--- title: "The SaaS Campaign That Spent 81 Days in Learning" description: "A SaaS account changed its offer four times in one quarter. Here’s how unvetted changes kept its Google Ads campaign learning instead of building a stable pipeline." image: "https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab7345e686cd4d54c331680_ec82f74e-f18d-4ae7-8178-d2ec9c0873f9.png" --- September 26, 2026 • min read # The SaaS Campaign That Spent 81 Days in Learning ![Young man with curly hair wearing a black shirt outdoors against green foliage background.](https://cdn.prod.website-files.com/6821efca072e48f6f495a47e/68562d390107b3921a6e3d68_1743932904108.jpg) **Alexander Perleman**, Head Of Product @ groas Ex-Goldman Sachs and Stanford Computer Science alex@groas.ai [LinkedIn](https://groas.com/post/the-campaign-that-never-left-the-learnin#) ![Cover image for: The SaaS Campaign That Spent 81 Days in Learning](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab7345e686cd4d54c331680_ec82f74e-f18d-4ae7-8178-d2ec9c0873f9.png) The bid strategy spent 81 days of a 90-day quarter in `Learning`. I ran the SaaS account, and the mistake was not turning on Smart Bidding. It was letting offer and landing-page changes hit a live campaign without a hold. The client changed its core offer four times that quarter. The campaign spent $42,000, generated sixteen qualified sales opportunities, and finished with a cost per acquisition nearly triple its target. On the monthly executive calls, leadership blamed Google's automated bidding. They called it erratic, said it couldn't understand enterprise software buyers, and wanted to move the account back to manual cost-per-click management. I understood the frustration. I also watched what happened before each swing in performance: someone changed the offer, the conversion action, or the page the ads sent people to. The bidder kept receiving a new definition of success just as it began to learn the old one. Here is the quarter in order. #### Week one: the campaign had a workable baseline The client sold contract workflow software to in-house legal teams at an average contract value of $18,000. It had a thirty-day sales cycle and a mid-market sales motion. We launched two focused search ad groups around high-intent queries such as `legal workflow automation` and `contract lifecycle management software`. Both led to a dedicated demo request page whose language matched the searches. We ran Maximize Conversions with a $240 Target CPA, below the client's $300 economic ceiling. By day seven, the campaign had logged twelve qualified demo requests at an average CPA of $195. Search terms were clean, delivery was steady, and the bid strategy showed `Learning`. That status made sense for a new campaign. As [Google's guidance on bid strategy statuses](https://support.google.com/google-ads/answer/6263057) explains, a strategy needs time and conversion data to calibrate. The point of those first seven days was not that we had beaten the target for good. We had a promising start and, more importantly, **one consistent conversion goal**. The campaign could gather data about which auctions produced demo requests while the sales team checked whether those requests belonged in the pipeline. We needed to leave that definition alone long enough to see what it meant. We did not. #### Day ten: a report download became a “demo” The client held an executive offsite on day ten. The VP of Sales came back convinced that a twenty-minute demo was too much to ask of in-house counsel arriving from cold search traffic. Without checking with the performance team or reviewing the conversion setup, the marketing director replaced the primary website call-to-action with a gated benchmark report. Then they pointed the existing Google Ads conversion action at the report download page. They kept the tag name so the internal weekly dashboard would not break. On that dashboard, a conversion still looked like a conversion. In the auction, the event behind the label had changed completely. ![Campaign notes and conversion graphs scattered across an office desk.](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab73460686cd4d54c3316da_2a164bb2-8add-4000-ac54-a1aaf223e467.png) Report downloads converted at 14%; live demo bookings had converted at 3.5%. Smart Bidding could see the higher rate, but it could not know that the company had lowered the bar from *talk to sales* to *get a free PDF*. It bid more aggressively across broader query variations and top-of-page placements. Within forty-eight hours, the campaign burned through three days of budget, and average CPC rose from $14.20 to $28.50. Four days later, the sales team found sixty new form submissions in Salesforce. None was an enterprise decision-maker ready to evaluate the software. They were junior legal assistants and law students downloading a report. The VP ordered an immediate return to the demo offer. That reversal produced the first visible reset: the bid strategy returned to `Learning (Bid strategy changed)`. The recent conversion data now included cheap PDF downloads, while the campaign was once again asking for demo bookings. If