Search ads reach people who already know what they want. Demand Gen goes looking for software buyers who haven't searched yet, while they watch YouTube, scroll Discover, open Gmail, check Maps or read a site that carries Google's display ads. For a software company, that sounds like the B2B dream. Put the product in front of operations managers while they watch a tutorial, and catch them before they ever type the category name.
The catch is who you can aim at. Software founders who've run LinkedIn ads expect to pick a job title, a seniority level, maybe a list of target companies. Google offers none of those. This chapter covers what Google can target for a software company, how to build audiences from the two things it does well, and how Demand Gen compares with LinkedIn for a team deciding where the next dollar goes.
LinkedIn Can Target a Job Title. Google Can't. Its Closest B2B Filters Are Three Company Sizes and Eight Industries.
Google's audience list includes what it calls detailed demographics, and a small part of it is about work. There are three company size bands: small employer (1 to 249 employees), large employer (250 to 10,000) and very large employer (more than 10,000). There are eight industries: construction, education, financial, healthcare, hospitality, manufacturing, real estate and technology. Google's audience page marks these as available in Demand Gen.
That's the whole B2B menu. We found no Google help page offering targeting by job title, job function, seniority or company name. LinkedIn's advertising pages, by contrast, describe targeting "based on traits like their job title, company name or industry," drawn mostly from the profile data its members enter.
So Google can tell a 40-person company from a 4,000-person company. It can't tell that company's controller from its receptionist. For a payroll product sold to finance leads, or a sales tool sold to heads of revenue, that's a real gap. For a product that any small business owner buys, like scheduling or a simple CRM, it matters less, because the buyer and the business are often the same person.
The company size bands still earn their place. A scheduling app built for small shops can lean on the 1-249 band, and an HR platform built for 500-person firms can lean on large employers. Industry helps where the product is built for one, like a construction CRM aimed at the construction segment.
Most Software Buyers Compare Two Products, Not Ten. A Custom Segment Built From Your Rival's Site Finds the Other Half of That Shortlist.
What Google does well is intent: what people search for and which sites they visit. Demand Gen uses both through custom segments. You list search terms, website addresses or apps, and Google builds an audience of people whose recent behavior matches. One option is literally "People who searched for any of these terms on Google." These segments work in Demand Gen, YouTube and Display campaigns, not in Search.
Software buyers make this unusually effective, because their shortlists are short. In the Capterra reviews we read, 84 reviewers listed the alternatives they considered, and 47 of them named just one. On Reddit, buyers asking for advice named a median of two products. The buyer is usually comparing the market leader with the tool they already use.
Build the segment from that shortlist:
- Your rivals' brand names and "alternative" searches, including the rival names Chapter 5 kept out of your main Search campaigns as negatives.
- Comparison searches like "[rival] vs [rival]" and "[rival] pricing."
- Your rivals' website addresses. Google is clear that listing a URL "doesn't mean your ads will show on those URLs." It means Google looks for people who browse sites like them.
One payroll buyer comparing Gusto and Rippling on Reddit described exactly how this feels from the other side: "Now Rippling keeps coming up in my LinkedIn feed." That's the job a custom segment does on Google: put your product in the feed of someone already weighing two others.
Google Recommends a Demand Gen Budget of 10 Times Your Target Cost per Lead. Few Young Software Accounts Can Feed It That.
Demand Gen learns from conversions, and it needs a lot of them. Google's performance guide says the campaign "needs to generate at least 50 conversions to get a good foundational understanding," and that "Your budget should be at least 10 times your target CPA."
Do the math with a qualified lead. If your target cost per qualified lead is $150, an illustrative figure, Google's guidance points to a daily budget of about $1,500. Fifty qualified leads at $150 each is $7,500 before the campaign has learned much at all. For most software accounts in their first months, that budget belongs to the exact match Search campaign, where every dollar lands on a search somebody chose.
That's why Demand Gen comes after Search in our order. Search proves which searches and which messages bring qualified leads. Demand Gen takes those proven messages to a wider audience once the account can afford to teach it. Once it runs, Google's guide counsels patience. Spending around four-fifths of the daily budget is a sign of health, raises should stay at or under 15%, and any change needs 30 to 50 conversions behind it before anyone judges it.
