Search vs Performance Max: Which Campaign Fits an SME?
Search, Performance Max and Demand Gen are not three rungs on a ranking ladder. They solve different problems: Search responds to demand that people have already put into words, Performance Max optimises a conversion goal across multiple types of Google inventory, and Demand Gen connects visual messages with audience and channel decisions. For a small or medium-sized business, the right question is therefore not “Which campaign type is best in general?” but “Which campaign can test our next business hypothesis most cleanly with the budget, data and creative assets we actually have?”
This guide provides a campaign selection matrix for SMEs. It does more than assign campaign types: every decision requires evidence and receives a status. Use STOP when the measurement or offer foundation is missing, START for a justified primary campaign, TEST for a controlled expansion, and SCALE only after the overall effect has been confirmed. The matrix is neither an automatic budget calculator nor a promise of results. Its purpose is to expose the precise condition on which a campaign idea stands or falls.
Start with demand, not with the campaign name
Three business questions to answer before opening Google Ads
First determine whether people are already actively looking for the offer. A specific search such as “bathroom renovation Wiesbaden”, “tax adviser for a limited company” or an exact product name presents a different starting point from a new product whose category is barely known. Google’s official Search campaign setup guide describes Search as a way to reach people while they are looking for products or services. That does not automatically establish purchase readiness, but it does provide an observable language-based signal that an SME can examine through keywords, search terms, ads and landing pages.
The second question concerns the desired result. A submitted form is a useful objective for a local service company only if it represents a genuine enquiry. For an online store, revenue, margin or contribution profit may matter more than order count alone. The third question is how much search and channel breadth the business can permit. Search offers a narrower starting point. Performance Max looks across channels for conversions or conversion value. Demand Gen is appropriate when a visual message and a defined audience hypothesis are themselves part of the test.
This produces the matrix’s first rule. If clear search demand, an appropriate landing page and a verifiable objective all exist, Search will often receive START. If existing demand is insufficient and the business has reliable goal signals and suitable assets, Performance Max may be added as a TEST. If the market first needs to see or understand the benefit, Demand Gen may run the initial controlled demand test. If either the offer or conversion measurement is unclear, all three options receive STOP.
Search, Performance Max and Demand Gen serve different roles
A comparison table creates a shared language, but it does not grant approval
All three campaign types can support top-level objectives such as leads or sales, but they pursue them through different selection mechanisms. A team should therefore not simply place the Conversions, Cost and ROAS columns side by side. It must also record the type of demand, the reach and the control included in each option. Only then can it tell whether two campaigns genuinely perform the same task or address different parts of the demand journey.
| Criterion | Search | Performance Max | Demand Gen | Decision signal | Typical wrong decision |
|---|---|---|---|---|---|
| Primary task | Capture expressed search demand | Optimise a business goal across channels | Build attention and drive action on visual surfaces | Where does the specific business opportunity arise? | Compare every campaign only by the last click |
| Selection signal | Search query, keyword, ad and landing page | Conversion goals, values, assets, feeds and signals | Audience, creative, format and channel | Is demand already visible in language? | Treat Performance Max as a larger Search campaign |
| Reach | Google Search and other Search Network surfaces; Search Partners are configurable, and enabling Display Network expansion adds Display inventory | Google inventory including Search, Shopping, YouTube, Discover, Gmail, Display and Maps | Visual and entertainment-oriented Google surfaces | How widely may the system look for results? | Equate more reach automatically with additional demand |
| Control | Keywords, exclusions, ads, landing pages and settings | Goal, brand, URL, search and asset controls | Audiences, creatives, formats and channel controls | Which control is indispensable to the business? | Confuse signals with guaranteed restrictions |
| Creative foundation | Relevant copy, assets and a consistent landing page | Channel-ready copy, images, logos, video and, where relevant, a feed | Distinct image, video or carousel hypotheses | Can the team supply genuine creative variants? | Treat mandatory assets as a strategy in themselves |
| Decision evidence | Search terms, lead quality and commercial viability | Improvement to the total portfolio while guardrails remain stable | Downstream quality and, where suitable, lift | What evidence determines the next stage? | Present attributed platform performance as causal proof |
The table deliberately does not declare a winner. A business may need Search for explicit service enquiries, Performance Max for additional profitable reach and Demand Gen for a new visual proposition at the same time. It may also only be able to fund one campaign. In that case, the winner is not the campaign type with the broadest feature set, but the hypothesis that is both most commercially important and most tightly testable.
