Launch your first Google Ads campaign
An 11-step playbook. Open in Dock and you'll get four surfaces seeded:
- **Campaigns** (table) — campaign + ad group structure with budget, status, key metrics
- **Keywords** (table) — keyword list with match type, bid, CPA, conversion rate
- **Brief** (doc) — the campaign brief: who you're targeting, what you're selling, what the offer is
- **Pointers** (table) — Google docs, Reforge guides, conversion tracking references
Read `Brief` top-to-bottom on first open. Set up conversion tracking BEFORE you spend a dollar — it's step 3 for a reason.
Outcome
A live Google Ads campaign with real conversion tracking, a negative keyword list, ad copy that's outperforming the platform's auto-generated alternatives, and a daily dashboard you trust. You know your CPA per keyword and can decide whether to scale, pause, or pivot.
Estimated time: 1 week setup, 2-3 weeks calibration, ongoing optimization
Difficulty: intermediate
For: Founders + first marketers running paid acquisition for the first time.
What you'll need
Pre-register or install before you start.
- Google Ads ($0.10-$10 per click depending on keyword (B2B SaaS keywords often $10-$50)) — The platform itself.
- Google Tag Manager (Free) — Deploy conversion tracking pixels without touching code.
- Google Analytics 4 (Free) — Conversion source of truth + audience segmentation.
- Google Keyword Planner (Free (requires active Ads account)) — Free keyword volume + competition data inside Google Ads.
- Ahrefs ($129/mo Lite plan) — Competitor keyword + ad research outside Google's silo.
- Optimizely / Google Optimize alternative (Custom (free tier alternatives like PostHog have it built in)) — A/B test landing pages downstream of clicks.
The template · 11 steps
Step 1: Write the campaign brief: who, what, offer, conversion event
Estimated time: 2-3 hr
A campaign brief defines who you're targeting, what you're selling, the offer, the conversion event (sign-up, demo, trial, purchase), and the CPA you're willing to pay. Without it, you'll optimize toward 'low CPC' instead of 'profitable customers'. Cheap clicks that don't convert are still 100% wasted spend.
Tasks
- Define the audience (job title, company size, industry, geo)
- Write the offer in 1 sentence (what they get, what they pay, why now)
- Pick the conversion event (sign-up form complete? demo booked? trial started? purchase?)
- Compute target CPA: (LTV * gross margin) / payback period in months. Conservative: target_CPA = LTV / 3.
- Decide the budget cap (start with $500-$1500 for the first 2 weeks of learning)
Pointers
- [Guide] Computing CAC payback (Reforge)
[!CAUTION] Gotchas
- Setting target CPA = LTV (no margin for risk) means you break even on every conversion. Use LTV/3 as the safer starting point.
- Targeting 'everyone in the US who might want this' produces $50 CPC and no conversions. Narrow targeting hard at the start.
Step 2: Set up conversion tracking BEFORE spending a dollar
Estimated time: 3-5 hr
If you launch without conversion tracking, you're spending money to learn nothing. Conversion tracking lets Google's algorithm optimize toward your goal and lets you measure CPA per keyword. Set up Google Tag Manager + GA4 + a Google Ads conversion action, then verify with Tag Assistant before the first dollar spends.
Tasks
- Install Google Tag Manager on the site
- Create a Google Analytics 4 property + GA4 tag in GTM
- Define the conversion event (e.g. 'sign_up_complete' fires on the success page)
- In Google Ads → Tools → Conversions: create a conversion action linked to that GA4 event
- Use Tag Assistant Companion to verify: trigger the event in test mode, confirm GA4 + Ads both receive it
- Wait for Ads to confirm 'Recording' status (24 hr) before launching the campaign
Pointers
- [Official] Google Ads conversion tracking setup
- [Tool] Tag Assistant Companion
- [Official] GA4 events to Google Ads import
[!CAUTION] Gotchas
- If conversion tracking is broken, Smart Bidding optimizes toward 'lots of useless clicks' because that's what looks good in the algorithm. Verify with Tag Assistant before launch.
