Google Display: assets, placements and lead quality

A display ad reaches someone while they read a website or use an app. That person was not necessarily searching for your product, and clicking an image is not a promise to become a customer. An affiliate therefore needs more than cheap traffic: the creative promise, actual placements and leads accepted by the advertiser must all be checked.

This guide builds a separate Google Ads Display campaign for ordinary inventory-management software in the United States. You will prepare assets, an eligibility brief, a placement record and two financial examples. It is not a Search, Performance Max or returning-visitor campaign tutorial. All examples are fictional, and no such software offer is promised in AffiliateCore. The amounts are teaching assumptions, not Google prices or earnings forecasts.

Establish US scope and account eligibility before spending

Google's official location target documentation includes US territories among its supported examples. Both scenarios use the US market and an English-speaking small-business audience. That establishes the targeting context, not permission for every reader to open an account, fund advertising or promote any product. Sources were checked on 18 October 2026.

Before funding, separately verify the account country and owner, payment method, required verification, seller's service area and the affiliate programme's permission for Display traffic. This guide does not treat Russia as an available buying or delivery market. Do not use a false address, borrowed account or disguised destination. Without confirmed personal eligibility, prepare the materials without launching.

Amounts in this version are in USD and remain hypothetical. Other translations use independent local-currency assumptions, not exchange-rate equivalents. All seven versions retain the United States as their market. Actual billing currency follows the account and agreement. The working ad and destination must make sense to the intended audience; translated explanations elsewhere do not automatically localise a US campaign for another country.

Start by matching the task to an eligible offer on our AffiliateCore platform. You can prepare a publisher profile and describe the proposed source. An account does not guarantee an appropriate offer, approval or sales. If no offer fits, keep the test brief or consider a direct programme that explicitly permits display advertising. AffiliateCore does not replace the Google advertising account.

Inventory, audience and outcome are different things

Google's Display Network description includes websites, apps and Google-owned properties. These provide different interaction contexts. Someone reading a business article and someone accidentally tapping an in-app banner may both generate a click but behave differently afterwards. Conversely, an app name alone does not establish poor traffic.

Our hypothesis is that a clear demonstration of reconciling two stock counts attracts a small US business seeking inventory help. The payable outcome is a new demonstration request accepted under agreed criteria: supported country, relevant business task and confirmed interest in discussing the software. Downloading a sample, viewing a page and submitting a form are not the same as an accepted lead.

Prepare a brief recording product limits, US service territory, permitted source, your page domain, commission, duplicate rules, attribution window, review timing, reversal reasons, media and non-media budgets, and the person who can stop spending. Version each condition. Discussion of next month's sales does not rewrite this cohort's acceptance rules after the event.

This is prospecting for new audiences. It does not use customer uploads, owned contact lists or remarketing. Those require separate permissions, legal grounds and configuration. Adding them automatically to a small test for more signals changes both the question being tested and the responsibilities for data handling.

Build assets that remain accurate in different combinations

Example short headline: Stock counts without guesswork. Another: Compare stock records. Long headline: See how a small team checks stock differences before choosing software. Description: View a sample workflow, check requirements and request a product demonstration. These avoid unsupported promises of savings, profit or automatic error correction.

Create your own landscape and square illustrations showing three fictional products in an illustrative interface, without customer data. Use a logo only with permission. The official responsive display ad instructions cover asset upload and previewing. Check each possible combination: a headline should remain accurate if its description is absent or an image is cropped.

The page must deliver an independent benefit: show how expected and counted stock are compared. Product A expects 25 units but has 23, a difference of −2; B has 10 against 10, difference zero; C has 8 against 9, difference +1. These are teaching data, not a real product screenshot. Explain unit matching, export dates and duplicate rows. Do not require readers to upload their warehouse data to see the example.

After the demonstration, state the software's limits, preparation requirements and situations where a product demo is unsuitable. Clearly disclose that the article contains an affiliate link and that its author may receive a commission. Use a concrete button such as “Check the supplier's demonstration terms”. Do not impersonate the developer or an independent auditor. Readers should know who publishes the guide and who receives the request.

Google's destination requirements reject destinations made only to send users elsewhere. Original usefulness is therefore substantive, not a token excuse for another step. The ad's destination is your useful page; an affiliate visit is a voluntary next action, not a hidden cross-domain redirect.

