Choose Search, Shopping, or Performance Max by the question your store needs to answer first. Search helps you examine active demand; Shopping tests the product-card route; PMax expands across channels using the goals and inputs you supply. Define one question, the evidence you need, and what stays on hold before setting up the account.
12 parts · 6 checkpoints
Show all 12 parts
- 01Start with how buyers express demand
- 02Define the product and measurement terms
- 03Give each campaign a defined role
- 04Choose a starting route from your evidence
- 05Practise four starts for the same product, has a checkpoint
- 06Record your business and readiness separately, has a checkpoint
- 07Match guidance to the route you selected, has a checkpoint
- 08Check four inputs before a Search launch, has a checkpoint
- 09Explain PMax inputs and scope before expanding
- 10Inspect the fields before continuing
- 11Repair the upstream gap first
- 12Check the decision with one action, has a checkpoint
Key point Give the first campaign one job
The first campaign should answer one main question. ROAS (return on ad spend) is revenue attributed to ads divided by ad spend. It measures revenue efficiency, not profit. Changing queries, product data, creative, and tracking together makes total clicks or blended ROAS hard to interpret.
Start with how buyers express demand
A Shopify store has a newly published product page for a leak-proof commute tumbler. Its hero product is the 20oz black or white version. The example targets English-speaking buyers in Canada and plans a 30 USD daily ad budget. That amount is an illustrative planning input, not a platform minimum or proof of eligibility. No earlier lesson is required to follow this case.
The owner expects searches such as leak proof travel mug and knows that product cards invite comparisons of image, price, stock, and delivery. Purchase tracking has not passed QA, and Merchant Center data has not been reconciled with the product page. The immediate work is a choice map and a verification task: hold Shopping and the retail PMax route used here. A separately controlled, very small Search query test can read demand while tracking is repaired; it cannot establish purchasing performance.
A campaign holds the objective, type, budget, and major settings. “Will these high-intent searchers buy from this product page?” identifies a reviewable chain: query, ad promise, first-screen content, and order signal. “Sell more” leaves those different questions mixed together.
Three buyer paths for the same tumbler
People already searching
- Buyer situation
- A shopper has already expressed the need in Google, for example "leak proof travel mug." This traffic is closer to purchase, but you still need to inspect the query that triggered the ad.
- Campaign fit
- Search is usually the clearer starting point. Read the query, ad promise, and landing-page fit before expanding reach.
- First checks
- Search term, keyword, ad headline, Final URL, Purchase / Lead, and above-the-fold page content.
- Common trap
- Do not make keywords broad just because Search is readable. Broad terms can attract clicks from people with little intention of buying.
People comparing product cards
- Buyer situation
- They compare price, image, stock, shipping, reviews, and size. Google mainly understands the product through product data.
- Campaign fit
- This usually points to Shopping, or feed-centered PMax. Merchant Center and product-page fields must match first.
- First checks
- Item ID, SKU, title, image_link, price, availability, GTIN, and Merchant Center Diagnostics.
- Common trap
- Do not blame bidding before the feed is accepted. If product data is wrong, the system may not even learn which product should show.
People the system tries to find
- Buyer situation
- They may come from Search, YouTube, Gmail, Discover, Display, or remarketing. The system combines your goals and inputs across several entry points.
- Campaign fit
- This is where Performance Max fits better: expanding after tracking, feed, assets, and budget window are already stable.
- First checks
- Conversion action, value, currency, transaction ID, asset coverage, product group, brand boundary, and learning status.
- Common trap
- Do not treat PMax as a beginner shortcut. Unreliable inputs can direct automated optimization toward the wrong goal.
Searching for buy leak proof travel mug is a different situation from watching a lifestyle video. Read the first row as a query–ad–page check, then compare it with the product-card row, where product facts matter directly. These are useful decision lenses, not mutually exclusive audiences: PMax can span the other routes.
- Should my first Google Ads campaign be Search, Shopping, or PMax?
- What can Search, Shopping, and PMax each help you check?
- Can I launch Shopping or PMax before the feed is ready?
- Which campaign can I inspect first when conversion tracking is not accepted?
- How much budget should I prepare for the first Google Ads campaign?
- Why should I avoid launching Search, Shopping, and PMax together in week one?
- When can I move from Search or Shopping to PMax?
Define the product and measurement terms
SKU connects a product with inventory, orders, and margin. Conversion goals tell the system what to optimize; CVR and ROAS help interpret the result. Identify the object and measurement definition before comparing numbers.
Terms used in the checks
- Google Ads
Google Ads is Google’s advertising system, not one ad slot. It can serve ads across Search, Shopping surfaces, YouTube, Display, Discover, Gmail, Maps, and more.
Where you see it You see it in the Google Ads account where campaigns, goals, budgets, bids, assets, and conversions are configured.If misread If every placement is treated as the same traffic, week-one data may look busy while the real source of results stays unclear.Example: For a 20oz tumbler, a person searching "leak proof travel mug" and a person seeing a YouTube video answer very different validation questions.- campaign
A campaign is the layer that carries the business goal, budget, campaign type, and major settings. It answers what this spend is supposed to learn first.