- iOS 17+ + Safari ITP block third-party cookies. Use enhanced conversions (server-side or first-party) to recover the lost attribution.
- Counting both 'sign-up complete' and 'page view of /thank-you' as conversions double-counts. Pick one canonical event.
Step 3: Research keywords with Keyword Planner + competitor research
Estimated time: 4-6 hr
Keyword research at the start is about finding 20-50 high-intent queries, not 5000 vaguely relevant ones. High-intent = the searcher is ready to buy or sign up. 'Best CRM for solopreneurs' beats 'CRM' by 100x in conversion rate. Pull from Keyword Planner, then layer Ahrefs / SEMrush to find competitor terms Google's silo doesn't show you.
Tasks
- Brainstorm 5 seed terms based on the campaign brief
- Run them through Keyword Planner: get volume, competition, suggested bid
- Filter to keywords with: clear intent, search volume 100-10000/month, your max bid <= 2x suggested bid
- Pull competitor ads via Google search + tools like Ahrefs / SpyFu
- Categorize keywords into 3-5 ad groups (one ad group per intent cluster)
- Pick 5-15 keywords per ad group, no more — tight ad groups = relevant ads = lower CPC
Pointers
[!CAUTION] Gotchas
- Single-word keywords ('CRM', 'analytics') burn budget on irrelevant queries. Use 2-4 word phrases minimum.
- Branded competitor keywords ('Salesforce alternative') convert well but you'll often pay $20-$50/click. Budget accordingly.
- Keyword Planner's volume numbers are bucketed (10-100, 100-1K, 1K-10K). Treat them as orders of magnitude, not exact.
Agent prompt for this step
Research keywords for this Google Ads campaign.
Read the Brief. Output:
1. 30-50 candidate keywords organized into 3-5 ad groups by intent
2. For each: search volume estimate, intent score (1-5), match type recommendation (exact / phrase)
3. A 'do not use' list — vague-intent variants of these keywords that will burn budget
4. 5-10 competitor ad headlines you observed in the search results
5. 20-30 negative keyword candidates to add as a starting list (free, careers, jobs, salary, login, customer-service variants)
Constraints:
- Skip keywords with <100 search volume — too thin to test
- Flag any keyword with suggested bid >$20 — those are expensive to test for first-time advertisers
- Group by intent, not by topic ('best vs cheap vs how-to' are different ad groups)
Step 4: Pick match types deliberately — broad match is the default trap
Estimated time: 1-2 hr
Match type controls what queries trigger your ad. Google's default is 'broad match' which matches anything Google decides is 'related'. For first-time campaigns, broad match torches budget on queries you'd never approve. Start with phrase match + exact match. Add broad match only after you have 30 days of conversion data and a hard negative keyword list.
Tasks
- For each keyword: pick exact (
[exact match]) for proven converters, phrase ("phrase match") for testing - Avoid broad match for the first 30 days unless you have a watertight negative keyword list
- If using Performance Max or Smart Campaigns: understand they USE broad match aggressively. Skip them at the start.
- Document match-type choice per keyword in the Keywords table
Pointers
- [Official] Match types explained (Google)
[!CAUTION] Gotchas
- Google Ads' Smart Campaigns burn 80% of budget on irrelevant queries. They look easy and are a tax on first-time advertisers.
- Broad match modified (BMM) was killed in 2021. Don't use legacy BMM syntax — Google rewrote it as phrase match.
- Performance Max hides the search query data, which means you can't audit it. Skip until you have intuition for what works.
Step 5: Build a 50-200 negative keyword starter list
Estimated time: 2-3 hr
Negative keywords are the keywords you DON'T want to match. Without them, a 'CRM software' campaign will match 'CRM software jobs', 'CRM software salary', 'free CRM software', 'CRM software for nonprofits'. Negatives save 30-50% of budget on day 1 and improve ad relevance scores (lower CPC).