Google Tells Demand Gen to Optimize for Light Actions Like Site Visits. A Software Account That Does Buys Browsers, Not Buyers.
Google knows most accounts can't produce 50 deep conversions quickly, so its advice bends. The Demand Gen overview page suggests optimizing for light conversion events, naming site visits and add-to-cart clicks as examples, and its performance guide suggests adding shallow goals as non-biddable conversions to help the campaign ramp.
For a software company, that's the trap from Chapter 1 in a new place. Optimize for site visits and Demand Gen finds people who visit sites. Optimize for trial sign-ups and it finds people who sign up for free things. Neither is a buyer. One software founder on Reddit described spending $2,000 on Reddit ads and getting "14 signups. 1 paying customer." A reply explained the pattern: people on a feed are "in browse mode not buy mode."
Our approach: if Demand Gen needs a shallower goal to get started, use it for bidding, but judge the campaign only on qualified leads from the CRM. If a month of spend produces site visits and sign-ups but no qualified leads, the campaign isn't learning slowly. It's learning the wrong thing.
Google Builds a Lookalike From Whatever List You Give It. Seed It With Paying Customers, Not Every Free Sign-Up.
Demand Gen's lookalike segments start from a list you provide, called a seed, and find people who resemble it. Google offers three sizes, Narrow, Balanced and Broad, which it describes as 2.5%, 5% or 10% of users in your target location. Through 2026, starting in March, Google has been rolling out a suggestion mode as the default, which treats the lookalike as a hint rather than a limit and drops the old 100-person minimum. Advertisers who opt out keep the strict version, where those size settings are hard limits and the seed lists together need more than 100 active matched people.
The seed decides everything. Upload every email that ever started a free trial, and the lookalike resembles everyone who signs up for free things: students, competitors and the bots from Chapter 10. Upload the customers who pay you, ideally the ones who stayed past their first renewal, and the lookalike resembles buyers.
Two related rules from earlier chapters still apply. Every policy-compliant advertiser can use a customer list as an exclusion, so keep paying customers out of the Demand Gen audience itself (Chapter 9). Aiming ads straight at an uploaded list is locked until an account has 90 days of history and over $50,000 spent, so for a software company in its first year, the lookalike is the way that list gets used.
LinkedIn Offers the Job-Title Targeting Google Lacks. Test It Beside Demand Gen, Never Instead of Search.
So where does LinkedIn fit? It does the one thing Google can't: put an ad in front of a specific role. Its advertising pages set a minimum audience of 300 members and suggest aiming for 50,000 or more, a minimum daily budget of $10 and charges by click or by impression depending on the campaign's objective.
For a software company, the order still starts with search:
- Google Search first. People typing "payroll software for small business" are already shopping. No feed campaign beats that intent.
- Demand Gen next, once Search is trained and the account can fund it, for reach on YouTube, Gmail and Discover built from your rivals' audiences.
- LinkedIn when the job title is the product's whole market. If you sell to controllers, heads of sales operations or IT directors, LinkedIn's targeting may be worth testing beside Demand Gen.
Set the test up so the comparison is fair. Tag each channel's links so the CRM records where every lead came from, run the channels over the same months, and compare them only after sales has had time to qualify the leads, which can take weeks after the click.
Whichever runs, judge it the same way: qualified leads in the CRM per dollar spent. A LinkedIn lead from the right job title who never takes a demo is worth no more than a Demand Gen click from the wrong one.
Next: Login Searches Run at More Than a Third of BambooHR's Brand-Name Volume. Your Brand Campaign Shouldn't Pay to Reach Existing Customers.
Chapter 12 turns to the searches for your own name, and how to make sure a brand campaign wins new customers instead of paying for existing ones to log in.
Every chapter, from tracking to brand campaigns, competitor campaigns and the rules for free trials, is on the guide's home page. Chapter 10 covers Performance Max, Google's other campaign that reaches beyond the search box.
For the software companies in our Google Ads management work, Demand Gen starts only after the exact match campaign has a few months of qualified leads, and its first audience is built from the client's two closest rivals. Write down the one or two products your last five customers compared you with, and you have the start of that segment.
