Build your first Google Ads Search campaign

This is the lesson where money starts moving, and it moves on your schedule. You build the whole campaign without spending anything, and nothing serves until you publish. Budget about 90 minutes.

Two things have to be true first. You need the account from Lesson 1, with billing configured and auto-apply off. And you need the tracking from Lesson 2 proven rather than installed: you submitted a test lead on your own site and watched the account record it.

One piece of housekeeping. If your signup offered no way to skip the guided campaign and you parked what it built, that campaign gets rebuilt here rather than repaired. Leave it paused and start fresh.

Build the campaign in a notebook before you build it in the product. The wizard asks for decisions in its own order, and a decision you already made is immune to that.

One campaign, one goal. Splitting a small budget across four city campaigns leaves four campaigns that each starve. That is our position; Google publishes nothing about geographic granularity against budget size. Run one campaign over the whole service area, read the locations report after a month, then adjust which towns you target.

Inside it, one ad group per service, because the ad group is where keywords sit beside the ads written to answer them. Somebody with a burst pipe at 6am and somebody pricing a tankless install want different sentences.

Now the money. Start from what you can spend for two months without changing how you run the business, then divide by 30.4: $600 a month is $19.74 a day. Test it against a click price. WordStream and LocaliQ's 2025 benchmark report puts home and home improvement at $7.85 per click with a 7.33% conversion rate, across 16,446 US campaigns from April 2024 to March 2025, a third-party aggregate rather than Google data. On that arithmetic, $600 buys about 76 clicks and five or six leads.

Five leads is a real month for a small shop and a thin month of data. Our rule: a budget that cannot buy 15 clicks a week leaves nothing diagnosable, and the first thing to shrink is the service area rather than the ambition. Our guide to local paid ads runs this by trade.

Open Campaigns in the left menu, use the create control, and choose Campaign. Google asks for your goal before your campaign type, which is the reverse of what older walkthroughs describe. The goal screen is a grid of objective tiles and it moves: Google has consolidated these more than once, and its own objectives article no longer prints the eight-tile grid most tutorials show. Read the descriptions rather than counting positions. A local service business wants Leads, which Google describes as encouraging "relevant customers to express interest in your products or services by signing up for a newsletter or providing their contact information." If nothing on your screen matches that, take the option to create a campaign without a goal's guidance, which Google's own help content points to for this problem and which reveals all available campaign types.

Then comes the type, and this is the most valuable navigation sentence in the lesson: choose Search. If Search is not visible, back out and re-enter through the no-goal path. Performance Max, the type Google presents prominently, deserves a description rather than a warning: you hand over assets, a budget and some signals, and Google decides where the ads run. It is still the wrong first campaign for a contractor, because you cannot read search terms the way a Search campaign shows them, cannot set match types, and cannot learn what your customers type.

Before the settings, check the conversion goals the campaign will optimize toward. Google fills them in from the account goals you created in Lesson 2. A contractor should see phone call leads and lead form submissions there. If that list is empty, stop and fix conversions.

Setting one is Networks, where two checkboxes decide whether a Search budget buys search clicks. Google is plain about the first: "When you create a campaign for the Search Network, search partners are included by default." Partners are thousands of sites and apps outside Google that show search results. Switch it off for the first 30 days to keep the data clean, or leave it on knowingly. Not knowing it was on is the problem.

Setting two is locations, and for a local business it is where money leaks. Target the cities and counties you serve by name rather than drawing a circle: a 20-mile circle around Hollister covers empty range land while clipping half of Salinas, because service areas follow highways and city lines. Then open the collapsed link marked Location options.

Setting three is the pair of AI switches that did not exist when most of the guides you will find were written. The Broad match keywords card converts a campaign wholesale: turning it on "will convert all existing phrase and exact match keywords into broad match". Google gates it anyway, stating the setting "is only available if the campaign is using conversion based smart bidding", so a new campaign cannot switch it on. The AI Max section is one toggle that enables several things at once, and Google states that turning it on enables text customization by default and final URL expansion by default: rewritten headlines, and a click that may land on a different page. Leave both off.

That has a second effect Google has already published. Starting in September 2026, campaigns using text customization or the campaign-level broad match setting will be upgraded to AI Max automatically. A campaign built the way this lesson describes uses neither, which is the reason for the build rather than a prediction about Google. Diary a check for October 2026 and open the AI Max section to see what your account is doing. The fourth setting that matters is bidding, and it gets its own section because it is where the default costs the most.

Google asks what you want to focus on and offers conversions, conversion value, clicks or impression share. Conversions is tempting, and Google is permissive: "Advertisers can start using Target CPA with no conversion history." Google also states, on evaluating that strategy, "For evaluation, we recommend you measure performance for the last 30 days, including at least 30 conversions." For a contractor closing eight leads a month, that is a four-month window on a strategy neither of you can read.

