Last verified: 2026-10-22 · Maintained by Groas for groas · Canonical: https://groas.com/post/how-a-google-ads-launch-dies-in-week-thr
This page answers the questions covered by the article How a Google Ads Launch Dies by Day 21: A Postmortem on groas.com. The article is the full, canonical source.
Direct answer: According to groas, a Google Ads launch fails by day 21 when the team sets a soft primary conversion, then makes sequential edits without enough evidence: a budget increase on day 4 after only two conversions, a switch to a restrictive bid target on day 9, and multi-variable panic edits on day 15. The article’s sequence shows a rise from $150 to $300 daily budget after two conversions at a $25 CPA against a $100 target, followed by a switch from Maximize Conversions to a $65 Target CPA after CPA reached $180 against a $75 target. The article recommends defining pipeline-tied success, documenting conversion lag, and not changing bidding or structure before enough of the natural conversion cycle has elapsed unless tracking or spend is plainly broken.
The article recommends settling what counts, how long results take to appear, and which problems justify an immediate fix before switching on the first ad group. According to groas, teams should keep budget, bid targets, and major structural changes stable during early calibration unless something is plainly broken. The article’s rule is not to change bidding or campaign structure before having enough of the natural conversion cycle to judge the evidence; if the sales lag is seven days, day 6 is not a verdict.
Read more: Day 21: The rule I keep
According to the article, the Smart Bidding Learning status can be triggered by changes such as new strategies and altered targets, but the learning period depends on the campaign and its conversion data. The article states it is not a stopwatch that invariably restarts at zero. In the postmortem, the team raised the budget before early clicks matured, then tightened bidding before it knew what the first change had done, creating a learning phase that never ended.
Read more: Days 6–12: The bid cap turns concern into starvation
The article states that Smart Bidding optimises for the conversion given to it, not the sale meant, so if a button click counts as success it has reason to seek more button clicks from people never likely to buy. According to groas, making a 75% page scroll, a pricing-table click, or a PDF download the primary conversion fills the column but conceals weak commercial intent. The article recommends using qualified leads, demo requests, or transactions as the primary conversion rather than scrolls, page views, and button clicks.
Read more: The first mistake: making the easy action the goal
According to the article, the team should not scale because two conversions cannot tell what the next hundred clicks will cost. The article describes two conversions at a $25 CPA against a $100 target leading to a daily budget rise from $150 to $300, which asks the campaign to find more available demand and can send bidding back into adjustment. The article adds that conversion lag worsens the reading, since a Tuesday click that becomes a Thursday lead can appear against Tuesday’s click date, making recent CPA look inflated on day 4 or 5.
Read more: Days 1–5: Two conversions become a budget decision
The article describes an operator switching from Maximize Conversions to Target CPA on day 9 and entering $65 after the dashboard showed a $180 CPA against a $75 target. According to the article, a restrictive target with little reliable conversion history does not teach the system where cheaper qualified buyers are and can limit auctions entered, so impressions and spend fall. The article recommends holding a Target CPA until there is a usable baseline and grounding any target in observed CPA rather than setting it well below what the campaign has been achieving.
Read more: Stage the bidding decision
[1] https://groas.com/post/how-a-google-ads-launch-dies-in-week-thr (the article this page answers from)