Google Ads Demand Gen: When to Test It and How to Scope Creative and Measurement

Proposed title: Google Ads Demand Gen: When to Test It and How to Scope Creative and Measurement Proposed canonical path: /en/blog/google-ads-demand-gen/ Category: Google Ads

Demand Gen is worth testing when a business needs to introduce or advance an offer before someone is actively searching, has a credible landing experience, and can measure an action that matters. It is not an automatic replacement for Search and it cannot promise sales. A responsible first test is deliberately narrow: one decision moment or audience, one offer, a controlled set of creative angles, and one verified primary conversion.

Demand Gen uses visual ads across Google environments including YouTube, Shorts, Discover, Gmail, Maps and the Google Display Network. Google combines available signals and creative assets to deliver ads, but the advertiser still has to supply an honest message and a sound conversion goal. See Google’s About Demand Gen campaigns.

For editorial context only, Ahrefs estimated the Morocco volume for “Demand Gen Google Ads” at 10 on 22 October 2026. That is a search-demand estimate, not a forecast of ad results, traffic, leads, revenue, or rankings.

Make the go-or-no-go decision

Conditions that make a test useful

Demand Gen is a reasonable candidate when the job is to make an unfamiliar offer understandable, show a product in use, or continue a journey after an earlier interaction. The business should be able to express the problem, the practical benefit, the available proof, and the next action in a short visual sequence. It is more likely to produce usable learning when the ad and destination make the same promise.

Check the post-click journey as carefully as the media plan. A compelling image cannot compensate for an unclear page, an inaccessible mobile form, or a sales team that has no way to process incoming enquiries. A structured Google Ads service can connect the campaign question to the offer and landing experience rather than treating traffic as the outcome.

Conditions that call for a pause

Delay the test if there is no meaningful action to record, no distinct visual material, or no owner for reviewing lead quality. It is also unhelpful to package incompatible objectives—such as brand awareness, catalogue sales, and high-intent appointment requests—into one undifferentiated experiment, then credit or blame the campaign type for every movement.

Where people are already expressing a specific need through search, Search may be the more direct response to that need. Do not reallocate a working channel simply because Demand Gen is available. Define the decision the experiment must inform and the real alternative it will be judged against.

Define a test that can teach the team something

Turn a marketing idea into an observable hypothesis

State the relationship between audience, message and action in plain language. For example: “For people who viewed this product category, a brief use-case video will produce more verified information requests than a static price-led message.” It is a question to investigate, not a predicted outcome.

Choose a single primary conversion that fits that question. For a store, this could be a correctly captured purchase. For a service business, a lead is only useful when its definition is stable: a submitted enquiry, a call meeting an internally chosen threshold, or a booked consultation. Page views, product views, and basket additions can remain observation signals when they do not represent the result the business wants bidding to pursue.

Bound the commercial context

Start with one service line, product family, or decision stage. Customer lists, site visitors, video viewers and lookalike segments may provide direction, but none of them repairs a vague offer. Re-check current account options and Google’s Demand Gen campaign documentation during build, because controls and placement availability can change.

Assign every ad group a destination and call to action that match its promise. Sending a specific product or problem-led ad to a generic home page discards the context paid media just created. If the destination or events are not reliable, assess them through Tracking & Analytics before interpreting clicks as evidence of business value.

Build a creative system rather than one hero ad

Give each asset a focused job

Demand Gen creative has to do early explanatory work. Develop a handful of angles that all remain true to the offer: a product demonstration, a relevant use case, an objection handled, a visible process, or a verifiable piece of proof. Each variation should contain one understandable visual idea, a clear subject, and an action that the landing page can honour.

Google recommends using both video and image assets and providing at least three images or videos in each vertical, square, and landscape ratio per ad group to expand eligible reach and engagement. That is creative-preparation guidance, not a performance guarantee. Read the official Creative Excellence Guide for Demand Gen.

Design for ratios and placements from the brief onward

Do not rely on a late crop of a text-heavy horizontal visual. Google’s image guidance includes recommended dimensions of 1200 × 1200 for 1:1, 1200 × 628 for 1.91:1, 960 × 1200 for 4:5, and 1080 × 1920 for 9:16; 9:16 is recommended for YouTube Shorts. Video assets have their own duration and safe-zone considerations. Confirm technical requirements and format eligibility in the Demand Gen asset specifications and guidelines.

Include the logo, headlines, descriptions, final URL, and brand treatment in the creative brief. A polished image can still fail as advertising if the value, brand or next step vanishes when it is resized. Sensitive categories and individual placements can have restrictions, so policy and eligibility checks belong before, rather than after, production.

