Google Ads Demand Gen: a data guide for small businesses

Contents

• Where Demand Gen fits in your Google Ads setup
• Demand Gen or Performance Max: which one to pick for your stage
• The formats and creatives that actually make Demand Gen perform
• What budget to plan for a profitable Demand Gen launch
• Measuring the real profitability of Demand Gen in GA4
• The mistakes that sink a Demand Gen campaign at a small business
• Deciding whether Demand Gen deserves a spot in your budget this quarter
• Before you spend a euro on Demand Gen, get your measurement right
• Frequently asked questions

Where Demand Gen fits in your Google Ads setup

Demand Gen sits at the top of the funnel, where your future customers aren't typing any query related to your product yet. That's exactly what sets it apart from Search, which harvests demand that's already been expressed.

When someone searches for "emergency plumber Lille," they have a conscious, immediate need. Your Search ad answers an existing intent. Demand Gen works the other way around: it shows a visual or video ad to someone watching a DIY video on YouTube, with no active purchase intent. You create demand instead of capturing it.

That logic has a direct impact on your expectations. A Demand Gen click almost never turns into a same-day sale. The journey spreads across several sessions, often several days. I never start turning this format on until an advertiser's Search is already profitable and stable, because otherwise you're spending budget on awareness before you've even captured the easy money.

Where do Demand Gen ads appear?

Demand Gen ads serve on four Google surfaces: YouTube (including the In-Feed stream and Shorts), Discover (the personalized article feed in the Google app) and Gmail (in the Promotions tab). According to the Google Ads help page on Demand Gen campaigns, these placements add up to a massive audience, with YouTube alone topping more than 2 billion signed-in users every month worldwide. According to DataReportal, YouTube ranks among the most-used platforms in the world, with an addressable ad audience counted in billions of users.

How does it differ from Search and the Display network?

Demand Gen differs from Search in the kind of demand it addresses: latent versus existing. Against classic Display, it leans on native creatives, more integrated into the content, and on lookalike audiences (profiles similar to your current customers). Traditional Display is still often seen as an intrusive banner, whereas Demand Gen aims for a subtler visual fit, close to the editorial experience of YouTube or Discover.

For an owner handling acquisition in-house, keep one simple image in mind: Demand Gen fills the tank, Search empties it. Flipping the order is like watering before you've dug the hole. If your priority right now is to grow your website traffic with the right levers, start with existing-demand channels before you open up latent demand.

Demand Gen or Performance Max: which one to pick for your stage

The right choice depends on your goal and your volume of conversion data, not on how new the format is. Performance Max targets conversion across all Google networks, Demand Gen targets awareness and discovery on visual surfaces.

Performance Max (often shortened to PMax) is an automated format that serves everywhere: Search, Shopping, Display, YouTube, Gmail, Maps. Its engine optimizes toward a conversion value and needs volume to work. Without enough conversion history, its algorithm spins its wheels. Demand Gen, by contrast, focuses on four visual surfaces and accepts goals higher in the funnel, such as page views or add-to-carts.

What's the difference between Demand Gen and Performance Max?

Performance Max chases the widest possible conversion across the entire Google ecosystem; Demand Gen aims to generate demand on discovery surfaces. PMax arbitrates between Shopping and Search to chase the purchase wherever it is. Demand Gen never touches the search network: it works the audience upstream, before it knows you.

Can you run both at the same time?

Yes, and it's often recommended once your account is mature. PMax harvests conversion, Demand Gen feeds the top of the funnel that will fuel your future conversions. The risk to watch: overlap on YouTube and Gmail, where the two formats can compete for the same placements. Google handles part of that arbitration, but you need to keep their goals distinct so you don't count the same conversions twice.

My prioritization rule is simple. Below €1,000 to €1,500 of already-profitable monthly Search spend, I touch neither Demand Gen nor PMax. Between that threshold and roughly €3,000, I open PMax to deepen conversion. Demand Gen comes after, once the margin freed up lets you invest in awareness without weakening the acquisition that pays the bills. McKinsey's work on marketing performance is a reminder that advertisers who invest in the top of the funnel without a solid capture base dilute their ROI rather than build it. Forrester analyses confirm that coordination between discovery and capture channels weighs more on returns than the choice of any single format.

