Get your Demand Gen setup reviewed before you spend
Key takeaways
Most Demand Gen campaigns are lost at setup, long before any optimisation.
Choose the conversion goal first; changing it later resets all the learning.
Leave Display off at launch; image inventory buys clicks, not real buyers.
Set your holdout and brand baseline before launch, not after results disappoint.
Raise branded Search budget with Demand Gen, or competitors collect your demand.
A Google Demand Gen campaign that works for B2B SaaS looks like this: one campaign per market, three audience ad groups, video-first creative, a single high-volume conversion goal tied back to your CRM, and a budget of at least 20x your target CPA. Then you leave it alone long enough to learn.
Most failed Demand Gen campaigns we audit were lost at setup, not in optimisation. Fyxer's first two tests ran one ad group each, with no exclusions and almost no video. They spent about £23k for 7 conversions. The rebuild used the same product, the same channel, but a proper setup. It produced 2,226 conversions and got monthly CPA down to £207. The full story is in our Fyxer Demand Gen case study.
This guide is the setup we use for B2B SaaS and AI companies, setting by setting, with the reason behind each choice.
The B2B SaaS Demand Gen setup at a glance
[table]
Setting | What we use | Why
Campaigns | One per market | Each market gets its own budget and CPA
Goal and conversion | Leads, with one campaign-specific primary conversion | Stops page views and micro-events from training the bidding
Bid strategy | Max Conversions, then Target CPA after about 50 conversions | Gives Google room to learn first
Budget | At least 20x target CPA per day | Google's own minimum for conversion bidding
Devices | All | Mobile converted as well as desktop at Fyxer
Channels | YouTube, Discover, Gmail to start | Display ads: cheap image inventory
Ad groups | Prospecting, lookalikes, remarketing | Each audience can be read and funded alone
Creative | Three to five short videos in 16:9, 9:16 and 1:1 | Video drove 90% of Fyxer's conversions
Exclusions | Customers and zero-value signups in every ad group | Stops Google re-converting people you have
Measurement | CRM import, brand search baseline, holdout plan | Last-click can't see what Demand Gen does
[/table]
Before you build: what needs to be in place
Don't open the campaign builder until these five things are true. Each one has killed a Demand Gen launch we've seen.
- Search is saturated. You've covered every keyword that converts at an acceptable CPA, and more Search volume only means a higher CPA. If not, scale Search first. Our Demand Gen vs Search guide covers how to tell.
- You have a conversion with volume. Demos are usually too rare. Pick the deepest event that still fires a few hundred times a month.
- Your CRM can see the lead. GCLID capture or enhanced conversions for leads, passed into HubSpot or Salesforce.
- You can fund learning. Fyxer spent roughly £165k before efficiency arrived. Your number will differ, but plan for a costly first month.
- You have video. Not one brand video. Several short cuts.
How much budget do you need for a Demand Gen campaign?
Google recommends a daily budget of at least 20x your target CPA for conversion bidding, with $500 a day per ad group for efficient performance and $100 as a floor. A $60 sign-up target means about $1,200 a day per campaign.
At peak, Fyxer ran £2,500 a day in GB and £1,111 a day in the US. If your budget can't clear 20x your target CPA, change the conversion event before you launch. Don't launch underfunded and hope.
Step 1: Choose the conversion goal before anything else
The conversion goal is the most important setting in Demand Gen, because it's what Google learns to find. Choose it first and don't change it once the campaign is live.
Google's Demand Gen performance guide calls a conversion goal change on a live campaign "the biggest reset to learnings." We learned that the hard way. Fyxer's spring relaunch migrated conversion goals while also changing ad groups, creative, and bidding. CPA rose to £1,776, and we couldn’t isolate the cause.
How to choose:
- Use campaign-specific goals, not the account default. Account defaults often include page views and newsletter sign-ups.
- Pick one primary conversion. Fyxer's US campaign optimised to Work Email Signup, roughly 400 a month. GB optimised to Work Inbox Connected, an activation event with similar volume and more value behind it.
- Filter for work emails if your product allows personal sign-ups. A Gmail sign-up and a company-domain sign-up are not the same lead.
- Import downstream stages as secondary conversions. You won't bid on SQLs at first, but you need to see them.
PAM AI shows why the CRM link matters. With Google Ads connected to HubSpot through GCLID, every deal traced back to a campaign, and $92.6k in spend tied to $1.27M in pipeline in one quarter (PAM AI case study). Without that link, we only had Google's word for it.
Step 2: Set campaign-level settings
Goal: Choose Leads. It filters the options to lead-gen conversion goals.
Devices: Leave all devices on. B2B teams often assume desktop wins. At Fyxer, mobile took 73% to 83% of spend at CPAs in line with desktop, and a desktop-only Max Clicks test spent about £7k for 4 conversions. Connected TV took 3% to 6% of spend at £327 to £462 CPA, so we left it on.
Location and language: Demand Gen sets these at ad group level by default. We run one campaign per market instead of mixing countries. Fyxer's GB and US campaigns used the same structure. By January, GB ran at £146 CPA and US at £264. Combined, Google would have pushed budget to GB, and the US would never have had a fair test.
Bidding: Start on Maximize Conversions, or Target CPA if you already know your CPA from earlier data. Google says to allow about 50 conversions of learning before changing bidding. Fyxer's GB campaign ran Target CPA and the US ran Max Conversions from launch through February with no goal changes. CPA fell every month from launch to January.
Never use Maximize Clicks. Click bidding buys the cheapest inventory on Discover and Gmail, which is mostly accidental taps.