Set up measurement around the business decision

Validate the conversion path before using it for optimisation

Google Ads can measure website actions, app actions, calls and offline conversions. Depending on the measurement design, a conversion action can use Google Analytics or the Google tag. Before launch, run the journey yourself: arrive with the expected parameters, make a consent choice where required, submit the form or complete the purchase, confirm the event, and check the relevant reporting surfaces.

Google distinguishes primary conversion actions, which can steer bidding, from secondary actions used for observation. The distinction matters. If a lightweight action is made primary even though it does not represent commercial value, automated bidding may learn toward an unhelpful outcome. Google explains the available sources and setup considerations in About conversion measurement.

Connect lead capture to lead quality

For a longer sales process, define how campaign information reaches the CRM: a click identifier or documented source convention, then simple statuses such as received, qualified, meeting, opportunity, and known outcome. Offline conversion information or CRM reconciliation can improve the decision record, but it cannot remove sales-cycle delays, missing contact details, or inconsistent human follow-up.

Agree the review window, deduplication approach, and status owner before traffic starts. Consent choices, browser settings and privacy controls may limit what can be observed. Provide users clear information and obtain consent where required by applicable law or Google policy; that operational point is not legal advice.

Read performance in layers

Separate delivery, attention and business response

Begin with delivery: spend, impressions, clicks and policy or asset diagnostics. Then examine engagement in the context of its format, followed by the primary conversion and the quality signal available. A high click-through rate does not establish that leads are useful. A low click-through rate is not, by itself, proof that a creative idea has failed without understanding where it served and what action it was asked to trigger.

Demand Gen reporting can be segmented by network and ad format. Google notes that the current “in-feed” portion aggregates traffic from Gmail, Discover and YouTube in-feed, so it should not be described as a precisely separate channel. Asset, channel and format reports should inform the next creative hypothesis, not be presented as final causality. See Demand Gen campaign metrics and reporting.

Change an interpretable variable

Before launch, record what will justify a decision: replacing an asset, reworking a message, reviewing lead quality, or stopping the pilot. Do not simultaneously change audience, destination, creative, budget and bidding goal; the resulting movement will have no clear explanation. Google notes that Smart Bidding has a learning phase and that frequent changes to budget, targets, or conversion goals can reset it.

Where volume and account conditions permit, a creative A/B experiment can provide a cleaner head-to-head comparison than ongoing editing. Even then, the result belongs to the tested audience, dates, assets and setup. It does not prove that the same message will work for every audience, offer or channel.

Limits to keep visible

Demand Gen cannot observe every exposure or every part of a person’s decision journey. Engagement definitions vary by format, and a portion of a video watched is not automatically a site visit or buying intent. Google Ads, Analytics, and CRM totals can differ because their identifiers, attribution rules and time windows differ. Investigate those mechanics before adding numbers together as if they measured exactly the same thing.

Reporting does not replace product quality, inventory, pricing, sales follow-up, or the post-click experience. Google may make experimentation or lift tools available under particular eligibility conditions, but availability should be verified in the relevant account. Applicable advertising policies and insufficient eligible assets can also constrain delivery.

The useful question is therefore not “Is Demand Gen good?” but “Does this scoped test support a decision to expand, adapt or stop?” If the scope needs to reflect your data and resources, request a quote without assuming an outcome in advance.

Frequently asked questions

Does Demand Gen replace Google Search?

No. Search primarily answers intent expressed in a query, while Demand Gen works across visual browsing and content environments. They may complement one another when each has a defined role, offer and measurement plan.

Is video required to launch?

A campaign can use image assets in eligible formats and placements, but Google recommends combining image and video assets to cover more formats. Do not create video just to fill a field; it still needs to explain the offer clearly and meet the specifications.

Which conversion should a lead-generation business use?

Use the action closest to the value the business wants and can verify. An enquiry may be primary when qualification rules are consistently applied. A page view is usually an observation signal unless the stated campaign goal is specifically site traffic.

Should Google Ads conversions exactly match the CRM?

They can be reconciled, but they should not be assumed to match exactly. Attribution, reporting windows, consent, duplicate handling and sales-status definitions can differ. Create the reconciliation convention before reaching a conclusion.

How many creatives should a first test include?

Start with enough variation to cover the ratios and a small number of readable message angles. Google recommends at least three image or video assets in each vertical, square and landscape ratio per ad group. Quality and relevance matter more than filling the account with near-duplicates.