Most of what is written about Google Ads assumes you sell products: a feed, a cart, a ROAS column that settles every argument. Google Ads for lead generation runs on different machinery. The conversion is a form fill or a phone call, the real sale happens days or weeks later, and metrics that look healthy can hide a channel quietly losing money. Here is how the setup differs, piece by piece.
Why e-commerce playbooks do not transfer to service businesses
E-commerce Google Ads is built around a closed loop. A product feed tells Google what you sell, a purchase fires a conversion with an exact value, and the algorithm optimizes toward revenue it can see. Shopping and PMax exist because of that loop.
Service businesses have none of it. There is no feed, no cart, and the conversion event is a promise, not a payment. A form fill might become a 9,000 euro contract or a spam submission, and Google cannot tell the difference unless you teach it.
That gap changes every layer of the account: what counts as a conversion, which campaign types make sense, how bidding learns, and what you report. Run a lead gen account on e-commerce logic and you get the classic failure: a dashboard full of cheap conversions and a sales team asking where the customers are. It is the account-level version of the landing page problem we broke down in why your ads get clicks but not leads.
Conversion setup: where Google Ads for lead generation is won or lost
In e-commerce, a missed conversion understates performance. In Google Ads for lead generation, a sloppy conversion definition redirects the entire account toward the wrong goal, because Smart Bidding optimizes toward whatever you count.
Count every form submission as a conversion and the algorithm hunts for people who fill in forms, including bots and tire-kickers. The fix is a hierarchy: a primary conversion for qualified actions (a form that passes validation, a call over 60 seconds) and secondary conversions for softer signals that you observe but never bid toward.
Phone calls need their own tracking, not just clicks on the number. And if your CRM can mark which leads became customers, import those back as offline conversions, because that is the moment Google starts learning what a good lead looks like rather than a cheap one. The plumbing has to be verified end to end first, and the failure modes are the same ones we documented in our conversion tracking guide.
Campaign types: Search first, everything else on merit
Search is the backbone of lead gen and usually deserves the whole budget at the start. Someone typing "emergency electrician antwerp" is declaring intent in plain language, and Search is the only campaign type that lets you match that declaration keyword by keyword.
PMax for lead gen is a riskier proposition than for e-commerce. Without purchase values, it optimizes toward whatever conversion signal exists, and if that signal is weak it manufactures cheap junk leads with conviction. We only consider it once offline conversion data is flowing back, so it has something real to chase.
Display and video are demand creation, not capture. For a local service business with a finite budget, they are usually the last money spent, not the first. The sequencing instinct is the same one a well-run Google Ads engagement applies everywhere: fund the channel closest to declared intent first, expand outward only when it saturates.
Bidding in a low-volume environment: Target CPA without the chaos
Most service businesses generate tens of conversions per month, not thousands. That changes how bidding behaves, because Smart Bidding learns from conversion volume and thin data means slow, noisy learning.
The practical sequence: start on Manual CPC or Maximize Clicks to gather data while you confirm tracking fires correctly, move to Maximize Conversions once a steady stream exists, and only then set a Target CPA, anchored to what the data shows rather than what you wish the number was. Set a tCPA 40% below your observed cost per lead and the system simply stops entering auctions.
Expect turbulence after every change. The learning phase in a low-volume account can take weeks rather than days, so resist the urge to react to three bad days. Judge bidding changes on rolling 30-day windows, the only timescale at which a low-volume account tells the truth.
Lead quality vs lead volume: the metric that decides profitability
Cost per lead is the most quoted and most misleading number in lead generation. A 20 euro CPL sounds better than a 60 euro CPL, until you learn the 20 euro leads close at 2% and the 60 euro leads close at 20%. The expensive leads are three times cheaper per customer.
The only way to see this is tracking leads past the form: where each lead came from down to the campaign and keyword, and what happened to it. A spreadsheet updated weekly beats no system. A CRM that feeds closed deals back into Google Ads as offline conversions is the proper version.
Once quality data exists, the account changes shape. Keywords producing cheap unqualified leads get cut even though they flatter the CPL, and budget concentrates where customers come from. Without that feedback loop you are optimizing the wrong half of the funnel and calling it performance.
Keyword strategy: intent tiers from research to ready-to-hire
Service keywords sit on a ladder of intent. At the bottom, research queries like "how much does a new roof cost": high volume, low readiness. In the middle, comparison queries like "best roofing company reviews". At the top, hiring intent: "roof repair near me", "emergency roofer [city]", service plus location plus urgency.
Budget should be distributed accordingly. Own the top tier completely before spending a cent further down, because a ready-to-hire click is worth multiples of a researcher even at triple the CPC. High CPCs at the top of the ladder are not a problem. They are the market correctly pricing buyer intent.
The discipline that protects all of it is negative keywords. Service queries drag in job seekers, DIY searchers, and people looking for free advice, and every irrelevant click costs real money at lead gen CPCs. The audit process from our negative keywords guide applies unchanged: read the search terms weekly, cut the waste, repeat.
Reporting when there is no ROAS
E-commerce reporting settles into ROAS. Lead gen has no equivalent single number, so accounts get reported on what is easy: clicks, CTR, impressions. None of it says whether the channel makes money.
The chain worth reporting: spend, leads, qualified leads, customers, and revenue where the close data exists. Even when the last steps live in a spreadsheet, cost per qualified lead tells a usable story, and once average customer value enters the picture you can state plainly what a euro of ad spend returns. That is the number a business owner needs ahead of any CTR chart.
Set expectations on timeline too: the feedback loop in lead gen runs through a human sales process, so the channel takes longer to read than an e-commerce account, a curve we map in the profitability timeline FAQ. Pricing for this kind of management is covered in our cost FAQ, and if you want your current setup checked first, the free 48-hour audit covers lead gen accounts as well as stores.