So the starter is clicks with a ceiling. Choosing clicks produces Maximize clicks, which Google defines as "an automated bid strategy that adjusts maximum cost-per-click (CPC) and sets your bids to help get as many clicks as possible within your budget." Read that as written: uncapped, it buys whatever clicks the budget can reach. Underneath the strategy sits an option to set a maximum cost-per-click bid limit. Use it.

Set the ceiling near the going rate rather than under it. Google's warning is direct: "if your CPC bid limits are too low, your bid strategy might not be able to help you achieve your goal." Against the $7.85 category average above, $9 leaves room to compete and $3 removes you from the auction without telling you.

Graduate to conversion bidding when the account produces conversions rather than when a calendar says so, and 30 conversions in 30 days is Google's own bar. One correction for the guides you will read meanwhile. Enhanced CPC is gone: Google states that "Effective the week of March 31, 2025 Enhanced CPC (ECPC) is no longer available for Search and Display campaigns", and campaigns never migrated are "now effectively using Manual CPC". Anything telling you to select Manual CPC and tick Enhanced CPC predates that, and the rest of it deserves the same suspicion.

Budget is one field, and Google's instruction is plain: "Enter your daily budget. This amount is the average you want to spend each day." Enter the number from your notebook: $19.74, not $20, because 30.4 is the multiplier Google uses at the other end.

A keyword is what you tell Google to match. A search term is what somebody typed. Match type decides how far Google may travel between the two, and Google's definitions have drifted a long way from the plain reading of their names.

Google makes the case for starting narrow better than any critic could. It sells broad match on reach, then attaches a condition: "It's critical to use Smart Bidding with broad match", because "every search query is different, and bids for each query should reflect the unique contextual signals." Smart Bidding runs on conversion history, and your campaign has none this afternoon. Start on phrase and exact, and earn broad match with data. Notice too what a keyword typed with no punctuation is: broad.

Choose terms the way a buyer types them. Somebody standing in water searches "water heater leaking" and "emergency plumber near me". Somebody with a free Saturday searches "how to drain a water heater". Both are real people and one has a wallet open. The tells are urgency, a service word and a place name. Google's own guidance is modest, 5 to 20 keywords per ad group, and ten deliberate ones beat sixty accepted from a panel.

Negative keywords are the other half of the list and the half that gets skipped. They tell Google which searches you refuse to appear against, and on a small budget they beat any bid adjustment. Add them at campaign level so every ad group inherits them. A starter list, ours rather than Google's:

• People looking for work. job, jobs, hiring, salary, salaries, career, careers, apprentice, apprenticeship, union
• People studying the trade. school, schools, training, course, courses, class, certification, exam, license test
• People doing it themselves. diy, how to, tutorial, video, youtube, step by step, instructions, guide
• People who are not buying. free, cheap, discount code, used, second hand
• People buying parts, not service. part, parts, supply, supplies, wholesale, rental, manual, diagram

Negatives behave differently from what you would guess. Your positive keywords expand to catch searches you never wrote; your negatives do not. Google: "Negative keywords won't match to close variants or other expansions." Excluding "job" does not block "jobs", so add the singular and the plural of every word above, along with your own synonyms. The exception runs in your favor: Google states that "Negative keywords automatically account for casing and misspellings, so you don't need to add those separately", so building misspelling lists by hand is wasted work.

A shared negative keyword list lives in the shared library and applies to several campaigns at once. Building one and never applying it is a failure Google names outright: unapplied lists "need to be associated with specific campaigns to be active". Build it, apply it, then confirm it is listed on the campaign. Like Lesson 2's missing phone snippet, it has no error state.

You write a responsive search ad, and Lesson 1 covered its shape: up to 15 headlines of 30 characters, up to 4 descriptions of 90, with Google assembling the combinations and a minimum of one headline and one description selected. So every headline has to make sense beside every other one. The ad is a pile of interchangeable parts rather than a paragraph you chopped up.

A headline set for a trade covers four jobs, with room for three or four of each.

• Service and city. "Hollister Water Heater Repair", "Water Heater Repair Nearby". Somebody with a flooded garage checks whether you are close before reading anything else.
• Proof. "Licensed & Insured Plumbers", "Family Owned Since 1994". Both have to be true: Google reviews ads and customers check.
• Speed. "Same-Day Service, 7 Days", "We Answer After Hours". For emergency work this is the entire ad.
• The ask. "Call For A Free Estimate", "Book Online In 2 Minutes". Say what happens next.

