Google Search Campaign Setup Checklist: Pre-Launch QA

A Search campaign is ready to launch only when its measurement, settings, targeting, structure, ads, landing pages, and safety controls agree with the business goal. Creating the objects is the easy part. The real launch check is whether the campaign can spend in the right places, optimize toward the right outcome, and produce evidence you can trust.

Use this checklist twice: once while the campaign is paused, and again after the first meaningful traffic arrives. The first pass catches configuration mistakes. The second catches assumptions that only real search terms, eligibility signals, and conversion data can expose.

Do not enable the campaign until you can answer yes to these questions:

If any answer is unclear, the campaign is not ready. “Google accepted the settings” is not the same as “the campaign is correctly designed.”

A checklist cannot repair a vague brief. Write down the decision inputs before you open Google Ads or ask an AI assistant to build anything.

Name the outcome the campaign should influence: purchase, qualified lead, booked demo, phone call, store visit, or another measurable action. Do not use “traffic” as a substitute for an unclear commercial goal.

Confirm which conversion action should guide bidding. A form submission that produces weak leads can make the campaign look efficient while the pipeline quietly gets worse. Keep the Google Ads signal and the downstream business definition separate.

Choose one offer and the page that best fulfills it. The landing page should match the query intent, repeat the core promise from the ad, load correctly on mobile, and make the next action obvious.

Test the full path yourself. Submit the form, make the call, complete the purchase, or trigger the intended event. An enabled conversion action proves configuration exists; it does not prove the browser-side event fires correctly or that the lead reaches the CRM.

Record the target locations, exclusions, acceptable customer profile, sales constraints, daily budget, target economics, and any legal or brand restrictions. This becomes the review standard for every later setting.

Use a naming convention that a teammate can understand without opening the campaign. Include the market, campaign type, intent or offer, and any language or product distinction that matters.

Create the campaign paused. A paused campaign is not a draft plan; it is a real campaign object that can be inspected without serving. Review the live saved values before activation.

Check that the daily budget is affordable at the account level, not merely plausible for this campaign in isolation. Account for other campaigns drawing from the same budget, shared-budget relationships, and the possibility that daily spend will vary around the average.

Match bidding to the campaign’s goal and data maturity:

Inspect the campaign-level goals instead of assuming account defaults are correct. Confirm which conversion actions are primary, which are secondary, and which are included in the Conversions column used by bidding.

A campaign can be technically complete and optimize toward the wrong action. That is the most expensive kind of tidy setup.

Confirm where the campaign may serve. For a controlled Search launch, start with Google Search and decide separately whether Search Partners belongs in the test. Do not let Display expansion slip into the campaign simply because it was available during setup.

If you enable another network, define how you will evaluate it. A blended campaign total can hide a network that spends without producing comparable outcomes.

Check both the selected locations and the location-option logic. For most local or market-specific acquisition campaigns, presence-based targeting is the safer starting point than including people who merely show interest in the location.

Review exclusions, especially when the target is a large region with non-serviceable areas. After launch, inspect geographic performance and user-location evidence before expanding.

Confirm the start date, any end date, and the account time zone. A midnight launch follows the account’s clock, not yours.

Use ad schedules only when the business can act on demand or the evidence justifies the restriction. Treat device and audience settings as deliberate choices. Observation can collect evidence without narrowing reach; targeting restricts eligibility.

Each ad group should represent one coherent reason to search. If one ad cannot honestly speak to every keyword in the group, the structure is too broad.

Do not split ad groups merely to create administrative symmetry. Split when the query intent, value proposition, landing page, geography, or required message changes.

Review every keyword for:

Phrase and exact match can still reach close variants. Broad match can be useful, but it expands the importance of measurement, bidding, and search-term review. Match type is a control surface, not a promise that Google will interpret intent exactly as you do.

Add the obvious exclusions before spending begins: jobs, careers, free, support, login, definitions, DIY, or other themes that are irrelevant to the offer. Adapt the list to the business; generic negative lists can block valuable demand.

Check shared negative lists and campaign-level negatives together. A perfect new keyword can still be silently blocked by an older exclusion.

Google requires at least three headlines and two descriptions for a responsive search ad. A stronger launch usually provides more unique assets, but filling every slot with near-duplicates does not create useful variety.

Review the first three headlines as if they appear together. Make sure the set can communicate:

Use pinning only when a message must appear in a specific position—for example, a legal disclaimer, brand requirement, or strict message sequence. Over-pinning reduces the combinations Google can test.

