fraudulent: unmeasured impressions, classification error and differences in coverage may affect the estimate.

A related metric is the valid-impression rate:

This relationship is meaningful only when invalid and valid categories are mutually exclusive and collectively exhaustive within the defined measurement population.

2.4 Detection signals and limitations

Verification systems may examine IP reputation, user-agent consistency, device characteristics, browsing sequences, mouse or touch events, server logs, timing patterns, geographic anomalies, conversion quality and publisher-level behaviour. A signal becomes more reliable when corroborated by independent evidence.

No detector is perfect. False positives can exclude genuine users, particularly users behind corporate networks, privacy tools, mobile carriers or shared devices. False negatives allow sophisticated fraud to pass. Buyers should therefore ask vendors about methodology, accreditation, sampling, measurement coverage, confidence intervals where available, invalidation rules and dispute procedures.

3. Viewability, Exposure and Attention

Viewability concerns whether an ad had an opportunity to be seen, not whether a person actually noticed or remembered it. Common industry standards use minimum proportions of pixels in view for a minimum duration, with different requirements for display and video formats. The applicable standard should always be identified because definitions vary by format, environment and measurement provider.

A generic viewability rate can be expressed as:

The denominator is measurable impressions, not necessarily all served impressions. If 600,000 of 800,000 measurable impressions satisfy the selected viewability criterion, viewability is 75%. Unmeasurable impressions should not automatically be treated as viewable or non-viewable.

Viewability also has limitations. An impression may meet a pixel-and-time threshold while appearing below the fold, being obscured briefly, or receiving no meaningful attention. Attention metrics, user studies and outcomes such as brand lift may supplement viewability, but they should not be treated as interchangeable measures.

4. Domain Spoofing and Supply Authenticity

Domain spoofing occurs when an impression is represented as coming from a more valuable publisher or app than the environment that actually delivered it. In a classic form, a counterfeit site labels its inventory as a well-known news domain. In other cases, an intermediary alters metadata, an app is misidentified, or an unauthorised reseller represents inventory as direct access.

Spoofing harms buyers through inflated prices and misleading audience assumptions. It harms legitimate publishers by diverting demand and weakening their pricing power. It also undermines measurement because placement, audience and contextual data become unreliable.

4.1 ads.txt and app-ads.txt

The ads.txt initiative allows a publisher to declare authorised digital advertising sellers in a publicly accessible text file associated with its domain. A typical record identifies an advertising system, the publisher’s seller account identifier and the relationship type, commonly direct or reseller, with an optional certification authority identifier.

app-ads.txt extends the principle to mobile applications. Because the app itself may not have a conventional web domain, the authorised-seller file is associated with the developer’s claimed web domain and referenced through app-store or application metadata mechanisms.

These files answer an important question: Which sellers has the publisher authorised? They do not prove that every impression is viewable, human, brand-safe or accurately measured. They can also become stale, contain errors or omit legitimate commercial arrangements. Buyers should therefore check file recency, seller status, account identifiers and consistency with the transaction.

4.2 sellers.json and SupplyChain Object

Where ads.txt and app-ads.txt provide a publisher-side declaration, sellers.json provides a buyer-facing view of entities participating in an exchange or SSP’s supply chain. It can identify the seller account, the organisation name and whether the entity is a publisher, intermediary or other participant, subject to the standard’s implementation and privacy provisions.

The SupplyChain Object (often called schain) travels with bid-request information and records the entities through which a transaction has passed. It assists buyers in tracing the supply path and comparing declared intermediaries with the sellers.json record.

These mechanisms are complementary:

They improve transparency but do not create complete transparency. Resellers may be legitimate, ownership structures may be complex, and technical declarations may not match actual operational behaviour. Verification, contractual rights and log-level analysis remain necessary.

5. Made-for-Advertising Inventory

MFA inventory is designed primarily to generate advertising revenue rather than to provide substantial editorial, informational or user value. Typical characteristics may include high ad density, repeated or thin content, aggressive refresh, extensive use of low-cost traffic, click-oriented page design and a large number of intermediaries.

MFA is not identical to fraud. A real human may visit an MFA site, and the site may technically comply with trafficking rules. The concern is economic and experiential: the buyer may pay for impressions that generate little attention, low-quality outcomes or reputational value. MFA classification is also contestable. A narrow information site, content farm, comparison platform or user-generated forum may be valuable for one advertiser and undesirable for another.