When Search is the right primary campaign
Explicit intent, clear language and necessary control
Search is often the first choice when customers already describe their problem, the service, the location or the product in specific terms. This is particularly true for local services, urgent needs, tightly defined B2B services and product ranges with stable category searches. An SME can follow the enquiry from the wording of the search to the landing page, inspect search terms and align messages closely with the need. This does not make Search automatically efficient, but it makes the campaign diagnostically valuable: poor enquiries can more readily be traced to search terms, ad promises, locations or landing pages.
Audiences in Search require precise interpretation. Google’s distinction between Targeting and Observation shows that Observation does not narrow reach; it makes additional criteria observable within the campaign’s existing delivery. An audience in Observation therefore proves neither that only those people were reached nor that the audience caused the search intent. The search query remains the central context. Audience data supports diagnosis and may also provide signals to Smart Bidding.
Search and Performance Max also do not behave like two completely isolated machines within the same account. For the same query and domain, Google’s official prioritisation rules for ad groups and asset groups apply a sequence. An eligible exact-match keyword whose text is identical to the query is considered first; a corrected spelling counts as identical, while plurals and synonyms do not. At the next level, identical phrase- or broad-match keywords and identical PMax search themes have equal standing. If there is no identical candidate, Google first selects by AI-based relevance; Ad Rank decides between candidates at the same priority level. The match-type label “exact” is not sufficient on its own because exact match may also cover queries with the same meaning or intent. Low search volume, disapproved ads or destinations, unmet targeting or a budget constraint may also remove an expected Search keyword from consideration. This sequence does not apply to Shopping formats or other Google inventory. The matrix gives Search START because it has a verifiable job to do, not because it promises an exclusive claim on search queries.
When Performance Max can add value or scale
Channel breadth requires a reliable goal and suitable inputs
Performance Max is a goal-based campaign type that provides access to Google inventory from a single campaign. Google explicitly presents it as a way to complement keyword-based Search campaigns and find additional conversions or conversion value. The Performance Max overview covers conversion goals and values, assets, audience signals, data feeds and campaign controls. It does not offer manual selection of individual channels or fixed budget shares by channel; Google allocates campaign budget across available inventory, while the channel report diagnoses the resulting distribution after delivery. An SME is not simply handing over a budget; it is providing an objective function and the materials from which the system can assemble potential ads and placements.
A TEST makes sense when the primary goal is measured correctly, its commercial value has been defined and the business does not need to restrict delivery to one channel. For ecommerce, a clean product feed can provide a strong foundation. For lead generation, forms, calls or appointments must be filtered well enough that Smart Bidding does not favour cheap but worthless contacts. Assets must fit both the service and the destination. Final URL Expansion, automatically created assets, and brand and search controls should be deliberately approved or constrained.
Search themes are optional, supplementary context signals rather than conventional keywords or reach boundaries; an asset group can contain up to 50 distinct themes. For prioritisation within an account, however, identical search themes behave like identical phrase- and broad-match keywords. Google’s description of how search themes work explains that they can provide context that the website, assets or feed do not communicate clearly. Removing a theme does not automatically exclude the corresponding searches. Audience signals are likewise optimisation cues rather than guaranteed targeting. A business that launches Performance Max merely to copy the same keyword list into another field has not yet articulated a distinct PMax hypothesis.
Where an established Search structure provides the baseline, a controlled comparison is better than uncontrolled competition between campaigns. Performance Max experiments may, subject to eligibility, examine questions such as the uplift from adding Performance Max, certain upgrades or optimisation changes. Availability and evidential strength depend on campaign type, volume and sound experimental design. Without a suitable experiment, a portfolio comparison remains observational; even then, the decision should focus on the total effect rather than whichever individual campaign reports the more attractive number.
When Demand Gen supplies the missing layer
Creative and audience become a distinct business hypothesis
Demand Gen fits when a benefit needs to be demonstrated visually, the category is not yet searched at sufficient scale or a business wants to test a creative proposition beyond conventional search results. Google’s Demand Gen overview distinguishes the roles directly: Search maximises performance in search, Performance Max across Google channels, and Demand Gen on especially visual and entertainment-oriented surfaces. Demand Gen may optimise towards clicks, conversions or conversion value and should not be reduced to an awareness-only product; the bidding strategy actually available depends on the campaign objective and setup.