Where you see it You choose a campaign type when creating a campaign, then set conversions, budget, bidding, and assets.If misread If the first campaign handles acquisition, remarketing, product testing, and creative testing at once, the readout becomes blended.Example: A tumbler launch can start by validating high-intent Search terms instead of adding Display remarketing and PMax scaling at the same time.- Search
Search ads are triggered by search queries, so they are useful for seeing what people are actively trying to find.
Where you see it You see it in Search campaigns, keywords, search terms, and ad groups.If misread If keywords are too broad or the page mismatch is large, clicks may rise while purchase intent stays weak.Example: "buy leak proof travel mug" is a cleaner first purchase test than "coffee lifestyle ideas."- Merchant Center / Feed
Merchant Center manages product data for Google. A feed is the product data table with fields such as title, image, price, inventory, brand, and GTIN.
Where you see it Shopping and feed-based Performance Max read these fields, and Google uses them for product review and matching.If misread Wrong price, inventory, title, or GTIN can make Shopping/PMax learn from incorrect product signals.Example: If one tumbler has 12oz and 20oz variants, the feed must clearly carry size, price, inventory, and images for stable Shopping surfaces.- SKU
SKU is the store-owned identifier for a product or variant. It is not a Google-generated ID; it is the unit your team uses to connect inventory, price, margin, and ad readouts.
Where you see it You see it in Shopify products, order exports, product feeds, Merchant Center sample sheets, and margin reports.If misread When SKUs are messy, Shopping and PMax product performance is hard to match with real inventory, margin, and refunds.Example: A black 20oz tumbler and a white 12oz tumbler should use different SKUs, or ad spend, inventory, and profit get blended.- CVR
CVR means conversion rate: how many clicks or visits become key actions such as Purchase, Lead, or AddToCart.
Where you see it You see it in Google Ads, GA4, Shopify analytics, and page review sheets.If misread When CVR is low, the first problem may be keywords, page fit, price, shipping, or tracking quality rather than campaign type.Example: If Search brings 200 clicks and 0 orders, do not blame budget alone; inspect the queries and the first-screen page fit.- ROAS
ROAS is return on ad spend: revenue attributed to ads divided by ad spend. It is revenue efficiency, not profit.
Where you see it You see it in Google Ads, GA4, and business review sheets, but attribution can differ by platform.If misread Comparing Search, PMax, and Display ROAS in one table can hide that each type has a different job.Example: Display can have low ROAS while building a useful remarketing pool; high Search ROAS also does not mean unlimited scaling.- conversion tracking / key event
Conversion tracking records important user actions. Ecommerce usually cares about Purchase, but Lead, AddToCart, or InitiateCheckout may also matter.
Where you see it You verify it in Google Ads conversion settings, GA4, Shopify orders, and ad reports.If misread If value, currency, trigger timing, or deduplication is wrong, automated campaigns can learn toward the wrong goal.Example: Purchase should fire after a real order with value and currency, not repeatedly just because a thank-you page is visited.
Before comparing CVR, name the denominator: ad interactions, site visits, and Shopify sessions are different measures. Event definitions and included conversions must also match. ROAS divides attributed revenue by ad spend; it does not deduct product, delivery, or refund costs. A GTIN is also distinct from an internal SKU: use assigned identifiers and the requirements that apply to the product and market, rather than inventing a number.
Give each campaign a defined role
Search makes active demand easier to inspect; when product data is stronger, Shopping may be the clearer product-card test. Display and Video describe reach, education, and remarketing jobs here. They can also support conversion goals: a view alone is insufficient purchase evidence, but that does not mean video cannot sell.
In this table, advertising inventory means the places ads can appear, such as Search or YouTube. Product inventory means the stock available to sell, which you check against the feed and product page.
Compare campaign roles
Search
- Job
- Validate active search intent first
- Useful when
- People actively search for the product, problem, or use case. With a small budget, you need readable queries, a clear ad promise, and a matching landing page.
- Outside this test
- Demand has no clear query yet, or keywords are so broad that click quality cannot be explained.
- First evidence
- Search terms, CTR, CVR, Purchase/Lead events, first-screen message match.
Shopping
- Job
- Validate product data and price competitiveness
- Useful when
- The feed is approved, and title, image, price, inventory, shipping, and applicable product identifiers are stable.
- Outside this test
- The feed is new and unaudited, product page price/inventory changes often, or variant data is messy.
- First evidence
- Merchant Center diagnostics, product approval, title quality, price/inventory consistency, product-page quality.
Performance Max
- Job
- Expand across inventory after signals are stable
- Useful when
- Conversion goals are clear, and feed, assets, audience signals, and budget window can support automated learning.
- Outside this test
- Using it as a beginner shortcut, or giving it most budget while tracking, feed, and assets are unstable.
- First evidence
- Conversion quality, asset coverage, product grouping, Search overlap, brand exclusions/negative boundaries.