Assessment should combine:

• content originality and usefulness;
• repeat-visit and engagement quality;
• ad-to-content balance;
• refresh and auction frequency;
• traffic acquisition sources;
• placement visibility and user experience;
• conversion and incremental-outcome evidence;
• number and role of intermediaries.

A blanket exclusion may reduce reach or remove effective niche audiences. A more sophisticated approach applies tiers: approved premium environments, monitored environments, restricted environments and excluded environments.

6. Ad Clutter and User Experience

Ad clutter refers to excessive, intrusive or poorly arranged advertising that competes for attention and degrades the user experience. It may involve numerous simultaneous units, repeated creative, sticky placements, pop-ups, auto-play video, slow-loading scripts or ads that obscure content.

Clutter can reduce the effective value of an impression even when it is viewable. It may increase accidental clicks, cause users to abandon a page, encourage ad blocking and damage perceptions of both publisher and advertiser. It also complicates attribution: a click produced by accidental contact is not equivalent to deliberate interest.

Buyers should monitor ad density, simultaneous advertisements, refresh intervals, load latency, format compliance and frequency across devices. Contracts can specify maximum ad loads, prohibited formats, user-consent requirements and remedies for non-compliance.

7. Brand Safety and Brand Suitability

7.1 From safety to suitability

Brand safety traditionally refers to avoiding clearly harmful or unacceptable environments, such as extremist propaganda, child sexual abuse material, terrorist content or illegal activity. Brand suitability is broader and more contextual. It asks whether an environment is appropriate for a particular brand, campaign, audience and social moment.

For example, a public-health advertiser may accept responsible reporting about a disease but reject graphic imagery. A luxury brand may prefer an environment with a specific tone and editorial quality even though other environments are not unsafe in a narrow sense.

7.2 Risk categories and controls

Typical controls include:

1. Keyword and phrase controls: useful for obvious terms but vulnerable to false positives, negation errors and contextual misunderstanding.
2. Category exclusions: block broad topics such as violence, tragedy, adult content or political controversy.
3. Contextual classification: analyses page meaning, entities, sentiment and semantic relationships rather than isolated words.
4. URL and app lists: provide precise allowlists and blocklists but require maintenance.
5. Pre-bid filters: prevent bids on inventory classified as unsuitable before the auction.
6. Post-bid verification: measures the actual delivered environment and identifies leakage or misclassification.
7. Human escalation: reviews ambiguous or high-impact cases.

A strong programme uses a written suitability policy. It defines unacceptable, restricted, monitored and preferred environments; sets escalation thresholds; identifies who makes decisions; and documents exceptions. Overly broad blocking can create social and commercial bias by excluding legitimate journalism, minority voices or controversial but important subjects.

8. Verification and Measurement Architecture

Independent verification is usually supplied through tags, SDKs, server integrations, log analysis or combinations of these methods. A mature architecture separates three questions:

• Was the impression valid?
• Was it delivered in a measurable and viewable manner?
• Was the environment suitable and the transaction authentic?

Figure: Quality controls operate before, during and after programmatic delivery.

Verification reports should be interpreted with attention to methodology. Important questions include whether measurement is census-based or sampled, whether in-app and connected-TV environments are covered, how cross-device duplication is handled, which countries and browsers are measurable, and how discrepancies are reconciled.

9. Supply-Path Optimisation

Supply-path optimisation is the systematic selection of efficient, transparent and reliable routes from advertiser demand to publisher inventory. The cheapest path is not necessarily the best path. A path should be evaluated using a balanced scorecard including:

• authorised seller status;
• number of intermediaries;
• auction duplication;
• bid and win rates;
• take rates and total cost;
• latency and timeout performance;
• IVT and viewability;
• brand-suitability performance;
• log-level transparency;
• data quality and privacy compliance;
• publisher value and commercial relationships.

A useful economic measure is the effective cost per thousand valid, viewable impressions:

If spend is measured in currency and impressions are counts, the result is currency per thousand valid and viewable impressions. This measure is meaningful only when validity and viewability are measured consistently and the campaign’s objective supports their use. It does not measure attention, conversion quality or incremental sales.

Suppose Path A costs £12,000 for 1,000,000 impressions. Its valid rate is 98% and viewability among measurable impressions is 70%. Assuming the rates apply sequentially to the same impression base, valid and viewable impressions are 1,000,000 × 0.98 × 0.70 = 686,000. Thus, eCPM_VV = (£12,000 ÷ 686,000) × 1000 ≈ £17.49. A cheaper nominal CPM may therefore be more expensive per quality-adjusted impression than a premium route.