Launching requires more than a folder of images. The business needs a hypothesis about which audience will recognise the problem, which creative will communicate the benefit and which downstream action will demonstrate success. The Demand Gen campaign setup guide connects the objective, conversion goals, devices, languages, locations, bidding, audiences, ads and, where relevant, product feeds. A selected audience is not a hard reach boundary in every setup: with Optimized targeting enabled, Demand Gen may serve beyond the selected audience and demographic signals. If audience, creative, bidding and landing page all change at once, the resulting performance can rarely be attributed to one hypothesis.
Demand Gen therefore receives TEST only when at least one testable image or video idea, a defensible audience definition, an appropriate landing page and a reliable downstream metric are available. Impressions, views or inexpensive clicks may be intermediate signals, but they do not replace a qualified enquiry, a sale or an appropriate lift result. If there is no creative system, or if the budget cannot sustain a meaningful observation period, the decision is STOP or “develop creative”, not “try the algorithm”.
Assess channel choice and creative readiness together
A format is usable only when its message works
Search also requires creative work, even though it may not need a major video production. The headline, description, assets and landing page must answer the need visible in the search query consistently. Performance Max can use channel-ready copy, images, logos, video and, where relevant, feed content. Google may automatically generate video assets when none are provided, so not every campaign requires a self-produced video. The available assets and feeds nevertheless affect which placements the campaign is eligible to serve. Demand Gen does not merely need more files; it requires deliberately varied hooks, visual treatments, formats and offers.
Demand Gen provides a specific campaign-architecture option: channel controls are set at ad-group level. A team may use “All Google channels”, with the Google Display Network as an optional addition, or manually select YouTube in-stream, YouTube in-feed, YouTube Shorts, Discover, Gmail, Maps and the Google Display Network. Available surfaces depend on factors including the campaign objective, ad type, audience and other settings; Maps requires a suitable location asset. Selection guarantees neither delivery, minimum volume nor an even budget distribution. A narrow selection is therefore justified only when the creative and user context clearly fit a surface.
Measurement readiness is the first campaign-selection gate
Conversion, value and backend status must describe the same action
A campaign type can optimise only what it receives as an objective. For leads, the business must decide whether a submitted form, a reached contact, a qualified appointment or a signed order counts. In ecommerce, purchase value, currency, transaction ID, returns and, where relevant, margin must be processed consistently. Micro-conversions may support diagnosis, but they should not quietly receive the same priority as a commercial result. Otherwise the winning campaign may simply be the one that can generate the easiest intermediate action.
Before selecting a campaign, create a measurement sheet containing the conversion action, category, primary status, counting method, value source, reporting delay, responsible system and business owner. A test conversion proves the technical transfer. A cross-check in the store or CRM proves that the same real action was captured. Lead generation needs a quality sample; sales require an order-and-value reconciliation. Only when this chain is complete may the result support START or TEST.
If that evidence is missing, the matrix status is STOP, even when the account already displays data. More reach can accelerate a measurement error, but cannot repair it. A new campaign should likewise not be misused as a diagnostic tool for an old tracking problem. First decide which figure is allowed to support a decision; then select the campaign type whose mechanism can sensibly optimise that figure.
The business model changes the appropriate starting architecture
Local services, B2B and ecommerce require different evidence
A local emergency service with clear regional searches can structure Search tightly around service, location and availability. Qualified calls or jobs matter, not clicks. Performance Max may later test additional reach once locations, schedules, goals and exclusions are sound. Demand Gen is a credible option only when an earlier or visual phase of demand has commercial relevance, for example planned renovations rather than emergency repairs.
In B2B, search volume is often lower and the sales cycle longer. Search can capture highly specific problem queries, while CRM or offline signals are needed to distinguish genuine opportunities from downloads and unsuitable form submissions. Performance Max receives TEST only after the objective represents quality rather than raw lead volume. Demand Gen can build expertise, use-case understanding or category awareness, but it needs downstream measurement through qualified visits, contacts, pipeline or a suitable experiment.