Display
- Job
- Support reach, remarketing, and demand education
- Useful when
- There is a clear audience, remarketing pool, or education task, and the first round is not meant to prove purchase demand.
- Outside this test
- Using impressions or cheap clicks as proof that a product is validated, especially before purchase tracking passes QA.
- First evidence
- Audience source, frequency, visit quality, remarketing pool size, downstream purchases/leads.
Video
- Job
- Explain use cases, grow awareness, and learn from creative
- Useful when
- The product benefits from a demonstration, comparison, or story, and the team has the assets and a suitable destination page.
- Outside this test
- A tiny-budget first launch that expects stable purchases while asset inventory is thin.
- First evidence
- Video completion/engagement, search lift, landing-page visit quality, downstream Search/Shopping capture.
Choose a type with the marketing goal, brand strategy, and available team time in mind. Starting with Search on a small budget is this lesson’s practical recommendation for a readable test, not a Google-mandated sequence. Text and keywords can provide a first Search test without a video library or product feed. Shopping matches using Merchant Center product data. A product page by itself does not establish feed readiness.
Google’s current Demand Gen guide lists YouTube, Discover, Gmail, Maps, and the Google Display Network, and announces a staged move of Display campaigns into Demand Gen. The older “Display / Video” task labels here are not a promise that every account offers the same creation button today. Check the current account’s workflow, inventory, and eligibility. Demand Gen remains distinct from PMax; inspect its creative, channel controls, and goal separately.
Choose a starting route from your evidence
Select the problem closest to your business. The map gives a starting route, later options, a hold list, three evidence checks, and the first action. Its initial active-search example describes conditions; it does not confirm your account meets them.
What needs explaining first?
- Question
- Validate search intent and page fit first
- Suggested start
- Search
- Later routes
- Shopping / Performance Max
- Hold this round
- Display / Video as first purchase validation
- Required evidence
- You can list 10-30 high-intent queries, not generic interest terms; the product page first screen matches the ad promise and clearly presents price, shipping, and trust information; Purchase or Lead tracking has passed QA.
- First action
- Launch one small-budget Search campaign to answer whether active searchers click and convert.
- Question
- Validate product competitiveness on Shopping surfaces
- Suggested start
- Shopping or feed-centered Performance Max
- Later routes
- Search for high-intent capture, PMax for broader coverage
- Hold this round
- Scaling before feed review
- Required evidence
- Merchant Center has no major disapprovals, and price/inventory match the site; title, image, brand, GTIN/MPN, and variant relationships are explainable; Product pages clearly present price, shipping, returns, and reviews.
- First action
- Use feed quality as the launch gate, then decide Shopping or PMax product scope.
- Question
- Use automation to expand across inventory
- Suggested start
- Performance Max
- Later routes
- Keep Search for high-intent control and use Shopping/feed grouping for calibration
- Hold this round
- Unverified tracking, thin assets, too-short budget window
- Required evidence
- Purchase/Lead goal, value, currency, and deduplication are trustworthy; there are enough images, headlines, descriptions, and any video gap is documented; The team knows which brand terms, product groups, or pages should not be blended by automation.
- First action
- Document PMax input quality, overlap boundaries, and stop conditions instead of treating it as a shortcut.
- Question
- Support awareness and return visits
- Suggested start
- Display / Video
- Later routes
- Search captures new demand; Shopping/PMax captures product clicks
- Hold this round
- Treating impressions, views, or cheap clicks as product validation
- Required evidence
- There is a clear audience source, creative theme, and downstream page path; the review metric is defined: visit quality, remarketing pool, or later brand search; The first round is not expected to prove purchase ability directly from Display/Video.
- First action
- Frame it as education or remarketing, while purchase validation stays on a more explainable path.
The active-search example uses 10–30 high-intent terms, a matching page, and validated tracking. That range keeps the exercise manageable; it is neither an eligibility threshold nor a sufficient statistical sample. If product data is the stronger evidence, inspect the card route first. Automation requires more inputs and boundaries to be understood.
Test the reason against a common misread
Compare active demand, a mature catalog, and a new brand below. In the catalog example, “products exist” still leaves product-data and page sampling to do. The last column is a verification task, not permission to increase spend.
Different questions behind the same launch request
Situation 1
- Situation
- The product has active demand, such as people searching for "leak proof travel mug" or "portable coffee cup."
- Likely misread
- Launch Performance Max immediately because it reaches more placements and looks easier.
- Better interpretation
- Start by validating search intent and page fit; Search makes the purchased queries easier to inspect.
- Next action
- Write 10-30 high-intent terms, ad promise, and landing page before launching Search.
Situation 2
- Situation
- The catalog is complete, with stable price, inventory, images, and variants.
- Likely misread
- Assume Shopping or PMax is ready just because products exist, without inspecting the feed.
- Better interpretation
- For Shopping, Merchant Center must receive accurate product facts. Having a catalog alone is insufficient.
- Next action
- Sample 20 SKUs for title, image, price, inventory, GTIN, and product-page consistency.
Situation 3
- Situation
- The brand is early, with little branded search and no stable remarketing audience.