The formats and creatives that actually make Demand Gen perform

For a small business with little audience data, the quality of your visual creatives drives most of the performance, well ahead of targeting signals. That's the opposite of what instinct suggests, which pushes you to spend hours on audiences.

The official documentation leans heavily on audience signals and lookalike audiences. That's not wrong for a large advertiser with thousands of conversions to work with. But when you start with thin history, the algorithm has almost nothing to chew on from the audience side. What makes the difference then is the strength of the image or video that shows up. A flat creative doesn't convert, even perfectly targeted.

What Demand Gen ad formats are available?

Demand Gen offers three main formats: single image, multi-image carousel and video (ideally vertical and short for Shorts). According to the Google Ads Developers technical documentation, you combine assets (images, headlines, descriptions, logos) that the algorithm assembles depending on the placement. The 9:16 vertical format has become the standard for grabbing attention on Shorts and in the mobile feed.

How many creatives to plan for launch?

Plan at least 5 to 8 distinct visuals and 2 to 3 short videos from launch, to give the algorithm something to test. A single visual chokes learning: Google can't compare what works. Vary the angles, the hooks and the formats (square, vertical, horizontal) to cover every surface without rough automatic cropping.

My advice here: never launch Demand Gen with a single banner hastily cropped from a Facebook ad. That's the surest way to burn budget without a usable signal. Better to be two weeks late with native creatives built for the vertical format. The analysis in Think with Google's resources on video creative confirms that the first three seconds of a video weigh more than any targeting setting when it comes to holding attention. Nielsen Norman Group studies on visual attention show that users judge content in a few seconds before deciding whether to keep going or move on.

What budget to plan for a profitable Demand Gen launch

Plan a daily budget of at least 10 to 15 times your target conversion value so the algorithm can exit the learning phase. Below that threshold, the campaign stalls and produces no reliable data.

The logic is mathematical. Google's automated bidding needs a minimum volume of conversions to calibrate its model. If your target conversion is worth €40, a daily budget under €400 to €600 slows learning down. For a small business, that looks high, but the math has to be read at the scale of the campaign, not of a single day. Aim for a more modest conversion value, a lead at €15 for instance, and the threshold drops proportionally.

Is a €10/day budget enough on Google Ads?

For Demand Gen, €10 a day rarely gets you out of the learning phase on a mid-value conversion. That budget suits a pure awareness test, with no expectation of a measurable conversion in the short term. If you're after trackable sales or leads, aim for at least €30 to €50 a day, set against your actual conversion value.

How long does the learning phase last?

The learning phase usually lasts 7 to 14 days, the time it takes for the algorithm to gather enough conversion signals. Google recommends avoiding any major change (budget, bids, creatives) during this period, because every change resets the counter. Waiting is counterintuitive for an owner in a hurry, but cutting too early is like judging a sowing before anything sprouts.

1. Calculate your target conversion value Multiply your average conversion rate by your basket or lead value to set the budget's starting point.
2. Index the daily budget Aim for 10 to 15 times that conversion value per day to feed learning properly.
3. Lock the settings for 14 days Don't touch bids or creatives during the learning phase, or you'll restart it.
4. Check conversion volume If the campaign generates fewer than 15 to 20 conversions per week, the budget is probably below the critical threshold.

One thing I always point out: budget isn't judged on its absolute size but on how well it fits the value you're chasing. Plan half a day per quarter to recalibrate these thresholds cleanly in-house. The Lysible plans start from there: analysis time, not raw data, is the real cost for a small operation.

Measuring the real profitability of Demand Gen in GA4

The raw ROAS shown in Google Ads almost always understates Demand Gen, because it misattributes multi-session conversions. The real contribution shows in GA4, through assisted conversions and conversion paths.

Here's the mechanism. Demand Gen reaches a user who, three days later, types your brand into Google and converts via Search. In Google Ads, that sale is credited to Search, not to Demand Gen, which then looks unprofitable. You risk cutting a campaign that was in fact feeding your funnel. It's the costliest reading mistake on this format.

Why is the ROAS shown in Google Ads misleading?

The ROAS shown in Google Ads mostly reflects direct conversions, not Demand Gen's assisted contribution to longer journeys. Because this format works upstream, its impact often materializes on another channel, later. Raw ROAS captures that priming role poorly. To isolate it, you have to leave the Google Ads interface and cross-reference the data in GA4.