you want the mechanics behind that status, [our guide to Google Ads Smart Bidding learning periods](https://groas.com/post/google-ads-smart-bidding-strategies-2026-learning-period-guide) goes deeper. The operational point here is simpler: **a conversion tag is not a stable goal if the action behind it keeps changing**. #### Weeks three to five: the page changed, then the offer changed again In week three, demo bookings were beginning to form a modest baseline again. Then the client's internal team migrated its CMS as part of a sitewide brand overhaul. On a Tuesday evening, they replaced our focused landing page with an interactive product tour on a new URL path. The original headline, `Contract Lifecycle Management for Enterprise Legal Teams`, gave searchers a direct answer to what they had looked for. Its replacement, `Reimagine How Work Flows`, did not. The demo form also moved behind a multi-step modal that took three clicks to reach. The redesign may have served the brand presentation, but it made the paid-search path harder to follow. ![A focused landing page beside a cluttered redesign with weaker performance metrics.](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab73460686cd4d54c3316e1_93f5c83c-30ba-45ec-af7d-e215fed55e65.png) The new destination URL required ad re-review, temporarily reducing impression volume during weekday auction hours. The page also lost much of the language that had connected the ads to the searches. Within seventy-two hours, average keyword Quality Score fell from 8/10 to 4/10, and Landing Page Experience ratings moved from `Above average` to `Below average`. Quality Score was a diagnostic warning, not a switch that directly set our CPC. Poorer relevance and landing-page experience can weaken ad quality in the auction, though, and our effective CPC doubled as we tried to hold the same positions. [Google's Quality Score documentation](https://support.google.com/google-ads/answer/6167118) explains the diagnostic, and [our analysis of AI-driven Quality Score optimization](https://groas.com/post/google-ads-quality-score-ai-driven-optimization-beyond-basic-rules) covers the broader account work. In this campaign, the warning was hard to miss: the page built for those queries had disappeared. At the end of week five, with CPCs climbing and weekly demo volume falling, the executives held an emergency strategy session. They decided enterprise legal buyers were intimidated by the long sales cycle and introduced a third offer: a self-serve 14-day sandbox trial. They added a conversion tag for trial signups, marked it as an equal `Primary` action in Google Ads, and left the demo action active too. Now the bidder could count a curious solo practitioner starting a trial the same way it counted a twenty-seat enterprise demo request. With demo volume scarce, the lower-friction action became the easier one to find. The campaign spent $3,800 over the next six days chasing free signups that never activated. The offer had changed again, but the more damaging decision was **giving two very different outcomes equal weight**. By week eight, the sales development team had seen enough. The trial users were hobbyists with personal email addresses who vanished when their fourteen days ended. Leadership pushed a fourth offer: a fifteen-field `Enterprise ROI Assessment`. At the same time, trailing CPA had passed $580, so they cut the daily campaign budget from $500 to $200. A 60% budget cut would have disrupted delivery on its own. Paired with a form asking for company revenue and IT architecture before showing a number, it left the campaign with almost no conversion signal. For six consecutive business days, it recorded zero conversions. The team wanted to cap exposure. Instead, it changed the conditions again before the previous test had produced a useful answer. #### Day ninety: the graph showed a reset loop The ninety-day bid strategy report made the sequence plain. Smart Bidding uses auction-time signals to estimate which clicks are likely to produce the conversion it has been told to pursue. Device, location, time of day, query, and conversion history all help it make that estimate. But yesterday's conversion data is less useful when today's “conversion” means something else. The client changed a core variable every two to three weeks. The bid strategy showed `Learning` for eighty-one of the ninety days. That label did not mean every bid was wrong; it meant we had not given the strategy a sustained period under stable conditions. The standard deviation of daily CPC was 210% higher than the established baseline. We spent $42,000 across a quarter in which each apparent recovery was followed by another change to the thing we were measuring or the page we were asking people to use. ![Timeline of offer changes and repeated learning periods across the ninety-day campaign.](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab73460686cd4d54c3316ea_b784020f-a5a6-4b93-a4cd-70040983ff9e.png) I cannot blame the bidder for failing to infer the client's real priority from contradictory instructions. A demo request became a report download, then a demo again. A focused page became a product tour. A sandbox signup gained the same status as an enterprise demo. Finally, a demanding assessment arrived with a sharply reduced budget. **The account never had a long, stable run at one definition of success.