Descriptions carry what a headline cannot, and 90 characters is two short sentences: "Water heater out? We answer the phone and most repairs are done the same day." and "Serving Hollister, San Juan Bautista and Tres Pinos. Free quote before we start." One takes the problem, one takes the service area.

Pinning fixes a headline or description to a position. Content pinned to headline positions 1 and 2, or description position 1, will consistently display, while Google states that content pinned to headline position 3 and description position 2 is not guaranteed to show. It costs you combinations, so pin sparingly. For a contractor one pin earns its cost: your city in headline position 1, when appearing to serve towns you do not cover would waste your money and somebody else's afternoon.

Pinning also drops your Ad strength rating, which matters less than the product suggests. Google's sentence: "The Ad strength rating of an ad doesn't directly influence your ad's serving eligibility." The ratings run Incomplete, Poor, Average, Good and Excellent, and they report on your creative rather than grading your auction, exactly as Lesson 1 described Quality Score. An Average ad naming the town and the service beats an Excellent one padded to move a meter.

Two things keep the ad yours. Text customization, which older articles call automatically created assets, lets Google generate headlines from your site; it lives inside the AI Max panel now, so leaving AI Max off leaves this off too. And call-only ads are finished: Google removed the ability to create them in February 2026 and existing ones stop serving in February 2027, so a phone-first Search ad now means a responsive search ad with a call asset.

Assets are the extra lines that make an ad taller: links, short phrases, your location, your number. A taller ad takes more of the screen and gives somebody more reasons to choose it, and none of them cost anything. Google offers them inside the creation flow, which is when to add them.

Start with the call asset. For a trade it is the difference between an ad that gets clicked and one that gets dialed.

1. 1

   Add a call asset at the campaign level

   Google lists it among the optional enhancements in the Search campaign flow. Campaign level means the number rides on every ad.
2. 2

   Enter the number a customer already calls

   Google substitutes one of its own forwarding numbers in the ad and routes the call to yours, which is what makes it countable.
3. 3

   Confirm that number appears as selectable text on your landing page

   A number baked into a logo image does not count. This is the most common reason a contractor's call asset fails review.
4. 4

   Check that Lesson 2's calls-from-ads action is still recording

   The asset supplies the number. The conversion action counts what happens on it.

Sitelinks come next, and there is a floor under them. Google: "You need at least 2 sitelinks (for desktop), and at least 2 sitelinks (for mobile) for the sitelinks to appear in the ad." One sitelink shows nobody anything. Google recommends at least six with descriptions, and six is easy: Services, Service Area, Reviews, About Us, Free Estimates, Contact. Text caps at 25 characters, so "Emergency Drain Service" fits and "Emergency Drain Cleaning Service" does not.

The last screen summarizes the campaign and flags configuration problems. Read four lines before you select Publish: Networks, Locations, the bid strategy and the budget. This is the last cheap moment to catch a default you inherited without noticing.

Publishing does not start the campaign. Google's review does. Google states that "Most ads are reviewed within one business day", and advises checking the status if an ad is still under review after 2 full business days, or contacting support past a week. Nothing serves and nothing spends while an ad waits, so publish on a Friday evening and Monday may bring your first impression. One habit to avoid: editing an ad restarts its review.

• Under review. Google is still checking the ad, and it becomes eligible to serve once it complies. It costs nothing while it waits.
• Eligible. Google: the ad "complies with the Google Ads policies, so it can be shown to all audiences."
• Eligible (limited). The ad "can run but will be restricted on where it can serve to comply with policies", which can turn on location, device, the viewer's age, or your eligibility to advertise a restricted service.
• Disapproved. The ad breaks a policy and shows nowhere. Read the Policy details column beside the Status column rather than guessing.

Then the choice nobody writes about. Publishing enabled is the normal path and carries little risk, because nothing spends until review clears and the budget holds the day. Publishing paused makes sense when something else is unfinished, or when the campaign would go live before a weekend you are away. A missed call from a paid click costs what a booked one costs, so publish it, pause it, then enable it Monday at 7am.

The first days look worse than they are, for arithmetic reasons rather than performance ones. Clicks arrive slowly on a small budget, and conversions are dated to the click and lag by hours, so the last two days of any report are understated by design. Nobody publishes a learning period for a new Search campaign, Google included: the seven-day and two-week figures repeated across the web are convention rather than documentation.

Lesson 4 covers the first 90 days: which numbers to read, which to ignore, and how to tell a slow start from a broken campaign. Accounts are rarely lost in the build. They are lost in month two, when one bad week gets treated as a verdict and somebody changes six things at once. If you would rather hand this over, that is what our paid advertising work does, starting from the account you built rather than replacing it.