Most of the broken Google Ads accounts I have audited were broken on day one. Not by a bad keyword or a weak ad, but by a launch sequence that skipped the boring parts. Someone opened the account, took the guided setup, accepted every default, and started spending before anyone had decided what a customer was worth or whether the conversion numbers meant anything. Six weeks later the account has data, but the data is fiction, and every decision made on top of it is guesswork. Having worked inside a team that managed thousands of accounts and eight figures of annual budget, I can tell you the accounts that work are almost always the ones where the first two weeks were spent not spending. This is the sequence I would follow today, in 2026, for a business starting from nothing.
Google Ads is a demand capture channel first. It is very good at putting you in front of people already looking for what you sell, and much weaker and more expensive at convincing people who have never thought about the problem. If nobody is searching for your category, paid search will not create that demand at a sane cost. Demand Gen and video can build interest, but they are patient plays and a poor first campaign for a business that needs leads this quarter.
Three questions worth answering honestly before you spend anything:
If the answer to any of these is no, fix that first. Ads amplify what already exists. They do not repair it.
This is the number that governs everything after it, and it is the step most businesses skip. You cannot judge a cost per lead of forty euros as good or bad without knowing what a lead is worth to you.
Work backwards from your own numbers. Take gross profit per sale, not revenue, multiply by repeat purchases if you have that data, then multiply by the share of leads that become customers. That is the most you could pay for a lead before breaking even. Your target is a fraction of it, because you also have to cover overheads and management time.
A simple version: if a closed job brings 900 euros of gross profit and one in four qualified leads closes, a lead is worth 225 euros gross. If you want ads to run at three to one, your target cost per lead is 75 euros. That is the number the account is judged against. Not clicks, not impressions, not click through rate. Write it down before launch, because if you calculate it afterwards you will fit it to whatever the campaign produced.
An account without trustworthy conversion tracking is not an advertising account. It is a donation. Smart Bidding, which now runs almost every campaign type, optimizes toward the conversions you report to it. Feed it garbage and it will very efficiently buy you more garbage.
Track the action closest to money that you can count reliably. For lead generation that means form submissions and phone calls long enough to be real conversations. For ecommerce it means purchases with the transaction value passed through.
Do not count page views, time on site or every button click. And do not mark six soft actions as primary, because the bidding system will optimize toward whichever is cheapest and most frequent, which is almost always the least valuable. Keep one or two primary conversion actions. Everything else goes to secondary, where you can still see it in reporting without it steering the bidding.
If you have any European traffic, Consent Mode v2 is now a hard requirement rather than a nice to have. Without the consent signals for advertising data and ad personalization being passed correctly, Google drops remarketing audiences, conversion modeling and enhanced conversions for that traffic, and accounts can pick up policy warnings. Set up a certified consent management platform, wire it into your tag manager, and confirm the consent signals fire before you launch. Then add enhanced conversions, which send hashed first party data such as an email address with the conversion so Google can match conversions it would otherwise lose. Enhanced conversions need weeks of data before the effect shows in reporting, which is one more reason to have it live from day one rather than bolted on later.
Submit the form yourself. Call the number yourself. Then check the conversion appears in Google Ads within a day, with the right value and the right count setting. Use one conversion per lead for lead generation and every conversion for ecommerce. I have opened accounts where the thank you page fired the conversion tag on every refresh and the reported cost per lead was a third of the truth.
Structure exists so you can read the account later and so budget goes where you want it. Nothing else. Elaborate structures with thirty campaigns and one keyword each are a relic. Modern bidding needs conversion volume in one place to learn, and hair thin campaigns starve every one of them. For a first account, I would build:
Leave display expansion and partner networks off. The search partners and display network settings are on by default in some setups and they will spend a share of your budget on inventory you did not choose.
Start narrow. Twenty to forty well chosen keywords beat four hundred. Use the Keyword Planner, but also read how customers describe the problem in their own words in reviews, support emails and sales calls. Those phrases convert better than the industry vocabulary.
On match types, my default for a new account is phrase and exact only. Broad match plus Smart Bidding can work well, but it needs conversion data to steer it and a new account has none. Broad match on day one is how you pay for searches with only a loose thematic relationship to your business. Add it later, in its own ad group, once conversions are consistent.
Build a negative keyword list before launch, not after: free, cheap, jobs, salary, DIY, course, and competitor names you do not want to bid on. Set audiences to observation rather than targeting on search, so you learn who converts without narrowing reach. Create remarketing and customer match lists on day one anyway, because they take time to populate.
Responsive search ads are the format. Give each ad group at least ten headlines and four descriptions that are genuinely different, not ten rewordings of one claim. Pin only the headline that must always appear, usually your brand or a legal requirement. Over pinning removes the only advantage the format has.
Write about what the customer is worried about: price transparency, response time, guarantee, coverage area, qualifications. Specific beats clever. Put your strongest proof in the ad and make sure the same words appear on the landing page, because a mismatch between ad and page is the fastest way to burn budget.
The landing page matters more than the ad. Send traffic to a page about the specific service in the ad, not the homepage. One clear action, visible without scrolling. A tappable phone number. A short form, because every extra field costs you leads. Load in under three seconds on mobile data, not on your office fibre.
Set a budget you can sustain for at least three months. A campaign that runs two weeks and stops teaches you nothing. As a rough floor, your daily budget should buy a handful of clicks a day at your category click price, and ideally produce fifteen to thirty conversions a month once things work. Below that you will never separate signal from noise.
Start with Maximize conversions and no target, or manual CPC if volume is very low and you want tight control. Do not set a Target CPA on day one. You do not yet know what is achievable, and a target set too low simply throttles the campaign until it barely serves. Once you have roughly thirty conversions in a month, move to Target CPA or Target ROAS, set the target near your actual achieved cost, then tighten in steps of ten to fifteen percent.
One thing to know in 2026: Google changed how target based bidding behaves in budget limited campaigns. Campaigns that used to look like they were beating their target while capped by budget now spread that budget across a wider set of auctions and report closer to the target you set. Expect reported cost per acquisition to move toward your target rather than sitting below it. That is the system doing what you asked, but it does mean targets need to be set deliberately rather than left at whatever the setup wizard suggested.
Week one is about safety, not optimization. Confirm ads are serving and approved. Confirm conversions are recording and match your inbox or CRM. Read the search terms report daily and add negatives. Check spend is landing where you expected across campaigns and devices. Do not change bids or budgets yet. Changing settings in the first days resets the learning and you will never know what caused what.
Week two is the first real read. Search terms again, then which ad groups and keywords are getting impressions and which are getting nothing. Check impression share lost to budget and to rank. If you are losing most impression share to rank on your best terms, bids or quality are the problem. Pause keywords that have spent more than a third of your target cost per acquisition with zero conversions and no relevance. Leave the rest alone.
Week four is when you are allowed to have an opinion. Compare actual cost per acquisition against the number you wrote down in step two. If you are within reach, move to a target based bid strategy and start expanding: add converting search terms as their own keywords, test new ad variations, widen match types cautiously. If you are far off, the answer is usually not in the bidding. It is the offer, the landing page, or the fact that the keyword intent is wrong.
If you are launching, do the first three steps this week and do not open the campaign builder until they are done. Demand, target cost per acquisition, tracking. Everything after that is easy to change. Those three are the ones that quietly decide whether the account is readable a year from now.
If you have already launched and skipped them, you have not lost anything permanent. Pause spending for a few days, fix the tracking, calculate the number, then turn it back on. A week of no data is cheaper than three months of data you cannot trust. If you want a second pair of eyes on the setup before you spend, you can book a free consultation and we will go through it together.