- Likely misread
- Treat cheap Display or Video clicks as the first sales validation.
- Better interpretation
- Cheap clicks may answer a reach question, not a purchase-intent question.
- Next action
- Define Display / Video as education or remarketing, separate from Search purchase validation.
Each comparison asks you to make the missing evidence interpretable before expanding. Cheap clicks, wider reach, or a larger catalog cannot settle that decision alone.
Practise four starts for the same product
These four scenarios are examples, not observed ad-account results. The first has 24 high-intent queries, making Search a sensible start. The feed example has a passing sample of 20 SKUs. Read the setup and make your own choice; switching scenarios does not choose the answer for you. The role table above explains what each action is for.
Choose the first campaign
- Setup
- You can list 24 high-intent queries such as leak proof travel mug and 20oz insulated tumbler with lid. The first screen shows leak proofing, size, price, shipping, and reviews; Purchase has passed QA.
- First question
- Whether active searchers will click and purchase.
- Hold
- Hold PMax, Display, and Video; add Shopping after feed sampling is complete.
- Evidence to retain
- Before launch, write 10-30 queries, matching ad promise, first-screen proof, and stop conditions.
- Setup
- A 20-SKU Merchant Center sample passes: title, image_link, price, availability, GTIN, and product page match. Query research is still rough.
- First question
- Whether product cards and price are understandable in Shopping surfaces.
- Hold
- Hold broad Search and PMax expansion; do not let automation blend unexplained demand yet.
- Evidence to retain
- Before launch, keep Merchant Center diagnostics, SKU sample sheet, product-page fields, and price/inventory match notes.
- Setup
- Search already found stable purchase queries, the feed is sampled, Purchase value / currency / transaction ID are trusted, the asset library covers images, short headlines, long headlines, and documented video gaps, and channel performance, Search themes, brand exclusions, and URL boundary can be explained.
- First question
- Whether reach can expand while boundaries stay clear.
- Hold
- Do not blend brand, remarketing, high-margin, and low-margin products into one boundaryless budget pool.
- Evidence to retain
- Before launch, document conversion goal, product groups, channel performance checkpoints, brand exclusion/negative boundaries, URL boundary, asset gaps, and 7-day review rules.
- Setup
- The product promise is commute use, leak proofing, and car-cup-holder fit, but search volume is low. The team has comparison videos and use-case assets, and the goal is education.
- First question
- Whether creative can educate the use case and create downstream visit quality.
- Hold
- Purchase validation still belongs to later Search, Shopping, or a trusted Purchase path.
- Evidence to retain
- Before launch, define post-view page depth, branded search change, remarketing pool size, and downstream capture path.
“Launch all types together” mixes intent, feed, assets, tracking, and audience in the same result. The exercise is about a clear testing sequence, not the number of buttons enabled. Before using feed-only PMax on a small budget, ask whether the budget supports cross-inventory learning, whether top sellers and a feed sample exist, and who will inspect search terms and channel performance each day.
Record your business and readiness separately
A service, hero product, multi-SKU catalog, and new product need different first evidence. The business profile selects an example; product or service scope records what this round actually covers. These are separate inputs. Changing the example does not rewrite the scope, so reconcile them before saving.
Business profile
- Business setup
- For example, Shopify consulting, installation, or a custom service. The key is not a product card, but whether people have a clear problem, the page explains the service, and qualified leads are recorded correctly.
- First question
- First validate whether people with active demand leave qualified leads.
- Business setup
- For example, one 20oz leak-proof commute mug. Separate a question about queries and page promise from a question about product cards, price, and inventory facts.
- First question
- Decide whether active search or product cards answer the current question more clearly.
- Business setup
- For example, multiple mug styles, colors, and sizes. The more products you have, the more clearly you need to define product scope, variants, price, and inventory before using one blended result for a product decision.
- First question
- First prove that product data is understandable, then decide whether to expand to more products and inventory.
- Business setup
- For example, buyers do not yet understand the use case for a leak-proof commute mug. You may need to teach the use case first, but impressions or views are not purchase proof and still need a readable downstream path.
- First question
- First define the education goal and how purchase intent will be validated later.
Records stay in this browser. Core choices, question notes, review notes, and page observations each have their own Save, Restore, Clear, and JSON export. A button affects its own group. JSON backs up fields. Keep credentials, customer information, real order numbers, and account identifiers out of these notes.
Scope entry
Documented input count
The initial hero-product scope contributes 1 recorded input, while all six readiness fields remain “not checked yet.” AOV means average order revenue. Contribution profit deducts relevant costs; together they help the responsible person set an affordable acquisition cost rather than treating ROAS as profit.
Match guidance to the route you selected
Readiness counts inputs; it does not choose the route. Explicitly select a start or a hold. “Hold campaign creation” retains that decision. PMax with the retail inputs required by this exercise receives PMax guidance. Every result still needs current account, page, and order verification.
Starting route this round
- What this choice means
- Write the question, evidence, and hold list before choosing a type.
- What this choice means
- Use it to read active search, ad promise, and page fit.