Apply the campaign selection matrix for SMEs
Move from the starting condition to evidence-based approval
The matrix combines six fields: starting situation, demand, data readiness, creative readiness, provisional decision and approval evidence. The decision remains provisional by design. Industry labels, competitor activity or Google recommendations do not make it final; the observed result does. Every row needs an owner and a review date. If the offer, tracking, feed or objective changes, the assignment must be reassessed as well.
| Starting situation | Demand signal | Data readiness | Creative readiness | Provisional decision | Approval evidence |
|---|---|---|---|---|---|
| Local service with specific search queries | High and explicit in search language | Low to medium, but verifiable | Ads and a matching service page are available | Search START | Relevant search terms, reachable contacts and qualified jobs |
| New account without a validated conversion | Unclear or arbitrary | Low | Any | All options STOP | Test conversion and backend cross-check agree |
| Store with a clean feed and reliable values | High, with additional cross-channel potential | High | Complete product and brand assets | Continue Search, Performance Max TEST | More profitable total value without harmful displacement |
| Established offer with limited Search reach | Medium to high | Medium to high | Channel-ready assets are available | Performance Max TEST | Additional qualified results while guardrails remain stable |
| New visual product with little search demand | Low | Medium | Strong image and video hypotheses | Demand Gen TEST, Search for emerging demand | Downstream quality or suitable lift rather than reach alone |
| B2B offer with a long sales cycle | Small but specific | Sufficient only with CRM feedback | Explanatory assets are available | Search START, expansion later TEST | Pipeline quality and value after the full sales cycle |
| Budget too small for three parallel campaigns | Any | Medium | Uneven | Give only one primary hypothesis START or TEST | Budget sustains the observation window and conversion lag |
| Strong brand demand but weak generic demand | Distorted by brand familiarity | Medium | Medium | Protect the brand baseline, test expansion separately | Non-brand effect and total marginal cost remain visible |
The matrix’s practical value lies in its rejections. A store with a large catalogue does not automatically qualify for Performance Max if purchase values are double-counted. A visual product does not automatically qualify for Demand Gen if only one untested image exists. A local service business does not automatically qualify for Search if its landing page promises a different region or service. The matrix assesses operational readiness, not merely theoretical suitability.
Budget constrains observation and optimisation, not according to a percentage formula
One viable hypothesis is better than three underfunded campaigns
There is no universal allocation such as “60% Search, 30% Performance Max, 10% Demand Gen” that makes sense for every SME. Budget follows the task. First define the target cost or target value, realistic demand, conversion lag, data volume and maximum tolerable risk. Then determine whether the campaign can produce enough decision-relevant observations or commercial outcomes within a useful period. A campaign whose delivery is regularly constrained by budget provides a different evidence base from a consistently funded test.
For a primary campaign, define a baseline budget that covers existing demand reliably where doing so is commercially sound. Give an expansion test its own predetermined, capped test budget and a stopping rule. This allows the team to see whether Performance Max or Demand Gen improves the portfolio as a whole without quietly starving the Search baseline. Shared budgets and frequent reallocations are problematic while the decision is being evaluated because they can blur the control and test conditions.
A small budget is not a judgement on the quality of the business, but it is an architectural constraint. If it can support only one hypothesis, start where both demand and business value are clearest. If it cannot cover the relevant conversion lag, assign STOP or extend the observation window. Scaling does not begin merely because daily budget remains unused; it begins when additional spend appears capable of maintaining result quality at acceptable marginal cost.
Read cross-campaign results without false precision
Platform reports support diagnosis but do not prove impact on their own
Search provides search-term, keyword, ad and landing-page signals. Performance Max provides views into assets, search and channels, among others. Demand Gen offers reporting by campaign, ad, audience, asset and channel. Google’s Demand Gen reporting documentation, for example, describes segmentation by network and ad format, alongside additional video and asset information. These views show where reported performance occurred; they do not automatically explain why it occurred or what would have happened without the campaign.
Use a shared assessment sheet across all three campaigns. The primary metric is the business outcome, such as a qualified order, contribution margin or confirmed pipeline. Guardrail metrics expose side effects: brand share, lead quality, returns, frequency, Search impression share, total CAC or enquiries from irrelevant locations. Diagnostic metrics such as CTR, view rate, asset performance or channel share help improve campaigns, but they must not replace the decision metric.
Every test needs exactly one decision question
A test contract prevents retrospective storytelling
Before launch, document the question, baseline, controlled change, primary metric, guardrails and decision rule. Google’s guidance on A/B experiments for Demand Gen recommends that campaigns differ in only one variable where possible; depending on the experiment, creative assets, audience, product feed or bidding may be examined. The experiment type that is available and evaluable depends on campaign status, setup and data volume. An A/B comparison between two Demand Gen variants does not automatically demonstrate incrementality relative to no advertising. If budget, audience, creative and landing page all change together, the team may observe performance but can rarely isolate one cause.
