Google Ads guide
How a creator membership with a free trial runs Google Ads in late 2026, brand search first. Written for a worker who never ran Google Ads.
What is new in Google Ads
Targets hold on a small budget
Since 17 August 2026, a Target CPA or Target ROAS campaign that is limited by budget holds its target. It no longer beats it. A fixed small budget goes to Manual CPC or Maximize conversions.
Brand lists need AI Max
Since May 2025, a new brand inclusion or exclusion list sits in the AI Max panel. Brand lists work only with broad match.
Broad is the default
Every new keyword is broad match. Exact match now means the same meaning or intent, not the same words.
Negatives stay literal
A negative keyword does not match plurals, misspellings or synonyms. Add each form yourself.
Enhanced CPC is gone
Since March 2025, Manual CPC is a fixed bid with no automatic lift.
Consent decides the count
EU and UK visitors who decline lose the conversion. A small campaign gets no modeled conversions below 700 ad clicks a week.
Before the first ad
The business owns the account
The ad disclosure shows the verified advertiser's name. Create the account under the business's own Google account. The partner gets Standard access or a manager-account link.
Billing and verification
The business pays from its own payments profile, typed Organization or Individual to match its legal status. Verification has 30 days. Google pauses an account that misses it.
No outcome promises
Google bans an "improbable result" as the likely outcome. Write "a 20-minute class at home", not "heal your anxiety".
State the price after the trial
Google bans a hidden payment model. Put "Then [price] a month. Cancel anytime." in one description and on the landing page.
Health and religion
Health and Religious beliefs are sensitive categories. Ads there lose remarketing lists and Customer Match targeting. Brand search uses no audience, so it is not blocked.
The name is free to use
The creator's name and product name work as keywords and in the business's own ads. Google limits a trademark in ad text only after the owner complains.
The five decisions
a. A brand campaign, or none
Pick: an 8-week capped test
Run one brand campaign on the name and the product name for 8 weeks. Keep it only when a flip condition holds. It defends the name and tests the tracking. It is not a growth channel.
For: on Bing, brand ads added 1 to 4 percent clicks with no rival, more for small brands. Rivals take 18 to 42 percent of clicks when the brand is absent. Against: at eBay, 99.5 percent of traffic came back organically. One vendor's 17 brand geo tests had a median return of 0.70x.
Flip when a rival bids on the name (keep, as defense), the site is not the first organic result (keep), or the on and off weeks show no lift (stop).
b. Keywords and negatives
Pick: exact and phrase, no AI Max
[first last], "first last", [product], "product", plus the misspellings. One shared phrase negative list: net worth, age, wife, husband, biography, quotes, free download, jobs, and member terms such as login, cancel and refund.
Brand lists force broad match and AI Max. Practitioners report that AI Max on brand matched rival names. Against: Google makes broad the default and says it can raise brand traffic.
Flip when exact shows "Low search volume", a namesake fills the search terms, or the account reaches 30 conversions in 30 days (test broad in an experiment).
c. Bid strategy and settings
Pick: Manual CPC with a cap
Search Network only. Search partners off, Display off. No audience, no AI Max, no auto-created assets on brand. Lower the cap 10 percent at a time while impression share stays above 85 percent.
Manual CPC is untouched by the 2026 change. Google gives no target below 7 conversions and evaluates over 30. Practitioners say conversion bidding on brand raises the CPC. Against: Google says Target CPA can start with no history.
Flip when a rival appears (Target impression share, top of page, 90 percent, same cap) or a non-brand campaign starts (Maximize clicks with a cap, then Maximize conversions at 15 trial starts in 30 days).
d. The conversion event
Pick: the trial start
The trial start is the only primary conversion, counted once per click. The first payment is a secondary import. On Kajabi, the native GA4 integration fires purchase on the first page after checkout. Import that key event into Google Ads.