Which GA4 reports reveal the real contribution?

Two GA4 reports reveal the real contribution: the "Conversion paths" report and the attribution model comparison. According to the Google Analytics help on conversion paths, the first shows every touchpoint before a conversion, including assisted views. There you can see how many times Demand Gen appears early in the journey. The model comparison pits "last click" attribution against "data-driven" attribution, and that's where Demand Gen often regains value. The open-source alternative Matomo offers comparable multi-touch attribution reports, useful if you keep full control of your data.

My reading framework comes down to three moves. One: I look at GA4 ROAS in "data-driven" attribution, never Google Ads ROAS alone. Two: I pull the conversion paths report and count how many converted journeys contain a Demand Gen touchpoint. Three: I compare the conversion rate of users exposed to Demand Gen against the others. Cross-referencing the two interfaces takes about thirty minutes a month by hand, or it runs continuously if you automate the tracking of these metrics.

This measurement work applies to any channel that acts upstream of the conversion, not just Demand Gen. If you want to master this reading, the traffic statistics worth actually tracking lay the groundwork before you get into fine-grained attribution.

The mistakes that sink a Demand Gen campaign at a small business

The top cause of failure isn't the format itself, but three management mistakes: misconfigured attribution, budget below the learning threshold, placements not excluded. Each one is enough to make your results lie.

Misconfigured attribution comes first. Judging Demand Gen by last-click ROAS is like condemning an awareness channel with a conversion channel's scorecard. Many campaigns cut "because they weren't profitable" actually were, read on an assisted basis. Budget below the threshold comes right behind: a campaign that never exits learning only produces noise, not a usable signal.

The third pitfall is quieter. Without placement exclusions, your ads can serve on content or apps unrelated to your target, sometimes mobile games or irrelevant videos. You pay for impressions that reach no one useful. Checking the placements report every week during the first month prevents that leak.

Should you cut a campaign that isn't profitable at 30 days?

No, 30 days is rarely enough to judge Demand Gen, especially if the purchase journey spreads over several weeks. Before cutting, check three things: has the campaign exited learning, does the budget reach the minimum threshold, and have you read ROAS in data-driven attribution in GA4. If all three conditions are met and the assisted contribution stays at zero, cutting is justified.

We regularly see advertisers run Demand Gen without having verified their conversion tracking upstream. If your tracking is shaky, no attribution reading will hold. That's why I always set up tracking via Google Tag Manager and consent before opening a latent-demand format. The CNIL is a reminder that collecting consent conditions the gathering of measurement data, and therefore the very reliability of your attribution. Without a clean measurement foundation, Demand Gen becomes a blind bet.

Deciding whether Demand Gen deserves a spot in your budget this quarter

Demand Gen deserves a spot in your budget only if your Search is already profitable, if you have margin to invest in awareness, and if you accept a horizon of several weeks of patience. Without these three conditions, wait.

The reasoning is done in risk and margin, not in media jargon. As long as your Search isn't capturing all the qualified demand that exists, every euro logically goes to it: it's the easiest money. Opening Demand Gen before that point means funding discovery while you leave obvious sales on the table.

At what ad revenue should you test Demand Gen?

Test Demand Gen once your Search spend exceeds about €1,500 a month with a stable, positive ROAS. Below that, focus your budget on capturing existing demand, far more profitable in the short term. The Direction générale des entreprises notes that small operations gain first from mastering their profitable channels before diversifying their acquisition. The France Num program observes that small businesses progress more by consolidating their existing digital tools than by multiplying channels all at once.

A business with a seasonal catalog should also schedule this test outside its sales peaks, when existing demand no longer fills the order book. For a shop whose SEO traffic already works the top of the funnel, data-driven e-commerce SEO can make Demand Gen redundant. Measure before you decide.

Before you spend a euro on Demand Gen, get your measurement right

By now you know this format isn't judged in Google Ads alone, nor at 30 days, nor on raw ROAS. The only verdict that holds is read in GA4, in data-driven attribution, counting assisted conversions. And cross-referencing your interfaces is exactly what costs a small operation time, month after month. Centralizing your Google Ads, GA4 and Search Console data into a single reading lets you see each channel's real contribution without pulling three reports by hand. That's what Lysible does for owners who run acquisition in-house: you decide to cut or keep a campaign on consolidated data, not on a hunch.