Tasks
- Pull standard negative seed list (free, jobs, careers, salary, login, customer-service, definition, meaning, login, support)
- Pull competitor brand names if you don't want to bid on competitor traffic
- Pull industry-specific irrelevant terms (e.g. for a B2B SaaS: 'consumer', 'personal', 'home use')
- Group negatives into a shared negative keyword list at the account level
- Apply the list to all campaigns by default
Pointers
[!CAUTION] Gotchas
- Without a negative keyword list, 30-50% of clicks on day 1 will be irrelevant queries that don't convert.
- Negatives applied at the ad-group level don't propagate to other ad groups. Use account-level shared lists.
- Don't add too many negatives at once — you'll over-prune. Add aggressively in the first week, taper after.
Step 6: Write 3-5 ad copy variants per ad group
Estimated time: 2-3 hr
Responsive Search Ads (RSAs) are the only ad format Google offers now. You write up to 15 headlines + 4 descriptions; Google mixes them. The strongest performers usually have: (1) the keyword in headline 1, (2) a specific outcome in headline 2, (3) social proof in headline 3, (4) a CTA in headline 4. Test 3-5 variants per ad group, refresh weekly based on what converts.
Tasks
- For each ad group: write 5-8 headlines (30 chars max each)
- Write 2-4 descriptions (90 chars max each)
- Pin headline 1 and headline 2 if you need brand or compliance control
- Add at least 4 ad extensions: sitelinks, callouts, structured snippets, call extension
- Run 2-3 RSA variants per ad group, label them so you can compare
Pointers
- [Official] Responsive Search Ads guide (Google)
[!CAUTION] Gotchas
- 30 chars is brutal — write the headline, count, cut. Most first-draft headlines are 35-45 chars.
- Responsive Search Ads with only 3 headlines penalize ad strength score. Use all 15 slots even if some are slight variations.
- Pinning every headline kills RSA's performance benefit. Pin only when you must (legal / brand requirement).
Agent prompt for this step
Draft ad copy for this ad group.
Read the Brief + the Keywords table for this ad group.
Output:
1. 8 headlines (30 chars max each), each labeled by intent: keyword-match / outcome / social-proof / urgency / CTA
2. 3 descriptions (90 chars max each)
3. 4 sitelink suggestions (link text + 90-char description)
4. 4 callout extension suggestions (25 chars each)
5. A note on which headlines should be PINNED (brand or compliance constraints) vs let Google rotate
Constraints:
- Headlines under 30 chars (count carefully — emoji and punctuation count)
- No 'best ever', 'amazing', 'revolutionary'. Specific outcomes only.
- Include the keyword in at least 3 headlines for ad relevance score
- Don't claim 'free' if the offer is a free trial that requires credit card
Step 7: Make the landing page match the ad — quality score lives or dies here
Estimated time: 1-2 days for the first dedicated LP, fewer thereafter
The landing page determines half your CPC via Quality Score. A landing page that mentions the searched keyword in the H1, loads in <2.5s, and has a clear single CTA gets a high quality score = lower CPC. A homepage targeting 5 audiences gets a low quality score = higher CPC. Build a dedicated landing page per ad group if budget allows.
Tasks
- Build a dedicated landing page per ad group (or at minimum: homepage variant per audience)
- Match the H1 to the search intent (use the keyword)
- Single CTA above the fold; remove nav links if possible
- Page load <2.5s on Lighthouse mobile (compress images, lazy-load below-fold)
- Conversion event fires on form submit, not page load
- A/B test the headline + CTA via PostHog or Optimizely
Pointers
- [Guide] Landing page best practices (Unbounce)
- [Official] Quality Score explained (Google)
[!CAUTION] Gotchas
- Sending Google Ads traffic to your homepage is the #1 first-time mistake. Homepages target 5 audiences; ad clicks target 1 intent.
- Quality Score 1-3 means you're paying 2-3x what you should per click. Fix the landing page before you scale spend.
- Adding a chat widget that auto-pops can tank the conversion rate by 20-30%. Test before keeping.