The sequential assumption is a simplification. In real analysis, validity and viewability may be correlated, and the correct calculation should use impression-level records identifying impressions that satisfy both conditions.

10. Contractual and Operational Controls

Technical standards are effective only when supported by governance. Contracts and insertion orders should specify:

• permitted inventory sources and authorised sellers;
• definitions of IVT, viewability and suitability;
• verification vendor, methodology and reporting frequency;
• make-good, refund or credit rules;
• audit and log-access rights;
• data-protection responsibilities;
• prohibited traffic-generation practices;
• creative and landing-page requirements;
• incident notification timelines;
• subcontracting and reseller disclosure;
• suspension and termination rights;
• dispute-resolution procedures.

A buyer should avoid vague promises such as “premium, brand-safe inventory” without operational definitions. Conversely, contracts should not demand impossible guarantees. Measurement uncertainty, platform constraints and evolving threats should be acknowledged through review provisions and evidence-based thresholds.

An operating control cycle can include:

1. Establish campaign risk appetite and suitability categories.
2. Approve vendors and verification methodology.
3. Validate domains, apps, sellers.json and SupplyChain information.
4. Apply pre-bid exclusions and frequency controls.
5. Monitor delivery daily or at an appropriate reporting interval.
6. Investigate anomalies using log-level and contextual evidence.
7. Reconcile invoices against verified delivery.
8. Apply remediation and update lists, contracts and buying rules.

11. Applied Case: The Northstar Financial Campaign

The following is a fictional illustrative case.

Background

Northstar Financial, a regulated financial-services advertiser, launched a national video and display campaign through three DSPs. The campaign required trusted news, business and lifestyle environments, low IVT and strict exclusion of extremist, adult and graphic-content categories.

Problem

After two weeks, the campaign achieved strong reach but weak post-view engagement. One DSP reported a 1.5% IVT rate and 68% viewability; a third-party verification report identified 7% IVT on a subset of inventory. Several domains appeared in the campaign report with unfamiliar names, and the finance team questioned why the campaign had used more than 40,000 publisher URLs.

Analysis

The advertiser created a path-level inventory table. It found that one SSP supplied multiple reseller routes to the same publishers. Some bid requests contained incomplete SupplyChain information. Several sites had high ad density, frequent refresh and little original content. A keyword-only safety filter had also allowed a page containing graphic imagery because the relevant term was absent from the URL and headline.

The team then compared ads.txt declarations, app-ads.txt records where relevant, sellers.json entries and transaction-level SupplyChain data. It discovered that a portion of the inventory was authorised for resale but passed through four intermediaries. Another portion was not clearly matched to an authorised seller account. The discrepancy between the vendor’s IVT report and the independent report resulted partly from different measurement coverage and partly from the exclusion of in-app traffic from the first report.

Approach and solution

Northstar adopted four measures:

1. It required complete SupplyChain information and rejected materially incomplete records.
2. It reduced duplicate and opaque supply paths, retaining two audited routes with direct publisher relationships.
3. It combined contextual classification, category exclusions, URL lists and human review for sensitive content.
4. It revised the contract to define measurement populations, require incident reporting and permit credits for verified invalid delivery.

The campaign team also introduced a quality-adjusted dashboard showing spend, valid impressions, measurable impressions, viewable impressions, suitable impressions and eCPM for impressions satisfying the required criteria.

Outcome

Nominal reach decreased, but the share of spend on transparent paths increased. IVT fell, the number of intermediary hops declined and the rate of unsuitable placements was reduced. The campaign’s reported post-view engagement improved, although the team correctly avoided claiming that quality controls alone caused the improvement. A controlled comparison was recommended for future campaigns.

Learning points

The case demonstrates that reach is not a sufficient success criterion, vendor reports are not automatically comparable, and transparency standards must be interpreted together. It also shows why safety, authenticity, viewability and economic efficiency should be monitored as distinct but connected dimensions.

12. Ethical, Legal and Strategic Considerations

Quality controls affect users, publishers and society. Excessive tracking can create privacy risks, while aggressive fraud detection can discriminate against users of privacy-enhancing technologies or shared networks. Contextual controls may suppress legitimate news and political speech. Publisher exclusions can concentrate advertising revenue among large platforms and weaken independent media.

Organisations must comply with applicable privacy, consumer-protection, advertising and online-safety laws. Consent, legitimate-interest assessments, data minimisation, retention limits and vendor due diligence should be considered where personal data or device identifiers are used. Regulatory requirements differ by jurisdiction; this chapter does not replace legal advice.