Prepare the campaign in order

Confirm permissions. Keep the source, product and territory approvals. Dashboard access does not replace them. Assign someone responsible for pausing expenditure.
Select Display. Record campaign type, goal, language and US geography. Do not mix Search, Performance Max and Display into one experiment. Verify available controls in the current interface rather than copying an old screenshot.
Review geographic mode. This task needs people in the supported country, not merely interested in it. Google's location explanation distinguishes these signals without promising perfect accuracy. Validate requests using permitted information, not covert profiling.
Record targeting choices. Optimized targeting can go beyond selected signals. This bounded teaching test turns it off and records manual criteria; that is an experimental choice, not universal advice. With expansion enabled, do not describe delivery as a strictly manual sample.
Upload and review assets. Check combinations, mobile layout, final URL, imagery, identity and disclosure. Correct the reason for rejection rather than circumventing review.
Test measurement. Exclude authorised technical events from payment. Do not submit a payable fake request or make a purchase to manufacture commission. Match identifiers and times at both ends.

An average daily budget is not a hard cap on every day's expenditure. Google documents separate spending limits. Review actual charges and keep a reserve within your overall limit; pause before exhausting it, allowing for reporting delay. The scenario costs are final reconciled amounts, not a promise of an exact automatic test-budget cap.

Link spending to accepted requests

Record click date, campaign, asset version, device and known placement group, then submission, initial decision and final status. Use pseudonymous identifiers. Keep phone numbers, email addresses and stock information out of URLs and sub-IDs. Validate parameter support in both systems: a Google macro does not automatically become an AffiliateCore parameter.

The examples use a hypothetical direct programme and an external advertiser acceptance ledger. A demo request is NOT a confirmed built-in payable AffiliateCore event and cannot be relabelled reg or ftd. An AffiliateCore route is appropriate only after verifying an actually supported payable event, offer terms and integration. Otherwise keep the pilot external or prepared without launching through AffiliateCore. No automatic import of accepted results into Google is promised.

The Where ads showed report helps inspect placements but is neither exhaustive nor a billing substitute. Do not distribute unknown traffic among domains you prefer. The examples assume complete matching of two aggregate groups; a real report should retain an unmatched category and explain its share.

Both pilots buy traffic for 14 days and reconcile 21 days after the final click, assuming all contractual windows have closed. These are teaching terms, not Google or AffiliateCore deadlines. Late leads remain attached to their original click cohort. An accepted lead is not yet a sale, and earned commission is not necessarily money received.

Scenario one: demonstrating a task produces a modest positive result

A hypothetical supplier approves the US market, source and 28 USD commission per final accepted lead. The affiliate prepares the images, stock example and voluntary demonstration link described above. Fourteen days produce 120,000 impressions, 2,400 paid clicks, 1,920 page loads and 72 forms. These fictional observations are not CTR or conversion benchmarks.

Initial review excludes 12 repeat requests and six outside the agreed business task. That leaves 72 − 12 − 6 = 54 provisional acceptances. Reconciliation reverses nine qualification errors: three requests without demonstration intent and six previously missed duplicates. Final totals are 45 accepted, 18 initially rejected and nine reversed; each of the 72 forms appears exactly once.

Media: 2,400 × 0.3 USD = 720 USD.
Content: six hours × 20 USD = 120 USD; images: four hours × 20 USD = 80 USD.
Tracking: three hours × 20 USD = 60 USD; checks and reconciliation: two hours × 20 USD = 40 USD.
Allocated tools and hosting 20 USD; non-recoverable buying fees and taxes 40 USD.
Total: 720 USD + 120 USD + 80 USD + 60 USD + 40 USD + 20 USD + 40 USD = 1,080 USD.

Revenue is 45 × 28 USD = 1,260 USD; the result before individual income tax is 180 USD. Full accepted-lead cost is 1,080 USD / 45 = 24 USD, with a 16.67% return on expenditure. Final acceptance is 45 / 72 = 62.5%; reversals among provisional results are 9 / 54 = 16.67%. CTR is 2,400 / 120,000 = 2%, while paid-click-to-accepted-lead conversion is 45 / 2,400 = 1.875%. These metrics answer different questions.

Pre-agreed boundaries are full cost no higher than 25 USD, reversals no higher than 20%, and 1,080 USD total expenditure. They are met. Break-even requires at least 39 results: 38 × 28 USD = 1,064 USD falls short, while 39 × 28 USD = 1,092 USD covers costs. Seven additional reversals would leave 38 and a −16 USD result. Do not claim stable profitability from 45 observations; consider only a separately budgeted follow-up pilot.