** That is the graph's useful conclusion. Not “automation doesn't work,” and not “never test a new offer.” The problem was where those tests landed. #### The wrong call was letting tests touch the live model Growth-stage companies test positioning. They should. A gated report, a sandbox trial, and an ROI assessment were all offers the client was entitled to try. What they were not entitled to expect was a reliable read on the existing demo campaign while changing its goal underneath it. The decision that could have protected the quarter was an **operational hold on unvetted changes to the live campaign**. A new offer belonged in an isolated experiment campaign with its own conversion action, budget, and bidding strategy. The existing demo campaign needed to keep measuring demos. Otherwise, a test of a new proposition also became an unplanned test of what would happen when the bidder's training signal changed mid-flight. ![Ad auction bidding engine shown behind operational guardrails and change controls.](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab73461686cd4d54c3316f7_937531e0-a666-4a27-a7f3-81295d25fad3.png) I have seen the slower version of this failure in traditional account management: a web team ships a change on Tuesday, the media buyer reviews the account on Friday, and everyone spends the next meeting discussing “auction volatility.” In this case, the client-side changes were visible in the timeline. They still reached the active campaign before anyone held them for review. That is part of my frustration with percentage-of-spend agency work. A weekly check-in is a poor defense against a conversion tag repointed between check-ins, and the fee does not fall when the budget goes toward form fills the sales team cannot use. The useful work is catching the change and deciding whether the campaign can safely keep spending, not writing a polished explanation after the money is gone. #### The rule I use now When I launch a new search build or substantially restructure an active ad group, I use a **change-freeze window** for the core architecture. For the first twenty-one to twenty-eight days, or until the campaign has logged at least 40 qualified conversions under a stable conversion cycle, I hold the primary conversion action, landing-page architecture, and destination URL steady. I also limit daily budget adjustments to no more than 15% in a five-day window. That is an operating rule, not a guarantee that every account will finish learning on day twenty-eight. It gives the team a chance to distinguish an actual performance problem from the effects of its own edits. If leadership wants to try a lead magnet or new positioning during that period, it can use organic channels, email, paid social, or an isolated search experiment. The live demo campaign does not have to become the test bed for every idea from an executive meeting. This is where continuous execution has an advantage over human-managed spreadsheets and periodic account reviews. At [groas](https://groas.com/), specialized models work continuously across [paid search](https://groas.com/paid-search), including bidding, query pruning, and landing-page alignment, while a named human strategist owns direction and accountability. Guardrails and action logs make changes easier to catch and examine without asking a manager to notice every problem at the next scheduled login. That does not remove the need for judgment. It gives judgment a better chance to arrive before another day's spend follows a broken signal. I once thought a clean launch and sensible target CPA gave a campaign enough room to find its feet. That account changed my mind. The launch was clean; the boundary around it was not. My rule now is simple: **if the business cannot hold the core offer, conversion action, and landing-page architecture steady, I do not let an active automated campaign absorb the changes.** Test the new idea somewhere separate. Do not pay Google $42,000 to learn a goal the company keeps replacing. ## Related Posts [![Cover image for: YouTube Has More Ads. Why Pay to Reach the Same Viewer Eight Times?](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab7364eea584b86d640ef71_a5c5edbb-cda7-4dd1-8190-432eeaca171f.png) ##### YouTube Has More Ads. Why Pay to Reach the Same Viewer Eight Times? YouTube’s heavier ad load is obvious. The quieter cost is paying for repeat impressions while frequency settings and reach reports make the spend look productive. September 26, 2026 • min read Written by](https://groas.com/post/stop-asking-if-youtube-added-more-ads-in) [![Cover image for: Five Google Ads Predictions to Test by the End of 2027](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab735d4b9190b5f663f0312_dc0d1571-28e3-4802-9aaa-84f2d6666953.png) ##### Five Google Ads Predictions to Test by the End of 2027 Five dated calls on Search structure, learning phases, setup advice, YouTube