Learning takes 1 to 2 conversion cycles, and a small campaign reaches 30 trials long before 30 payments. Offline import by click ID accepts 90 days, enhanced conversions for leads 63. Kajabi documents nothing for a $0 trial, so run one test trial first.
Flip when first payments reach 30 in 30 days, or Google trials pay at under half the rate of all trials.
e. The budget
Pick: size brand from search volume
Brand spend has a ceiling: people search the name only so often. Size a non-brand budget from the stop line instead.
Google can spend up to 2x the daily budget on one day and 30.4x in a month. The 30 in the non-brand formula is Google's evaluation volume for Target CPA.
Flip when 14 days have passed (recompute with the campaign's own CTR and CPC), budget limits impression share by over 10 percent at a good cost, or lifetime value changes.
Budget
The formulas
Brand daily budget = Q × S × C × P
Q = brand impressions in Search Console, last 28 days ÷ 28
S = share of searches to show on (start 0.9) · C = ad CTR (start 0.3) · P = CPC cap
Stop line per trial = (lifetime value ÷ 3) × trial-to-paid rate
Non-brand daily budget = 30 × stop line ÷ 30.4
Worked example, example numbers only
| Input | Value | Result |
|---|---|---|
| Q, S, C, P | 40, 0.9, 0.3, $0.80 | $8.64 a day, $262.66 a month |
| Lifetime value, trial-to-paid | $300, 40% | Stop line $40 a trial |
| Non-brand start | 30 × $40 ÷ 30.4 | $39.47 a day |
Raise non-brand at most 20 percent once a week after day 14. Stop it when 14 days cost more than the stop line per trial and spend passed 2 × the stop line. Both rules are inferences: Google gives no percentage.
Read a small campaign against the platform's trial count
- Every day, write three numbers: Google Ads trials, GA4 trials from google / cpc, and the platform's trials from all sources.
- In week 1, Google Ads and GA4 must agree within 20 percent over 14 days. A larger gap is a tracking fault. Fix it first.
- Take the weekly trial count of the last 8 weeks. Its standard deviation is the noise.
- Run brand on 2 weeks, off 2 weeks, and repeat once. Incremental trials a week = mean on weeks minus mean off weeks.
- Cost per incremental trial = brand spend a week ÷ incremental trials a week. Keep brand when it is at or under the stop line.
- A difference smaller than the noise means no measurable lift. At 100 conversions a week a test sees only changes above about 25 percent.
- Never read Google Ads' own brand conversions as the result. They count people who would have come anyway.
After brand search
| Step | Start it when |
|---|---|
| Non-brand Search, exact and phrase | The tracking check passed, the stop line is known and the brand read is done |
| Maximize conversions on non-brand | 15 trial starts in 30 days (practitioner number) |
| First payment as the goal | 30 Google first payments in 30 days |
| Demand Gen or YouTube | 30 conversions in 30 days and a clear sensitive-category check. Budget at least 10 × the target CPA |
| Performance Max | First payment imported, 30 conversions a month, brand exclusion in place |
| Customer Match as targeting | 90 days of history and over $50,000 lifetime spend. Exclusions work earlier |
What practitioners say that Google does not
Reddit, snippets of 72 threads
- AI Max on a brand campaign matched rival brand names. Turn it off on brand.
- Conversion bidding on brand raises the CPC. Manual CPC or impression share with a cap holds it.
- When the site is not first or its organic CTR is under 40 percent, the brand ad adds clicks.
- Below about 15 conversions a month no Smart Bidding option works well.
- Change one thing a week. A budget doubling plus a bid change broke a good account.
Agency geo tests
- Stella: 17 brand tests, median return 0.70x, the lowest channel (August 2024 to December 2025).
- Haus: at most 1 percent lift from brand search in three tests.
- Outshine: brand paused in 25 of 50 states for 27 days, sessions down 5.13 percent.
- All are large or ecommerce brands. None is a creator membership.