AI can improve semantic classification, anomaly detection, creative analysis and supply-path discovery. Yet AI systems may reproduce classification bias, misunderstand satire, over-block minority language content or become targets for adversarial manipulation. Human governance remains necessary for policy design, appeals, high-impact decisions and model monitoring.

13. Contemporary Developments and Best Practice

The ecosystem is moving toward greater supply-chain accountability, authenticated inventory, privacy-preserving measurement and more rigorous attention to publisher economics. Buyers increasingly seek direct publisher deals, curated marketplaces and outcome-based evaluation rather than maximum automated reach. Connected television, retail media, gaming and digital audio introduce distinct measurement and authentication challenges; web standards alone do not solve them.

Best practice includes:

• define quality in relation to campaign objectives;
• use independent verification and understand its methodology;
• combine pre-bid prevention with post-bid measurement;
• validate ads.txt, app-ads.txt, sellers.json and SupplyChain information;
• evaluate paths on quality-adjusted value, not nominal CPM alone;
• maintain explicit suitability policies and escalation procedures;
• monitor ad experience, latency and clutter;
• use contracts with measurable definitions and remedies;
• review false positives and excluded content for fairness;
• retain human oversight over ambiguous and high-impact cases;
• conduct periodic audits because sellers, domains, threats and standards change.

Chapter Summary

Programmatic media quality is multidimensional. Validity concerns whether traffic is legitimate; viewability concerns whether an ad had an opportunity to be seen; authenticity concerns whether the represented publisher or app and selling entities are genuine; suitability concerns the relationship between content and brand risk; and accountability concerns evidence, governance and enforceability.

Invalid traffic includes general and sophisticated forms, while fraud may involve bots, click manipulation, spoofing, ad stacking and other deceptive practices. Domain spoofing and unauthorised resale can be mitigated through ads.txt, app-ads.txt, sellers.json and the SupplyChain Object, although none of these tools independently proves quality. MFA inventory and ad clutter may be technically valid yet economically or experientially poor. Brand safety should be expanded into brand suitability through contextual analysis, clear policies and human escalation.

Verification should operate before, during and after delivery. Supply-path optimisation evaluates transparency, cost, duplication, measurement, risk and publisher value. Quality-adjusted metrics such as eCPM for valid and viewable impressions can improve decisions, but they depend on consistent definitions and do not replace outcome measurement. Finally, effective quality management combines technology with contracts, audits, privacy safeguards, ethical judgement and continuous review.

Key Terms

Invalid traffic: Advertising traffic that fails defined requirements for legitimate measurement or delivery.

General invalid traffic: Invalid activity identifiable through routine, known or straightforward checks.

Sophisticated invalid traffic: Invalid activity requiring advanced analysis, corroboration or investigation.

Ad fraud: Deceptive activity intended to obtain improper advertising revenue or measurement benefit.

Viewability: The extent to which an ad had an opportunity to be seen under a specified standard.

Measurable impression: An impression for which the measurement system can determine the relevant viewability condition.

Domain spoofing: Misrepresenting the publisher, domain or app associated with an advertising impression.

MFA inventory: Inventory created primarily to monetise advertising rather than deliver substantial user value.

Ad clutter: Excessive or intrusive advertising that degrades user experience and attention.

Brand safety: Protection from clearly harmful, unacceptable or dangerous content environments.

Brand suitability: The contextual appropriateness of an environment for a specific brand and campaign.

ads.txt: A publisher-domain declaration of authorised digital advertising sellers.

app-ads.txt: An authorised-seller declaration for mobile application inventory.

sellers.json: A marketplace file identifying seller and intermediary entities.

SupplyChain Object: Transaction information describing entities involved in a supply path.

Supply-path optimisation: Selection and management of efficient, transparent and reliable routes to inventory.

Pre-bid filtering: Excluding unsuitable or risky opportunities before bidding.

Post-bid verification: Measuring the actual delivered impression after the auction.

Quality-adjusted eCPM: Cost per thousand impressions satisfying specified quality conditions.

False positive: A legitimate item incorrectly classified as invalid, unsafe or unsuitable.

Review Questions

Short Answer Questions

1. Distinguish general invalid traffic from sophisticated invalid traffic.
2. What is the principal purpose of ads.txt?
3. Why is viewability not equivalent to attention?
4. Explain one difference between brand safety and brand suitability.
5. Identify two limitations of sellers.json or the SupplyChain Object.