Step 8: Launch with a small daily budget and monitor for 7 days
Estimated time: 1 hr to launch, 7 days of patient monitoring
Launch with a daily budget of $20-$50. Don't change anything for 7 days — Google's algorithm needs the data to calibrate. Resist the urge to pause keywords on day 2 because they're 'expensive'. After 7 days you'll have enough conversion data to start making optimizations.
Tasks
- Set daily budget to $20-$50 (Google may spend up to 2x daily on individual days)
- Start with manual CPC bidding, NOT smart bidding (smart bidding needs 30+ conversions to work)
- Set max CPC to your target CPA / 5 (rough heuristic, assumes 20% CR)
- Launch the campaign — verify spend starts within 24 hr
- Watch the search-terms report daily, build the negative keyword list as irrelevant queries appear
- Don't pause / change anything for 7 days unless spend is wildly off (2x budget) or conversions are 0 after 100+ clicks
[!CAUTION] Gotchas
- Google will show 'Limited by budget' warnings on day 1. That's normal. Don't react.
- Pausing keywords on day 2 because 'no conversions yet' is the #1 way to break a campaign before it has data. Wait 7 days.
- Smart bidding (Target CPA / Maximize Conversions) needs 30+ conversions in the last 30 days to work. Until then, manual CPC.
Step 9: Read the search-terms report daily — it's where the money leaks
Estimated time: 30 min/day for the first 2 weeks, then weekly
The search-terms report shows the actual queries that triggered your ads. It's where you find out 'CRM software for small business' triggered ads on 'CRM software definition' (which converts at 0%). The first 2 weeks should be aggressive negative-keyword building from the search-terms report — it's the single highest-leverage optimization you can do.
Tasks
- Daily for 14 days: pull the search-terms report
- For each search term: did it convert? Is it relevant? Should it be a new keyword OR a negative keyword?
- Add converting search terms as exact-match keywords (lock in the win)
- Add irrelevant search terms to the negative keyword list
- Track the 'wasted spend' metric: $ spent on queries that won't convert. Aim to drive it under 20% by week 3.
[!CAUTION] Gotchas
- The search-terms report only shows ~70% of queries (Google hides low-volume ones). What you see is the iceberg's tip.
- A search term that converted once isn't necessarily a winner — wait for 5+ clicks before promoting to its own ad group.
- Bot traffic spikes on cheap CPC keywords (gaming, downloads). If conversion rate looks too good, audit form-fill quality.
Agent prompt for this step
Process today's search-terms report.
I'll paste the export. Output:
1. Top 10 converting search terms — propose adding as exact-match keywords if not already
2. Top 10 wasted-spend search terms (high spend, no conversions, irrelevant intent) — propose adding as negative keywords
3. Any 'long-tail surprises': search terms that converted at unusually low CPA — flag for new ad-group creation
4. The week-over-week trend in wasted-spend ratio
Constraints:
- Don't propose negatives for terms that converted, even once
- Flag conversions that look like bot/click-fraud signals (sign-up < 10 sec on landing page)
- Output proposed changes as draft rows in the Keywords table; don't apply directly
Step 10: After 30 days: switch to smart bidding if you have 30+ conversions
Estimated time: 2-3 hr to switch + 14 days to recalibrate
After 30 days you'll know your CPA per keyword + ad group. If you have 30+ conversions in the last 30 days, switch to Target CPA bidding — Google will optimize bids per auction with more data than you have. If you have <30 conversions, stay on manual CPC and broaden your funnel (more keywords, longer time window, lower friction conversion event).
Tasks
- Audit conversion volume in the last 30 days
- If 30+ conversions: switch to Target CPA bidding at your target CPA + 10% (give algorithm room)
- If <30 conversions: stay on manual CPC, broaden the funnel
- Expect performance to dip for 7-14 days after switching while the algorithm recalibrates — don't undo the switch
- After recalibration: compare CPA + conversion volume against the prior manual-CPC baseline
[!CAUTION] Gotchas
- Switching to smart bidding too early (under 30 conversions / 30 days) makes Google bid blind — CPA spikes.
- Target CPA set 50% below current CPA causes Google to throttle spend hard. Set it at +10% of current to start.