Scenario two: downloads increase but commercial results lose money

A second independent pilot promotes the same software category through a useful CSV sample. The team's mistake is treating the download as its main success measure. The page genuinely supplies the promised file without a form; the programme still pays only for an accepted demo request. This is not a randomised comparison: assets, inventory and costs differ, so the outcome cannot be attributed to one factor.

There are 4,000 paid clicks, 3,000 loads, 320 downloads and 80 forms. Twenty forms are duplicates and another twenty concern coursework rather than business implementation. Forty are provisionally accepted; ten are reversed after an additional intent check. Final accepted leads number 30. Commission of 28 USD produces 840 USD, not 320 times commission or 80 times commission.

Media 600 USD; content five hours × 20 USD = 100 USD; images four hours × 20 USD = 80 USD; tracking three hours × 20 USD = 60 USD; review five hours × 20 USD = 100 USD; tools 20 USD; non-recoverable fees 40 USD. Total 1,000 USD. Average CPC is 0.15 USD, but full cost is 1,000 USD / 30 ≈ 33.33 USD. The result is −160 USD, or −16% of expenditure; reversals are 10 / 40 = 25%. Cheap clicks do not compensate for weak commercial intent.

Website group: 1,600 clicks, 320 USD media, 30 provisional leads, six reversals and 24 final acceptances. Allocate 40% of shared additional costs, 160 USD. Total 480 USD, cost per result 20 USD, revenue 672 USD, result 192 USD.

App group: 2,400 clicks, 280 USD media, ten provisional leads, four reversals and six final acceptances. Allocate 60% of additional costs, 240 USD. Total 520 USD, cost per result approximately 86.67 USD, revenue 168 USD, result −352 USD. Reconciliation: 480 USD + 520 USD = 1,000 USD, 24 + 6 = 30, and 192 USD − 352 USD = −160 USD.

These fictional groups do not mean apps are always poor. Click-proportional cost allocation was chosen in advance; it proves neither causality nor next test's preparation cost. Review task fit, devices and available placement information while retaining unknown rows.

Pre-agreed limits are 1,000 USD total spend, 160 USD tolerable loss, full accepted-lead cost up to 25 USD, and reversals up to 20%. The budget is exhausted, loss reaches its boundary and both quality measures fail: stop buying. Break-even requires 36 accepted leads; only 30 exist. A possible next step is a separately costed test with a clear demonstration proposition and correctly defined measurement objective, not a larger download budget.

What to fix and what reports cannot prove

High CTR with few loads calls for checks of speed, redirects, repetition and accidental taps. It is not automatic bot evidence. Many loads without forms may mean the content already solved the problem; ask whether the reader needs the commercial step. High reversals require advertiser reasons, not a new “free” headline when no free service exists.

Two clicks on a weak placement do not justify a sweeping exclusion list. Compare cohort maturity, volume and incomplete data. Ineligibility, broken measurement or a false promise require an immediate stop; waiting for algorithmic learning does not override those boundaries. Avoid changing country, page, optimisation and budget together if the next observation is meant to teach you something.

Both calculations value the author's work and include tools and non-recoverable expenses, assuming the final ad invoice matches the stated spend. Add intermediary or currency costs when applicable, without inventing an AffiliateCore tariff. Individual income tax is not calculated. The supplier's development and sales expenses are not mislabelled as affiliate costs. With zero accepted leads, CPA is undefined; with zero revenue, loss equals expenditure.

A reviewable outcome

Complete a brief for one US product, prepare two images and independent short headlines, then inspect the destination without its affiliate button: is it still useful? Build a status record and recalculate both scenarios with five additional reversals. Preserve the decision, original conditions and one next testable change.

Only after verifying a supported payable event and integration, and if an eligible offer exists, sign in to AffiliateCore, match source and payable event to your campaign terms and use its permitted link. Our first paid pilot guide provides background on spending discipline. A link alone does not guarantee automatic reconciliation between Google Ads and accepted results.

US targeting, account, payment and source are checked separately.
Creative combinations remain accurate.
The page is useful without mandatory registration or onward travel.
Optimized targeting and the test objective are recorded.
Downloads, submissions and accepted leads are distinct.
Costs, unknown placements, rejections and reversals remain visible.
Stop rules and a fresh estimate precede further buying.