frequency and agency fees, with a test for each one. September 26, 2026 • min read Written by](https://groas.com/post/what-breaks-in-google-ads-by-2027-five-f) [![Cover image for: The Google Ads Learning Phase Isn’t a Timer: 4 Beliefs That Kept Resetting Mine](https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab735c5686cd4d54c338554_3e5d10a3-beea-4290-a2a9-a828f25f2796.png) ##### The Google Ads Learning Phase Isn’t a Timer: 4 Beliefs That Kept Resetting Mine The Google Ads learning phase depends on conversion signals and account changes, not just elapsed days. Here are four beginner beliefs that make calibration harder. September 26, 2026 • min read Written by](https://groas.com/post/what-i-believed-about-the-google-ads-lea) ![](https://cdn.prod.website-files.com/6821efca072e48f6f495a47e/6823c03839b2b618430d5ae6_Light-p-500.png) [![White stylized owl eyes with green background icon.](https://cdn.prod.website-files.com/6821efca072e48f6f495a47e/6a68a036a5dd5627409d3579_Groas%20Icon.png)](https://groas.com/post/the-campaign-that-never-left-the-learnin#) [contact](mailto:scale@groas.ai) Explore [Home](https://groas.com/)[Philosophy](https://groas.com/our-philosophy)[Blog](https://groas.com/blog) What We Do [Paid Search](https://groas.com/paid-search)[Earned Search](https://groas.com/earned-search) Who We Serve [Businesses](https://groas.com/for-businesses)[Agencies](https://groas.com/for-agencies) MCP [ChatGPT](https://chatgpt.com/plugins/plugin_asdk_app_6a9c2d7108f08191beb5b3d736be8655?q=groas)[Claude](https://claude.ai/directory/mcp-groas-ai) Legal [Privacy Policy](https://groas.com/legal/privacy-policy)[Terms of Service](https://groas.com/legal/terms) Get Started [Apply](https://groas.typeform.com/to/xC1bQNUT) © 2026 groas 🇺🇸 [![](https://cdn.prod.website-files.com/62434fa732124a0fb112aab4/62434fa732124a389912aad8_linkedin%20small.svg)](https://www.linkedin.com/company/groas/) ```json { "@context": "https://schema.org", "@graph": [ { "@type": "", "@id": "https://www.groas.com/post/the-campaign-that-never-left-the-learnin#article", "headline": "The SaaS Campaign That Spent 81 Days in Learning", "description": "A SaaS account changed its offer four times in one quarter. Here’s how unvetted changes kept its Google Ads campaign learning instead of building a stable pipeline.", "url": "https://www.groas.com/post/the-campaign-that-never-left-the-learnin", "image": { "@type": "ImageObject", "url": "https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab7345e686cd4d54c331680_ec82f74e-f18d-4ae7-8178-d2ec9c0873f9.png" }, "thumbnailUrl": "https://cdn.prod.website-files.com/6823bbd57170ea42b357cf81/6ab7345e686cd4d54c331680_ec82f74e-f18d-4ae7-8178-d2ec9c0873f9.png", "datePublished": "2026-09-26T02:56:33.756Z", "dateModified": "2026-09-26T02:56:33.756Z", "inLanguage": "en-US", "isAccessibleForFree": true, "articleSection": "", "keywords": "Google Ads, ", "mainEntityOfPage": { "@type": "WebPage", "@id": "https://www.groas.com/post/the-campaign-that-never-left-the-learnin" }, "author": { "@id": "https://www.groas.com/author/#person" }, "publisher": { "@id": "https://www.groas.com#organization" }, "about": [ { "@type": "Thing", "name": "Google Ads" }, { "@type": "Thing", "name": "Pay-per-click advertising" }, { "@type": "Thing", "name": "Performance marketing" } ], "mentions": [ { "@type": "Organization", "name": "groas", "url": "https://www.groas.com" }, { "@type": "Thing", "name": "Google Ads" } ], "speakable": { "@type": "SpeakableSpecification", "cssSelector": ["h1"] } }, { "@type": "Person", "@id": "https://www.groas.com/author/#person", "name": "", "description": "", "jobTitle": "", "email": "", "url": "https://www.groas.com/author/", "image": { "@type": "ImageObject", "url": "" }, "sameAs": [ "" ], "knowsAbout": [ "Google Ads", "Performance Max", "AI Max", "Pay-per-click advertising", "Conversion tracking", "Bid management", "Search advertising" ], "worksFor": { "@id": "https://www.groas.com#organization" } }, { "@type": "Organization", "@id": "https://www.groas.com#organization", "name": "groas", "alternateName": "groas.com", "url": "https://www.groas.com", "logo": { "@type": "ImageObject", "url": "https://cdn.prod.website-files.com/6821efca072e48f6f495a47e/6834a970e03a7016560e3515_Logo%20Design%20256x256.png" }, "description": "Dedicated strategists run your account and a proprietary engine trained on $500B in profitable ad spend optimises the execution underneath them", "sameAs": [ "https://www.linkedin.com/company/groas" ], "knowsAbout": [ "Google Ads", "Performance Max", "AI Max for Search", "Pay-per-click advertising", "Autonomous campaign management", "AI advertising agents" ], "areaServed": "Worldwide" }, { "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://www.groas.com" }, { "@type": "ListItem", "position": 2, "name": "Blog", "item": "https://www.groas.com/blog" }, { "@type": "ListItem", "position": 3, "name": "", "item": "https://www.groas.com/category/" }, { "@type": "ListItem", "position": 4, "name": "The SaaS Campaign That Spent 81 Days in Learning", "item": "https://www.groas.com/post/the-campaign-that-never-left-the-learnin" } ] } ] } ```