LinkedIn
A $50/day Search campaign and a $500/day one should not follow the same day-seven checklist. One may have 40 clicks when the other has 4,000. I’ve watched a small account get judged before it had enough searches, and a faster one buy junk traffic while everyone waited politely for “learning” to finish. This list is for anyone launching a Google Ads Search campaign who wants to know what to touch without changing five things at once. The order is fixed. The speed is not. Watch the clicks, search terms and conversions; let each threshold tell you when to move.
1. Prove tracking works before you spend
Trigger: before the first paid click. Everything else in this list depends on the conversion signal. If a purchase or lead fires twice, or never fires on a browser your customers use, you spend the next stretch fixing data instead of buying customers. Smart Bidding cannot learn from a conversion it never sees.
Open Goals > Conversions > Summary. Launch Tag Assistant on the Unverified action, connect to the landing page, submit a real test, then hit Finish. Check that the action registers. Google may show No recent conversions before it shows Recording; the draft’s expected window is roughly 30 minutes to a day. Do not mistake a form thank-you page for proof that the conversion reached Google Ads.
I used to give launches a tidy two-week review date. That was convenient for my calendar, not the account. After a bid change, Smart Bidding can take up to around 50 conversion events or three conversion cycles to settle. A low-volume campaign needs more calendar time to collect the same signal. No confirmed conversion, no launch. Test the instrument before you read it.
2. Clear policy problems before you read performance
Trigger: ads have been submitted for review and their status is available. Check Ads and Assets for Disapproved and Limited by policy before you decide that one ad beat another. If an asset cannot serve, its apparent performance tells you very little about the message. The auction was not giving your options a fair run.
Fix the violation, appeal only if you are sure the flag is wrong, and replace an asset if it keeps getting limited. Then check that the ads you intend to compare can actually serve. This is a status check, not an excuse to rewrite every headline on day two. First step: clear the serving problem; judge the copy later.
3. Kill obvious junk at 50 to 100 clicks
Trigger: the first 50 to 100 clicks. This is not the point to declare a keyword a winner. It is the point to stop paying for searches that plainly cannot buy. Close variants and same-meaning matching can send a repair campaign toward DIY research. A launch checklist cited in the draft estimates 15 to 30% of week-one spend can go to traffic that was never going to convert.
Open Keywords > Search Terms and sort by cost. Look for unmistakable mismatches such as free, jobs, careers, tutorial, how to, reddit, wikipedia, DIY and course. Add campaign-level negatives where those searches cannot lead to a sale. Do not blacklist a borderline term because it got two clicks and no lead. Negate the impossible, not the merely unproven.
4. Sweep search terms after a full weekday-and-weekend cycle
Trigger: one full calendar week of search terms, or roughly 300 to 500 impressions per ad group. The first cleanup catches the obvious junk. This one catches patterns: locations you do not serve, competitor searches you do not convert on, or price shoppers whose intent does not fit a premium offer. Tuesday and Saturday may tell different stories. A quiet Monday cannot stand in for either.
Sort search terms by cost again and group the expensive mismatches by intent. The same launch guidance recommends weekly search-term reviews through the first month, because new variants keep appearing after the first pass. Add negatives at campaign level unless you have a reason to keep a particular ad group open to that query. First step: find the query family eating spend, not just the single odd search.
5. Pause the clearly weaker ad, then leave the bids alone
Trigger: each ad you are comparing has about 1,000 impressions or 30 clicks. Now you can start reading the ad comparison, provided both ads were eligible to serve. Check whether the headline matches the queries that brought people in. Search CTR around 5 to 10% is one benchmark; below 3% can point to a headline-query mismatch. Treat that as a clue, not a verdict. A click tells you someone noticed the ad, not that they became a customer.
If one ad is plainly less relevant, pause it and keep the stronger one serving. Do not pair that change with new bids, a budget increase and a keyword rebuild. I used to tweak bids here because it felt like account management. Mostly it made the next result harder to explain, especially during the early learning period when frequent changes are best avoided. Change the ad if the ad is the problem. Let the bid data accumulate.
6. Fix the page behind your top five queries
Trigger: you can name your top five queries by spend. Pull Campaigns > Keywords > Columns > Quality Score and add Quality Score, Landing Page Exp., Ad Relevance and Exp. CTR. Landing Page Experience is one part of Quality Score, so it is worth checking before you try to solve every CPC problem with a bid change.
Start with the promise. Does the landing-page H1 match what the ad and query led someone to expect? Is there one clear CTA? Fix that before you polish the design. Then run the top page through PageSpeed Insights and check mobile. The page-speed source cited in the draft reports a 21.6% lift in lead-form completions alongside a 0.1-second mobile improvement across 37 brands. That is a reason to inspect the page, not a promise that shaving 0.1 seconds will rescue yours. First step: make the page deliver what the query and ad promised.