| Test question | Baseline | Controlled change | Primary metric | Guardrail metric | Decision rule |
|---|---|---|---|---|---|
| Does Performance Max add to a stable Search structure? | Documented Search baseline | PMax with its own budget and the same business goal | Additional qualified value across the portfolio | Search quality, brand share and total cost | Scale only if the total result improves on the baseline |
| Does Demand Gen reach new relevant users? | Existing demand-capture channels | One creative–audience hypothesis | Qualified downstream action or suitable lift | Frequency, bounce rate and lead quality | Iterate the creative or end the test |
| Is expansion beyond brand worthwhile? | Separately documented brand performance | Defined non-brand or cross-channel test | Additional qualified demand | Brand cost and total marginal cost | Do not scale a mere displacement of demand |
| Does value-based bidding improve commercial performance? | Validated conversion values | Change only the bidding logic | Profitable conversion value | Volume, margin and value quality | Correct values before judging bids |
| Which visual proposition works better? | Control creative asset | One defined creative variant | Downstream conversion or agreed lift metric | Cost, frequency and quality rate | Select a winner only after sufficient data maturity |
The test contract does not prevent operational action in response to genuine risks. Incorrect prices, unavailable products, policy problems or unqualified leads may be corrected immediately. The change must, however, be logged, and the assessment window should restart if its effect is material. A “winner” without a stable baseline can at most remain in TEST; it cannot move to SCALE.
The decision path moves from STOP to SCALE
Seven gates replace a blanket campaign recommendation
Automated bidding needs time to calibrate after launch or a material change. Google’s official explanation of the learning period identifies conversion volume, conversion cycle and bid strategy as important influences; depending on available data, calibration can take up to three weeks or one to two conversion cycles. That is neither a mandatory waiting period for every test nor a promise of stability. It is a reason not to change decisions daily and to align the observation window with the real sales cycle.
- Measurement gate: Has the primary conversion been confirmed technically and commercially? If not, apply STOP and repair tracking and backend reconciliation first.
- Economics gate: Are target cost, value or quality criteria defined? If not, apply STOP, because no campaign can receive defensible approval.
- Demand gate: Is there specific, relevant search demand with suitable landing pages? If yes, give Search START for that task.
- Breadth gate: Should the account seek additional conversion performance across Google inventory, and are the goal, data and assets sound? If yes, give Performance Max TEST.
- Creative gate: Must demand be developed visually, and is there a testable audience–creative hypothesis? If yes, give Demand Gen TEST.
- Impact gate: After an appropriate conversion lag, does the test improve the primary metric without breaching its guardrails? If not, iterate or apply STOP.
- Marginal-value gate: Do quality and marginal cost remain viable under a controlled budget increase? If yes, apply SCALE; otherwise return the campaign to TEST.
The path may produce different results for different parts of an offer. Search may receive SCALE for a core local service while Performance Max remains in TEST for a wider product range and Demand Gen receives STOP because the necessary creative is missing. Statuses therefore belong to a specific hypothesis, not permanently to a campaign type.
STOP rules prevent expensive false confidence
When a campaign should not launch or continue scaling
Apply an immediate STOP if the primary conversion does not represent a real business action, is double-counted or does not pass a reliable value. The same applies to incorrect prices, unavailable services, unresolved data rights or a landing page that fails to fulfil the ad promise. More budget does not produce better knowledge under these conditions. It merely increases the volume of faulty signals or poor user experiences.
Apply a test STOP if guardrails are materially breached: lead quality falls, returns rise, irrelevant locations dominate, brand share displaces the expansion being tested, or repeated interventions prevent the campaign from completing a stable observation window. A seemingly attractive campaign CPA is also insufficient if the account as a whole gains no additional qualified demand. Before launching Performance Max, the business should therefore establish when the campaign type makes sense for its specific situation rather than treating it as a standard upgrade.
STOP does not necessarily mean a permanent shutdown. It may trigger “repair measurement”, “develop new creative”, “clarify the offer”, “concentrate the budget” or “correct the test design”. Make a new decision only after the missing evidence exists. This reversible process prevents both extremes: ending a campaign too early through impatience and continuing to fund it indefinitely through hope.