- Smart bidding ignores manual bid changes. Don't try to 'help' the algorithm by tweaking bids underneath it.
Step 11: Decide weekly: scale, pause, or pivot
Estimated time: 2-3 hr/week
After the calibration period, every campaign + ad group + keyword is in one of three buckets: scale (CPA < target, increase budget), pause (CPA > target * 2, no path to fix), pivot (CPA > target but trend is improving, change something specific). Make the call weekly. Indecision is the most expensive optimization.
Tasks
- Weekly: rank ad groups by CPA
- Scale: ad groups with CPA < target — increase budget by 20-30% (not 100%, that breaks Google's optimization)
- Pause: ad groups with CPA > 2x target after 30+ days — pause, don't 'try one more week'
- Pivot: ad groups with CPA > target but trending down — change one variable (LP, ad copy, audience, bid)
- Document each decision in the Brief with reasoning + date
[!CAUTION] Gotchas
- Doubling budget on a working ad group resets Google's optimization for ~7 days. Increase 20-30% at a time.
- Keeping a 2x-target-CPA ad group running 'just in case' is a $1000s/month tax. Pause and re-test from a clean slate.
- Pivoting on every variable simultaneously means you can't tell which change worked. Change one thing per week.
Hand the template to your agent
Paste the prompt below into your agent's permanent system prompt so the agent reads, writes, and maintains this workspace as you work through the steps.
You are an agent on the "Launch your first Google Ads campaign" playbook workspace.
Your role: maintain the four surfaces (Campaigns, Keywords, Brief, Pointers) as the campaign goes live and runs.
Cadence:
- Each morning: pull yesterday's spend + conversions + CPA per campaign + per keyword. Flag anything that's drifted >50% from baseline.
- When a keyword has spent 2x its target CPA without converting: propose pausing it.
- When the search-terms report shows new converting queries: propose adding them as exact-match keywords.
- When the search-terms report shows irrelevant queries that wasted budget: propose adding them to the negative keyword list.
- Weekly: refresh ad copy variants based on which headlines/descriptions converted best.
First MCP tool calls:
1. list_surfaces(workspace_slug="launch-a-google-ads-campaign")
2. list_rows(workspace_slug="launch-a-google-ads-campaign", surface_slug="campaigns")
3. get_doc(workspace_slug="launch-a-google-ads-campaign", surface_slug="brief")
Hard rule: never push changes directly to Google Ads. Propose them as draft rows the user reviews and applies.
FAQ
How much should I spend on my first Google Ads campaign?
Budget $500-$1500 for the first 2-3 weeks of learning. That's enough to get conversion-tracked data on 30-50 keywords without burning $5K on a calibration phase. Scale only after you have a working ad group with CPA below target — not before.
Should I use Smart Campaigns / Performance Max for a first launch?
No. Smart Campaigns and Performance Max abstract away the controls (match types, search-terms report, negative keywords) you need to learn what works. They burn 80% of first-time-advertiser budgets on irrelevant queries. Run a regular Search campaign with manual CPC + tight match types until you have intuition.
How long until I see results?
Day 1: spend starts. Week 1: enough click data to see search-terms patterns and start the negative keyword list. Week 2-3: enough conversion data to know your real CPA per keyword. Week 4+: optimization mode. If you don't have a single conversion in 100+ clicks by week 2, the funnel is broken (LP fit, conversion tracking, audience), not the bidding.
Can my AI agents help run the campaign?
Yes for the operational work: drafting ad copy variants, processing the daily search-terms report, flagging wasted-spend keywords, proposing negatives, watching CPA drift. Not for the strategy decisions (which keywords to target, what offer to test, when to scale). The agent watches; the human decides.
What's the most common reason first campaigns fail?
Three reasons in order: (1) conversion tracking was broken or never set up — you spent money learning nothing, (2) match types defaulted to broad and burned budget on irrelevant queries, (3) the landing page didn't match the ad intent so quality score was low and conversion rate was 0.5%. All three are preventable; this playbook's first 6 steps exist to prevent them.