7. Diagnose Limited by budget before adding money
Trigger: the status column shows Limited by budget for two days running. I read that label as a question, not an invoice. Did high-intent searches run out of budget, or did junk spend it first? Raising the cap funds both unless you separate them.
Sort search terms by cost and split qualified spend from waste. Then compare impression share lost to budget versus rank, along with hour, device and location performance. Look for core terms that convert inside your target but stop serving before peak converting hours. That is the case for raising budget: try a controlled 15 to 20% step. If weak queries consumed the day instead, add negatives and isolate the core campaign from tests before spending more. Fund the terms that earn more room, not everything in the account.
8. Find the device or location dragging down the average
Trigger: roughly 200 to 300 clicks or $500 in spend, whichever comes first. I am not rebuilding geo targeting from a tiny segment report. I am looking for a mismatch big enough to investigate: a place you cannot serve, or a device sending costly visits into a form people cannot use.
Segment by device, then by location. Compare conversion rate and CPA rather than CTR, and check the spend behind each result. If one city or device has roughly triple the CPA and no conversions on meaningful spend, inspect the query mix and landing-page experience before you exclude it or bid it down. Desktop may convert differently from mobile; the account average hides that. What the deck calls “granular optimization,” I call finding the postcode that never buys. First step: locate the expensive outlier before changing the whole campaign.
9. Expand match types only when the core is starved
Trigger: exact and phrase have at least 2,000 impressions, but the core still is not spending after the earlier cleanup. Check coverage and impression share lost to rank first. If rank is the blocker, adding a broader match type does not fix the reason the existing terms struggle. If the exact core is spending and converting, you may not have a match-type problem at all.
When the core truly needs more reach, test one additional phrase term where it fits and keep its results readable. I would hold a broad-match test until the campaign has conversion history: the launch guidance cited here recommends waiting until after 30 conversions before using broad, while starting lower-volume campaigns on Maximize Conversions without a target. If you do test broad, put it in a separate campaign if you need its own budget cap. An ad group cannot have one. First step: diagnose why the current terms are starved; expand only if reach is actually the problem.
10. Build a conversion sample before you switch to tCPA
Trigger: at least 15 conversions, with 30 in the last 30 days the stronger point for considering a switch. This is where a launch calendar can get expensive. Target CPA tells Google to avoid auctions it predicts will miss your number. With only a handful of conversions, those predictions have little campaign history behind them. Set a tight target too early and delivery can shrink just when you need more signal.
One bidding guide advises waiting for 30 conversions in 30 days before moving from Maximize Conversions to Target CPA, with 50-plus making the switch smoother. Another puts the working floor around 15 to 30 monthly conversions. If you have eight conversions after two weeks, stay on Maximize Conversions and keep cleaning the traffic. No usable sample, no target. Waiting for conversions is a decision, not neglect.
11. Set your first CPA target from what the account buys
Trigger: 30 conversions in the last 30 days. Now you have a more useful trailing CPA. Do not type in the CPA you wish the account had. Google uses the target to decide which auctions look worth entering; a target set well below current performance can cut eligible volume instead of bargaining down your costs.
Start 5 to 10% above actual trailing CPA: the draft’s example is a $320 target on a $307 actual. Let it run a week before you consider stepping down. Keep budget and bid edits in small 10 to 15% steps rather than stacking big changes that make the next result hard to read. First step: calculate the observed CPA, then set a target near it. A target describes what you can currently buy; it does not negotiate on your behalf.
12. Record the baseline before you tell the month-two story
Trigger: after the first target decision, once you have the conversion sample to judge it against. On a faster account, that may fit inside 28 days. On a slower one, do not jump past moves 10 and 11 just because the month ended. Keep a month-end record if you want one, then make this your first full baseline when the sample arrives. A $50/day account and a $500/day account run the same list at different speeds.
Log four things and one screenshot: CPA, conversion rate, impression share lost to budget versus rank, the top 10 search terms by cost, and the landing-page version that produced the results. No verdict in the document. Just the record. Otherwise, in month two everyone remembers month one as better or worse than it was and credits whichever change they already liked.
This is also where an autonomous setup pulls ahead of a Friday account review. groas runs daily checks and works on bids, keywords and budgets around the clock for a flat fee, with changes and their results recorded. The record is there when you need to judge a decision, rather than waiting for someone to reconstruct it. If you want more detail around the launch window, use the week-by-week plan for the first four weeks, the first two-to-four-week priorities and the 28-day checklist as references, not as dates that override your counters.
If I could make only one of these moves, I would still start with move 1. Bad tracking poisons the other eleven. A real test conversion comes before the spend, before the clever bid strategy and before the slide about what the campaign supposedly learned. Everything else can wait for its threshold. Tracking cannot.