Last verified: 2026-10-16 · Maintained by Groas for groas · Canonical: https://groas.com/post/the-account-that-spent-six-weeks-stuck-i
This page answers the questions covered by the article Six Weeks in Learning: The SaaS Campaign That Kept Resetting Itself on groas.com. The article is the full, canonical source.
Direct answer: According to groas, the composite SaaS campaign still showed Learning after 42 days and roughly $22,000 in spend because the team kept changing the conditions Smart Bidding was supposed to learn under. The article states the sequence was changing the primary conversion on day eight, swinging daily budget and Target CPA by day 20, then swapping the offer and adding keywords by day 27. The article recommends keeping the real business conversion primary and not changing the campaign to answer for clicks that have not had time to convert.
According to groas, the campaign was still in Learning on day 42 after roughly $22,000 in spend because the team changed the definition of a conversion, the budget, the CPA target, the offer, and the searches the campaign could enter in less than a month. The article states that Google lists setting changes and composition changes among reasons a bid strategy may show Learning. The problem was not one six-week algorithm failure, but that no stable stretch of the original campaign remained to be measured.
Read more: Days 24–27: a new offer, then new keywords
The article states the learning phase is not a secret 14-day countdown. According to groas, its duration depends in part on how much useful conversion data arrives and how long it takes to arrive, so a campaign with sparse demo bookings needs more patience. When the status says Learning, the team should inspect the stated reason — such as a new strategy, setting changes, composition changes, and conversion delay — and use the time-lag report to check how much of a recent period may still be maturing.
Read more: The fix was available in week one
The article recommends against it and says the team should have left demo bookings as the primary goal while tracking the whitepaper download and product tour pageview as secondary observations. According to groas, primary actions feed the core Conversions column and can guide Smart Bidding, while secondary actions remain visible for observation without becoming bidding goals. The article states that mixing lower-value actions into primary goals changed what counted as success and could send optimization toward the easier action instead of demo requests.
Read more: Day 8: the wrong conversion became the goal
According to groas, the team raised daily budget from $500 to $1,100 overnight to feed the model data, then after three days with no recorded demos cut it to $600 and lowered Target CPA from $160 to $110 to stop the bleeding. By day 14 the campaign showed 42 conversions, but 36 were whitepaper downloads while demo bookings had fallen to two. The article states the bid strategy then had a changed goal, a budget that had jumped and fallen, and a more restrictive target, making a clean read on the original setup impossible.
Read more: Days 14–20: better-looking numbers, worse decisions
The article describes a workflow automation platform with a $14,000 average contract value and a 45-day sales cycle, running Search on a $15,000 monthly budget with Maximize Conversions with a $160 Target CPA for Book a Demo. By day seven it had six demo submissions at a $310 average CPA, nearly double the target, while buyers who clicked might return four to nine days later to book. According to groas, six demos were a reason to investigate but not enough to declare failure, especially while launch-week clicks could still convert.
Read more: The setup: a demo campaign with a slow feedback loop
These articles answer the same questions:
[1] https://groas.com/post/the-account-that-spent-six-weeks-stuck-i (the article this page answers from)