- What this choice means
- Use it to read product cards, with trustworthy Merchant Center product data.
- What this choice means
- Use it to expand after goals, signals, assets, and product scope are clear.
- What this choice means
- Use it for use-case education or visual asset validation; leave purchase judgment to a later readable path.
- What this choice means
- When conversion, feed, page, or the question itself is unclear, repair evidence first.
Current starting guidance
Hold
Write this round’s question in one sentence first
Before choosing a type, state whether this round reads active search, product cards, post-education visit quality, or trusted automated expansion. Until the question is clear, the readiness score only counts entries. Hold campaign creation until the question, evidence location, and responsible lead can all be written down.
Treat conversion tracking as the starting condition
Whether you choose Search, Shopping, or PMax, the current goal, value, currency, and dedupe are not yet documented as reviewable evidence. Use the conversion-tracking lesson to check those inputs first. Do not treat clicks, impressions, or a system suggestion as purchase evidence.
Do not use Shopping as the first service-lead test
Shopping is evaluated through retail product data and product cards. A service business should first confirm the user problem, page explanation, and qualified lead before deciding whether a separate product path exists. Record it as held, then frame the service scope as a readable Search or qualified-Lead validation.
Make product facts withstand sampling before Shopping
Shopping uses Merchant Center product data to decide how and where ads show. When price, inventory, title, image, variants, or product pages are still unclear, repair product facts first. Hold Shopping and feed-based PMax, and retain the product-data diagnostic record.
PMax inputs are still incomplete
PMax optimizes around the goals, assets, audience signals, and data feeds you provide. When feed or assets are unclear, reduce scope and repair those inputs first. Write PMax as held and name which input must be completed before the next review.
Explain what users will search before Search
Search is valuable because people are actively looking for a product or service. When high-intent queries are still unclear, map keywords and page language first instead of forcing a sample with broad terms. Hold Search until you can write the question, query set, page promise, and pause condition.
Separate visual education from purchase validation
Demand Gen and video paths can support visual education or creative-performance review, but the current readout must say whether it measures use-case understanding, page depth, or a remarketing pool. Purchase judgment still needs a later readable path. Do not record impressions, views, or cheap clicks as proof that the product is validated.
A multi-SKU catalog can consider PMax after product scope is fixed
The current inputs are relatively complete. Document product groups, brand or page boundaries, the observation window, and the review lead before combining more of the catalog. Expand one explainable scope at a time and keep Search or Shopping as comparison evidence.
The product-card path can begin with a narrow validation
When feed and conversion records are explainable, first limit product scope and observe whether product cards, price, and inventory facts stay consistent. Do not turn it into an account-wide conclusion. Keep the reasons for holding Search and PMax until the current product path has a reviewable readout.
Use one explainable Search path to read active demand first
Current conversion and search demand are documented. Use one question, a high-intent query set, and the same landing page to validate the promise, then decide after the observation window whether scope belongs in Shopping or PMax. These Search results support a decision about this demand path. Feed readiness and automated expansion still need their own checks.
Keep the selected hold
You selected a hold. Keep the question, missing inputs, evidence location, and owner on record, then choose a route after review. Record the condition for the next review; a hold is not Search readiness.
Review PMax within a fixed scope
You marked tracking, feed, and assets as confirmed for this retail exercise. Inspect product groups, brand and page scope, the observation window, and owner before implementation. This recommendation uses self-reported inputs; it proves neither account eligibility nor profit.
The feed condition in this decision exercise applies to the retail product-data route. PMax also has separate online-sales, lead, and other goal workflows; a product feed is not a universal condition for every PMax use. A service without a retail product route should retain the unanswered checks and hold reason, then verify the appropriate conversion goal and setup rather than falsely marking feed readiness as confirmed.
Check four inputs before a Search launch
Search is useful when you can inspect the query, ad promise, and page together. The worked example has queries and page fit but lacks tracking and a budget window: 2/4, so gaps remain. Your checklist below starts empty on first use and after clear. The example contributes no evidence to your own count.
Search readiness
Hold
Hold the launch
Close the remaining gaps
A possible first validation route
In the example of 200 clicks and 0 orders, inspect actual queries, page content, price, delivery, and event records before blaming budget. A search-terms report is not a complete account of every buyer’s intent. Repair a mismatched page promise before buying more clicks.
If Search is the first validation route, begin with one main intent, one ad group, the same Final URL, and a small query set. Review actual terms, the ad promise, first-screen content, and order or lead samples together to find the layer that needs work. This controls variables for the current test; it is not a required structure for every account.
Explain PMax inputs and scope before expanding
PMax combines goals and inputs across inventory. More creative gives the system more combinations; it does not guarantee purchases or profit. For this retail example, review each condition below. Assign unresolved inputs to the relevant owner rather than expecting automation to repair them.
Four retail PMax checks
Conversion goals and value are trustworthy
- Why it matters
- PMax learns toward your stated goals. Wrong goals mean wrong optimization.
- Repair action
- QA Purchase/Lead, value, currency, and deduplication first.