Long Answer Questions

1. Explain how bots, domain spoofing, MFA inventory and ad clutter can reduce the economic value of programmatic advertising.
2. Compare ads.txt, app-ads.txt, sellers.json and the SupplyChain Object, including the perspective and limitation of each.
3. Design a brand-suitability governance framework for a regulated advertiser.
4. Discuss the role of independent verification in measuring IVT, viewability and content suitability.
5. Explain how contractual controls and supply-path optimisation should work together.

Analytical / Application Questions

1. An advertiser spends £18,000 on 1,500,000 impressions. Independent measurement identifies a 4% IVT rate and 65% viewability among valid impressions. Calculate the approximate cost per thousand impressions that are both valid and viewable, state the assumptions, and explain the limitations of the result.
2. A DSP reports low IVT, but an independent vendor reports much higher IVT. Develop a diagnostic plan that examines definitions, measurement coverage, in-app traffic, sampling, log data and reconciliation procedures.
3. A premium publisher is authorised in ads.txt, but its impressions arrive through five intermediaries with incomplete SupplyChain information and high ad clutter. Should the buyer retain, renegotiate or exclude the path? Construct a decision using quality, cost, transparency, risk and publisher-value criteria.

Multiple Choice Questions with Answers

Question 1

Which statement best describes sophisticated invalid traffic?

A. Traffic identifiable through a standard known-bot list

B. Traffic requiring advanced analysis or investigation to identify

C. All traffic generated from mobile devices

D. Any impression that is not viewable

Correct Answer: B. Traffic requiring advanced analysis or investigation to identify

Question 2

What is the principal purpose of ads.txt?

A. To measure whether an ad was seen

B. To classify the sentiment of page content

C. To declare publishers’ authorised advertising sellers

D. To calculate campaign conversion rate

Correct Answer: C. To declare publishers’ authorised advertising sellers

Question 3

Which tool primarily provides a buyer-facing record of seller and intermediary identities within an SSP or exchange?

A. sellers.json

B. app-ads.txt

C. A viewability tag

D. A frequency cap

Correct Answer: A. sellers.json

Question 4

Why should measurable impressions be used as the denominator in a standard viewability-rate calculation?

A. Unmeasurable impressions are automatically fraudulent

B. The measurement system cannot determine viewability for unmeasurable impressions

C. Measurable impressions always generate conversions

D. It removes all sampling error

Correct Answer: B. The measurement system cannot determine viewability for unmeasurable impressions

Question 5

Which example is most characteristic of domain spoofing?

A. A legitimate publisher reducing its ad load

B. A buyer excluding a sensitive content category

C. A counterfeit site representing its inventory as a famous news domain

D. A user viewing an ad below the fold

Correct Answer: C. A counterfeit site representing its inventory as a famous news domain

Question 6

What is a central concern associated with made-for-advertising inventory?

A. It is always generated by bots

B. It is designed primarily to provide public-interest journalism

C. It may provide little user or editorial value relative to the advertising it displays

D. It cannot contain a valid ads.txt file

Correct Answer: C. It may provide little user or editorial value relative to the advertising it displays

Question 7

Which approach is generally strongest for brand-suitability governance?

A. Using one isolated keyword list for every brand

B. Combining contextual analysis, exclusions, verification and human escalation

C. Buying all available inventory to maximise reach

D. Treating every controversial topic as unsafe

Correct Answer: B. Combining contextual analysis, exclusions, verification and human escalation

Question 8

An advertiser spends £10,000 on 1,000,000 impressions. If 2% are invalid and viewability among valid impressions is 50%, what is the approximate eCPM for valid and viewable impressions?

A. £10.00

B. £20.41

C. £50.00

D. £5.10

Correct Answer: B. £20.41

Question 9

What is the primary goal of supply-path optimisation?

A. To select the path with the largest number of intermediaries

B. To maximise nominal impressions regardless of quality

C. To select transparent, efficient and reliable routes to inventory

D. To eliminate all publisher-direct relationships

Correct Answer: C. To select transparent, efficient and reliable routes to inventory

Question 10

Which statement about verification systems is most accurate?

A. They eliminate all false positives and false negatives

B. Their results are comparable without regard to methodology

C. They should be evaluated for coverage, definitions and measurement methods

D. They make contractual controls unnecessary

Correct Answer: C. They should be evaluated for coverage, definitions and measurement methods