Open every final URL. Confirm redirects, HTTPS, mobile rendering, page relevance, and the final domain. The display path should help the user understand the destination; it does not change where the click lands.

If the campaign uses tracking templates, final URL suffixes, or custom parameters, verify the effective value at the correct level. More-specific settings can override broader ones. Test the resolved click URL and confirm the landing page still loads.

Add the asset types that genuinely help: sitelinks for deeper navigation, callouts for proof or differentiators, structured snippets for categories, and call, price, image, lead-form, or location assets when relevant.

Assets are not decoration. Each should either answer a buying question, add proof, or give the user a useful next route.

Verify auto-tagging or the chosen tracking method, submit a real test conversion, and confirm the event reaches the systems that matter. Where applicable, check Google Ads, Google Tag Manager, GA4, and the CRM separately.

Do not infer browser execution from configuration alone. A tag can exist and still fail because of consent behavior, triggers, redirects, cross-domain navigation, or page changes.

Check for disapproved ads, invalid URLs, account verification tasks, billing holds, and policy warnings. Some delivery blockers are visible only in the Google Ads interface. An AI-led account review should label those as interface checks instead of inventing certainty.

Before enabling the campaign, capture the campaign ID, budget, bidding strategy, conversion goals, locations, networks, ad groups, keywords, negatives, ads, final URLs, assets, and approver.

This is less glamorous than another round of headline ideation. It is also what lets you distinguish a bad decision from an accidental edit two weeks later.

Use this after the campaign has been created but before it is enabled:

The useful output is not a generic score. It is a decision queue with evidence, entity IDs, and a clear launch gate.

HireOtto can create a Search campaign paused, apply opinionated defaults when inputs are omitted, and return the created IDs for review. Its current campaign-creation guide documents Google Search-only delivery, presence-based location targeting, and a start date of tomorrow as defaults. Those defaults reduce blank-field mistakes, but they are not a substitute for checking whether the choices fit the business.

From the same AI conversation, a practitioner can inspect campaign settings, add ad groups and keywords, manage negatives, create responsive search ads, review disapprovals, manage supported assets and URL tracking, and make approved campaign updates. The safer loop is:

HireOtto does not replace landing-page QA, browser-side tag testing, billing review, or commercial judgment. It compresses the account inspection and execution work so the operator can spend more time deciding whether the campaign deserves to launch.

A clean setup is only a hypothesis. Real traffic provides the second QA pass.

Confirm impressions, spend, ad approvals, keyword eligibility, and obvious budget or rank constraints. If nothing is serving, diagnose the first broken layer before raising the budget.

Review early search terms for intent drift. Add clear negatives carefully, but avoid judging a low-volume campaign from a handful of clicks.

Confirm that expected conversions appear and reconcile them with site, analytics, and CRM evidence. Separate reporting delay from a genuinely broken event.

Review recent change history when the saved state no longer matches the launch record. Sequence is evidence; it does not prove causation.

Label decisions that require a mature sample. Avoid changing bids, match types, ads, and landing pages simultaneously. If every variable moves, you learn nothing.

At a practical level, you need a budget, bidding strategy, targeting, at least one ad group, keywords, an eligible responsive search ad with a valid final URL, and a conversion setup appropriate to the goal. Google’s interface may also require an eligible conversion goal and will flag missing or invalid components.

Treat it as a separate choice, not an automatic yes or no. A controlled launch often begins on Google Search only so traffic quality is easier to interpret. If Search Partners is enabled, evaluate it separately where reporting allows.

There is no useful universal number. Keep enough keywords to represent one intent without forcing the ad and landing page to serve unrelated needs. Structure follows message coherence, not a quota.

Only when conversion signals, bidding, negatives, budget, and search-term review can support the wider reach. Broad match is not inherently wrong; using it without the surrounding controls is.

No. AI can inspect the account state that its connected tools expose and compare it with your brief. It cannot automatically prove landing-page experience, browser-side tag firing, lead quality, billing clearance, or every policy and eligibility message.

No. Create it paused, inspect the saved entities and settings, fix discrepancies, test measurement, and enable it only after the launch gate passes.

The best Search campaign checklist is not a longer list of settings. It is a sequence of proof.

First prove the business goal and measurement. Then prove the campaign settings. Then prove the structure, ads, destinations, and safety controls. After launch, prove that real traffic behaves as expected.

AI can make the inspection faster. HireOtto can make the account work more direct. The operator still owns the launch decision.