Feed is no longer an unknown risk
- Why it matters
- Product title, price, inventory, and GTIN affect Google matching and product serving.
- Repair action
- Resolve Merchant Center diagnostics, variant relationships, and product-page consistency.
Assets can support multi-surface serving
- Why it matters
- Multi-surface serving is not only product cards; headlines, images, descriptions, and video gaps affect delivery.
- Repair action
- Prepare core images, short copy, and needed video, or at least document gaps.
Blending boundaries are known
- Why it matters
- Mixing brand, remarketing, high-margin, and low-margin products can look good while hiding scale risk.
- Repair action
- Document brand exclusions, product grouping, page scope, and stop conditions.
Audience signals guide automation; they do not restrict delivery to those people alone. Brand exclusions, negative keywords, product groups, and URL scope do different jobs. Writing “brand and remarketing separated” is not evidence that the controls achieve that boundary. Final URL expansion may send a visitor to another relevant page, so inspect the actual setting and generated content.
The 7-day window here is an initial operational review of tracking, product status, and query quality, not a promise that PMax has finished learning. Google’s current creation guide recommends at least 6 weeks and avoiding frequent changes. Allow for conversion lag, budget, and the sample condition agreed in advance. A tracking defect should be investigated before the example window ends.
Inspect the fields before continuing
The English names below identify objects to inspect; they are not one fixed click path for every interface version. Older Merchant Center “Diagnostics” references now mean product issues and Issue Details as applicable. Reports and controls can depend on account or goal. Record the observed date, status, and sample location without private values.
Three objects in this check serve different purposes. A keyword is a matching input you set; a query or search term is what the user actually typed. The Final URL identifies the landing page the ad points to. Compare the actual term, ad copy, and that page to see whether they answer the same buyer question.
Search / search terms / landing-page intent path
Purpose: Validate active demand first: people are already searching for the product, use case, or problem, and the ad plus page can answer it.
Objects and locations: Google Ads -> Search campaign -> Keywords / Search terms / Ads / Landing pages; then sample GA4 landing pages and Shopify orders.
Fields to record: Query, match type, keyword, ad promise, landing page URL, CTR, CVR, conversion action, order or lead sample, and negative keyword candidates.
Interpretation: Shows whether real queries, ad promise, above-the-fold page content, and Purchase / Lead form one explainable path.
When to hold: If terms are noisy, the page promise does not match, or Purchase is unverified, do not treat PMax or Shopping as the scaling answer yet.
Next work: Route into the Google Ads Pro search terms / negative keyword system, and reuse the insight for SEO keyword mapping.
Shopping / Merchant Center / feed readiness path
Purpose: Validate whether Google can understand the product card correctly: title, image, price, stock, GTIN, and product page consistency.
Objects and locations: Merchant Center -> Products / Needs attention / Diagnostics; then return to the feed, Shopify product / variant, and product page.
Fields to record: Item ID, SKU, GTIN, title, image_link, price, availability, shipping, approval status, diagnostics status, and product page URL.
Interpretation: Shows whether product data is ready to support Shopping or feed-centered Performance Max.
When to hold: If the feed has disapprovals, price or stock mismatch, unclear variants, or GTIN problems, fix the feed before blaming bids.
Next work: Route into Product Data Feed Operations, the GTIN / feed tools, and the Merchant Center diagnostics checklist.
Performance Max / input quality / conversion value path
Purpose: Expand coverage after goals, feed, assets, and conversion signals are trusted, instead of using automation to hide basic gaps.
Objects and locations: Google Ads -> Performance Max -> Asset groups / Listing groups / Goals / Insights; also inspect channel performance / inventory mix, conversion goals, brand exclusions, search themes, and final URL expansion.
Fields to record: Conversion action, value, currency, transaction ID, asset coverage, channel performance, inventory mix, final URL expansion, product group, audience signal, brand exclusion, budget window, and learning status.
Interpretation: Confirms that automation has trusted goals, enough assets, clear product scope, an explainable inventory mix, and a reviewable budget window.
When to hold: If conversion value, feed, assets, inventory mix, or brand / remarketing boundaries are unclear, do not let PMax carry the main budget yet.
Next work: Route into account setup and conversion tracking, plus Google Ads Pro PMax control, brand exclusions, and negative boundaries.
Budget / check window / hold-list path
Purpose: Make sure the first campaign has enough sample to answer one question, instead of splitting budget across unreadable variables.
Objects and locations: Google Ads -> Budget / Bidding / Change history; then review GA4, Shopify orders, and the team task sheet.
Fields to record: Daily budget, bid strategy, target CPA / ROAS if used, observation window, change history, hold list, pause condition, and responsible reviewer.
Interpretation: Shows that the team is reviewing by observation window, pause condition, and next task rather than changing ads by feel.
When to hold: If budget is fragmented, changed multiple times a day, or change history is not recorded, narrow terms, product scope, or geography first.
Next work: Route into budget and bidding basics and ROAS / profit review, then record the next action in the campaign notes.