Put the choice into a 90-day approval plan
Four phases connect the baseline, test and scaling decision
The plan does not promise that every business can scale after exactly 90 days. It creates a shared cadence for review and documentation. With long B2B cycles, the final impact assessment may come later; with high ecommerce volume, individual diagnostic questions may be answered sooner. What matters is that phases are not skipped and every material change is recorded with its date, hypothesis and a new observation start.
| Phase | Objective | Core actions | Required evidence | Gate |
|---|---|---|---|---|
| Days 0–7 | Confirm operational readiness | Review goals, values, tracking, CRM, feed, creative, landing pages and demand | Test conversion, backend cross-check and documented matrix | STOP or approval |
| Days 8–30 | Stabilise the primary architecture | Launch Search or preserve the existing baseline without major parallel changes | Search terms, quality, cost and conversion lag are traceable | Confirm START |
| Days 31–60 | Examine exactly one expansion | Launch Performance Max or Demand Gen with a separate test contract | Primary and guardrail metrics after a mature data window | Continue, correct or stop TEST |
| Days 61–90 | Assess total impact and marginal value | Review brand, channel displacement, quality, margin and budget response | Improvement over baseline with documented limitations | SCALE, further TEST or STOP |
Frequently asked questions about Search, Performance Max and Demand Gen
Eight answers for a defensible campaign choice
The following questions summarise the decision. Every answer remains tied to the business task, measurement readiness and approval evidence; no campaign type receives priority on the strength of its name alone.
Should a new Google Ads account start with Search or Performance Max?
When specific search demand, suitable landing pages and a validated conversion are available, Search often provides the more tightly testable starting point. Performance Max may make sense immediately when the business has a reliable objective, suitable assets or product feeds, and room for cross-channel optimisation. If measurement is not ready, neither campaign type should be approved.
Can Performance Max replace a Search campaign completely?
Performance Max can technically use Search inventory, but Google describes the campaign type as a complement to keyword-based Search. Whether a separate Search structure remains necessary depends on control requirements, search tasks, brand management and measurement design. Any replacement should be tested and judged at portfolio level, not inferred from one campaign metric.
Can Search and Performance Max run at the same time?
Yes. On Google Search, however, the selection levels described in section 3 apply. Only an eligible exact-match keyword whose text is identical to the query has first priority; identical phrase- or broad-match keywords and identical PMax search themes share the next level. If no identical candidate is eligible, Google uses AI-based relevance; Ad Rank decides between candidates at the same priority level. Eligibility constraints can remove an expected Search keyword before those rules are applied. A lower CPA in one campaign therefore does not automatically show that it generated additional demand rather than merely receiving a different allocation or attribution.
When does Demand Gen make sense for an SME?
When a visual proposition, a defined audience and a downstream action need to be tested together. That may apply to new products, offers that require explanation or categories with limited search demand. Demand Gen should not launch simply because video files exist; the creative hypothesis, destination, budget and success criteria must fit together.
How should an SME test Performance Max with few conversions?
First determine whether the low volume reflects genuine demand and the sales cycle or an overly narrow or faulty measurement setup. Do not inflate the data artificially with worthless micro-conversions. If a built-in experiment is not suitable, run a clearly documented portfolio test with a longer observation period and appropriately cautious conclusions.
How should budget be split between the campaign types?
Do not use a universal percentage formula. Adequately fund the commercially most important primary hypothesis first. Give an expansion a limited test budget without changing the baseline uncontrollably. Only confirmed marginal value justifies a lasting reallocation or increase.
How can brand cannibalisation or simple budget displacement be identified?
Observe brand and non-brand demand separately, together with total conversions, quality, cost and the original Search baseline. If the baseline loses performance to the same extent that the new campaign gains it, additional impact has not been established. A sound experiment provides stronger evidence than a simple before-and-after comparison.
When is a campaign ready to scale?
Only when a mature data window supports the primary metric, guardrails remain stable and a controlled budget increase appears commercially viable. A short ROAS spike, one large order or a creative with a strong click-through rate is not enough. If the cause remains unclear, keep TEST; if measurement is broken, apply STOP.
Next step: have the campaign architecture reviewed
Turn three campaign types into a verifiable investment plan
A sound campaign choice starts with a limited statement: Search will capture existing demand, Performance Max will expand the current portfolio under controlled conditions, or Demand Gen will test a visual demand hypothesis. The business then defines the goal, budget, assets, baseline, guardrails and stopping rule. The campaign selection matrix keeps these requirements together and prevents a product label from substituting for strategy.
An audit should therefore assess more than existing campaigns. It should establish whether the offer is already searched for or still needs explanation, whether the conversion and value match the backend, whether the creative fits the intended reach, and what evidence would justify scaling. The result is not a blanket “turn PMax on” or “turn Search off”, but a documented status for each business hypothesis: STOP, START, TEST or SCALE.