Step 3: Fix the conversion settings most teams skip
These sit at the conversion action level, not in the campaign. They decide what Demand Gen gets credit for.
- Click-through window: Match it to your signup cycle. We used 60 days at Fyxer because buyers often watched, left, and came back weeks later. A 7-day window would have hidden most of those conversions.
- Engaged-view conversions: These count people who watched at least 10 seconds of a video ad and converted later without clicking. Keep them on, but report them separately. They're real signals, and they're also the easiest to over-credit.
- Count setting: Use "One" for leads and signups, so a person who signs up twice counts once.
- Primary vs secondary: Only the campaign's chosen goal should be primary.
Just remember that engaged-view and view-through conversions carry no GCLID. Your offline upload will credit Search for many leads Demand Gen helped create. That's why Step 8 exists.
Step 4: Choose channels deliberately
Demand Gen can now run on YouTube, Shorts, Discover, Gmail, Maps, and the Google Display Network. We start with YouTube, Discover, and Gmail and leave Display off.
Our reasoning is about creative, not channels. Display adds a large pool of cheap image placements, and image inventory was where Fyxer lost money. Image ads ran at £1,070 CPA against £280 for video, and two image sets drew 178k clicks for 54 conversions. High CTR, wrong users. Add Display later as a separate test if your core campaign is stable.
Step 5: Build three ad groups, not one
Every campaign gets three ad groups: prospecting, lookalikes, and remarketing. Each one holds one type of audience, so you can see which one is working and fund it on its own.
- Prospecting: custom segments built from your converting search terms, competitor URLs, and competitor brand terms.
- Lookalikes: seeded from paying or high-value customers, never from all signups.
- Remarketing: site visitors, YouTube viewers, and branded search themes.
Expect prospecting to look expensive. At Fyxer, GB Prospecting ran at £700 CPA while GB Lookalikes ran at £126. Prospecting built the pools the cheaper ad groups converted. Cut it, and the cheap ad groups run dry within weeks.
Leave optimized targeting off for the first month, so you know which audience drove results.
Step 6: Load creative Google can actually use
Upload three to five videos per ad group, each in 16:9, 9:16, and 1:1 where you can. Vertical matters because Shorts is a large share of Demand Gen inventory. Missing ratios limit where your ads can serve.
What worked at Fyxer:
- A mid-funnel product explainer was the highest-volume creative in both markets.
- A paid UGC workplace skit converted at £250 to £340 CPA.
- Short cuts beat long ones. The 30-second versions were 82% cheaper per signup than the full-length originals.
- Fresh creative in proven audiences. A rebrand video set in GB Lookalikes hit £119 CPA.
Open on the problem in the first five seconds, show the product early, and send people to a page that matches the video's promise. If the video says "your inbox, sorted in a day," don't land them on a generic homepage.
For formats and placement specs, see our Demand Gen and YouTube ads playbook. If you're not sure video belongs in your mix yet, read our guide to see if YouTube ads are worth it for B2B SaaS.
Step 7: Add exclusions to every ad group
Without exclusions, Smart Bidding finds your cheapest conversions, which are often people who already pay you. Add these to every ad group, including prospecting:
- Existing paying customers, through a customer list
- Signups you know are low value (personal emails, students, a zero-value LTV bucket if you have one)
- Demographic limits, only where your CRM proves them
Fyxer excluded paid users and predicted-zero-LTV signups everywhere, limited age to 25 to 44 and removed the bottom 50% of household income. Copy the method, not the numbers. A developer tool with many 22-year-old buyers would lose real customers with Fyxer's age filter.
Step 8: Set up measurement before launch
Decide how you'll prove Demand Gen worked before you spend the first pound. Afterwards is too late.
Three things to set up:
- A brand search baseline. Record your branded search impressions, clicks, and impression share for the three months before launch.
- A holdout. Leave one region dark, or run a Google lift study. Fyxer's branded impressions rose from 62k a month to 169k (2.7x) during the flight and fell to 51k after it stopped. Strong evidence, but without a holdout, it stays correlation.
- YouTube viewers in Search. Add Demand Gen viewer lists to your Search campaigns as observation audiences, so you can see how they convert when they come back. Fyxer missed this, and we'd do it on day one next time.
Also raise your brand campaign budget at launch. During Fyxer's flight, brand impression share fell to between 41% and 62% because demand outgrew the brand budget. Competitors bidding on your name collect the demand you paid to create.
The first eight weeks: what to check and when
[table]
When | Check | Don't
Week 1 | Ad approvals, budget pacing, conversion tags firing | Judge CPA
Weeks 2 to 4 | Sign-up quality in the CRM: email domains, company size | Change bids, goals, or audiences
Weeks 4 to 6 | CPA trend by ad group and by creative | Cut prospecting for being expensive
Weeks 6 to 8 | Pause losing creatives, fund the best ad group | Change more than one thing at a time
Ongoing | Brand search impressions, blended account CPA | Compare Demand Gen CPA to mature Search CPA
[/table]
When it's time to scale, raise budgets by no more than 15% at a time, which is Google's own guidance. Fyxer's GB first month ran at £1,829 CPA. By January, it was £146. A team judging week three would have killed it.
Get your Demand Gen setup reviewed before you spend
Most of what decides a Demand Gen campaign happens before launch. We'll check your conversion goal, budget math, channel settings, audience structure, creative and measurement plan, and tell you what to fix first.
Get My Free Demand Gen Setup Audit
Still choosing a partner? See our top Google Ads agencies in the US.