Find the PMax Channel performance report under Campaigns → Insights and reports → Channel performance. Inspect the date range, conversion goal, product-data or video format, and channel Status diagnostics. No spend may reflect missing assets, restrictions, or the system’s allocation; an empty row alone does not identify the cause. Reported attribution still needs order and margin reconciliation.
Consistent UTM names also help connect the first campaign links with traffic, conversions, and channel reports. Tags distinguish sources; they do not replace Purchase tracking or prove incremental impact.
UTM Link Builder
Bring the landing-page URL and the agreed source, medium, and campaign names. Check the generated parameters, then use the same names in this round’s records.
Repair the upstream gap first
A CTA, or call to action, is a button or link that invites the shopper to act, such as “Add to cart.” When checking the page, ask whether the next step is clear and whether the nearby price and delivery details agree with the ad promise.
When a result is unclear, separate tracking, feed, page, and budget problems. Choose a missing input to see which types to hold, what to repair, and what to record. If several gaps exist, repair the upstream input first. Selecting a panel does not pause or change any real ad.
Current blocker
- Symptom
- Purchase / Lead count, value, currency, transaction_id, or deduplication still does not match Shopify / GA4 order samples.
- Hold
- Hold PMax, tROAS, Maximize conversion value, and any scaling that learns from Purchase value. Search can only run as a small query-reading test.
- Repair first
- Complete account and conversion tracking first. Verify the primary conversion, value, currency, deduplication, and test orders before returning to PMax or budget decisions.
- Key line
- Write in the campaign notes: fix tracking first; keep only explainable query samples this round; hold PMax / value bidding.
- Symptom
- Merchant Center has disapprovals, price / availability mismatch, internal-looking titles, unclear GTIN/MPN, or confusing variant relationships.
- Hold
- Hold Shopping and feed-centered PMax. Search can validate high-intent queries, but do not treat weak feed data as a bidding problem.
- Repair first
- Go to Merchant Center and product feed basics, then Product Data Feed Operations and GTIN / feed tools to fix fields, sample product pages, and record diagnostics.
- Key line
- Write in the notes: fix the feed first; hold Shopping / feed PMax; Search can read demand, but it does not validate the feed.
- Symptom
- Ad promise, hero copy, price, shipping, returns, reviews, and CTA do not line up; clicks arrive but AddToCart / Purchase / depth are weak.
- Hold
- Hold PMax scaling, Display / Video amplification, and large-budget Search. Keep only a narrow search terms and landing-page-fit readout.
- Repair first
- Go to CRO page roles, landing-page message match, or product-page trust work. Align the ad promise, above-the-fold content, and CTA.
- Key line
- Write in the campaign notes: fix page fit first; do not scale budget this round; record the query, ad promise, Final URL, and above-the-fold gaps.
- Symptom
- Daily budget is split across campaigns, regions, product groups, or bid strategies, with multiple same-day changes and no readable 7-day sample.
- Hold
- Hold parallel campaign types, automated PMax scaling, and complex bid changes. Narrow to one campaign, one main question, and one check window.
- Repair first
- Go to budget and bidding basics. Use ROAS / Pricing tools to prepare break-even ROAS, Max CPA, contribution profit, and pause lines before setting guardrails.
- Key line
- Write in the notes: narrow the budget sample; answer one question this round; discuss scaling only after the check window.
The risk selector checks how you are interpreting results: blended ROAS treated as comparable jobs, waiting for bad tracking to repair itself, changing bids before fixing a feed, or treating cheap views as sales. Read the cause and first action, then retain the relevant response in your notes.
Which launch risk fits?
- Symptom
- Week one has clicks, impressions, and a few events, but the team cannot tell which results came from Search, Shopping, PMax, or Display.
- Likely cause
- The first launch skipped the business question and tested intent, product data, assets, and audience at the same time.
- First action
- Pause the blended readout, choose one validation path, and move the other types into the hold list.
- Avoid
- Do not compare PMax, Search, and Display ROAS in one table as if they had the same job.
- Symptom
- PMax starts learning while Purchase / Lead count, value, currency, or deduplication still does not match Shopify / GA4.
- Likely cause
- Automation was handed an unverified goal, so the system may learn toward the wrong event or wrong value.
- First action
- Return to conversion tracking QA first: event trigger, value, currency, dedupe, and order samples must pass before PMax.
- Avoid
- Do not replace tracking acceptance with "let it run a few more days."
- Symptom
- Products are disapproved, price or inventory mismatches, or titles look like internal SKU names, but the team starts changing bids and budget.
- Likely cause
- Merchant Center and product-page facts were not validated, so Shopping / PMax is reading unstable product data.
- First action
- Sample titles, images, price, inventory, variants, GTIN/MPN, and product-page consistency before choosing product scope.
- Avoid
- Do not blame Smart Bidding or low budget while the feed is still weak.
- Symptom
- Impressions, views, or click cost look good while purchases, leads, page depth, and profit do not improve.
- Likely cause
- Reach or education traffic was treated as purchase validation, ignoring the job difference from active Search or Shopping intent.
- First action
- Define Display / Video as education, remarketing, or asset learning; keep purchase validation on a more readable path.
- Avoid
- Do not write impressions, views, or cheap clicks as "product validated."
Continue or hold before launch
| Stop here | Go |
|---|---|
Conversion tracking has not passed QA | Purchase/Lead, value, currency, trigger timing, and deduplication have proof |
| Evidence Field records and order samples from Google Ads, GA4, and Shopify | |
Search, Shopping, PMax, and Display all launch together | The first campaign answers one question |
| Evidence Selection map states which types are not touched this round | |
Shopping / PMax launches before feed review | Merchant Center diagnostics, price, inventory, and variants are explainable |
| Evidence Feed checklist and product-page samples | |
Impressions, views, or cheap clicks are used as product validation | Purchase validation stays with Search, Shopping, or a trusted conversion path |
| Evidence Review sheet separates visit quality, purchase quality, and profit quality | |
A fixed review order helps the team identify what needs work at the end of the week. These five steps carry the decision from its question into a usable map. Actual account work requires the appropriate authority and preparation; this lesson creates, submits, and launches no ads.
From a choice to a checkable plan
Write one business question
Identify active search, product cards, automated reach, education, or remarketing as the main question.
Shortlist from the buyer path
Find the clearest evidence among demand, product facts, tracking, and budget, then choose the route that can test it.
Compare the example with your situation
Use the four tumbler setups to practise a starting choice; state the reason and what keeps the others on hold.
Check current controls and gaps
When authorized, inspect current goals and types. For a retail route, check product data, website verification or claim, shipping settings, and account link. For measurement, inspect events and test orders or leads. Assign tracking, feed, page, budget, channel, or brand gaps to an owner.
Keep the decision and next check
Keep the question, starting route, fields, hold list, owner, review window, and continue/hold conditions for the relevant next lesson.
Check the decision with one action
Start with the poor example of launching everything and comparing blended ROAS. Read the reason, then answer for yourself. The question starts unanswered and shows the explanation for the option you choose. It is separate from the saved launch path; one does not rewrite the other.
What should the first round do?
Self-check not answered
The decision should explain whether to continue, hold, or revisit a particular input; actual results still need observation. At the next review, record CTR, CVR, CPA, ROAS, order or lead samples, refund reasons, and change history rather than selecting the most flattering number.
CTR is click-through rate: ad clicks divided by impressions. CPA is cost per conversion: ad spend divided by the conversions counted under your agreed definition. State whether that conversion is an order or a qualified lead before comparing costs for different outcomes.
No checkpoint finished in this lesson yet. Start with “Go to the exercise”.
Self-check not answered
Keep the decision for the next task
Use four lines for the question, starting route and reason, hold list, and review rule. An empty personal note remains unfilled; the example does not make the decision for you. Core choices and other note groups still save separately.
Common questions
- What is Google Ads for an ecommerce store?
It finds potential buyers through search intent, product data, creative, conversion goals, and budget rules. First identify an active-search, product-card, automation, or education task, then choose a route you can evaluate.
- Should the first campaign be Search, Shopping, or PMax?
Search is often the clearest active-demand test when queries and page fit are understood. Shopping can come first when product-data and page samples pass. Before expanding the retail PMax route, explain tracking, product data, assets, budget, and boundaries; do not generalize its feed check to every PMax business.
- Why avoid launching every type in the first week?
The types test different inputs and serve different jobs. Launching them together blends intent, product data, creative, tracking, and audience, making it harder to decide what to continue or repair.
- How does Search evidence differ from Shopping evidence?
Search begins with actual terms, keywords, the ad promise, and Final URL. Shopping begins with Merchant Center item ID, SKU, applicable identifiers, price, stock, and product issues. Relevant Search clicks do not validate product data.
- When should PMax stay off the main budget?
Keep it on hold when conversion value, currency, transaction ID, product data, assets, brand or URL scope, or channel results remain unclear. Search themes and audience signals help guide the system; they are not a guarantee of strict delivery boundaries.
- What does the tumbler practice teach?
The same product can need a different start when queries are clear, the feed is mature, automation inputs are ready, or the use case needs explanation. Identify the available evidence and holds; an example answer is not approval of your account.
- How much budget should the first campaign receive?
There is no universal amount here. Budget should support the defined scope, sample condition, and review window. If spend is fragmented, settings change frequently, or nobody reviews queries and channels, narrow the terms, product scope, or geography first.
- What should I do after this lesson?
Carry the map into account and tracking checks. Search then leads to a first Search campaign and query control; Shopping leads to Merchant Center and feed work. Later PMax work covers conversion-value quality and scale readiness, while education needs creative, page fit, and a later purchase test.
Choose the next task by the gap
The next lesson verifies account setup and conversion tracking. A Search route then moves into a first Search campaign and query control. A Shopping route carries SKU, GTIN, item ID, price, stock, and issue status. Later PMax work reviews conversion value, product groups, brand, page, and channel scope. An education route carries asset purpose, page, audience source, visit quality, and the later purchase test.
Series start
Repair the gap before expanding