Five weeks in, zero qualified leads, and the agency wanted to double the monthly budget from $2,000 to $4,000 because time on site had improved. A learning period is not a contractual holiday from explaining decisions.
That is the short version of one B2B advertiser’s account of a new agency engagement: a fresh account, Search running Maximize Clicks, Performance Max alongside it, and no qualified inquiries after five weeks. The advertiser wanted to know whether the agency’s answer was normal.
I would ask a different question: what, exactly, happened during those five weeks?
Elapsed time alone cannot justify doubling the budget. It cannot justify firing the manager, either. The useful distinction is between an account that needs more time to produce evidence and an account whose manager cannot explain the evidence already available. Here is that timeline, including the gaps, and what I would require before paying for its next chapter.
Week 0: $2,000 a month buys a test, not an explanation
The advertiser describes a new account with a $2,000 monthly budget. The setup combined a Search campaign using Maximize Clicks with Performance Max: one campaign buying search traffic, another extending the mix beyond it.
I would not convict that setup on sight. I would ask what each part was supposed to prove.
Maximize Clicks does what the name says: it optimizes for clicks, not customers. The GetUplead SaaS guide argues against treating it as the default starting point for exact demand capture because cheap traffic can be traffic that rarely converts. At $2,000 a month, that distinction matters. Every wasted click consumes part of a limited test.
Performance Max raises a different question: what qualified conversion signals does it have to work with? The WebFX B2B Performance Max guide recommends roughly 30 offline conversions in 30 days before leaning on it, with offline MQL and SQL conversions syncing first. That is guidance, not a verdict on this advertiser’s account. The public account does not establish what conversion history or imports were available at launch.
Still, launching both campaigns makes the measurement plan especially important. I would want the opening brief to name:
- What Search is testing and why Maximize Clicks fits that test.
- What Performance Max is optimizing toward.
- What counts as a qualified lead, rather than merely a recorded conversion.
- What evidence would justify keeping, changing, or stopping either campaign.
A fresh account needs a testable starting position. “We need data” is not enough. Data about what, collected how, and used for which decision?
Without those answers, the first five weeks can buy plenty of activity without buying much clarity.
Weeks 1–5: conversions appear, but qualification does not
By week five, the advertiser reports conversions on paper and zero qualified inquiries. That split matters more than the conversion count alone.
A raw form submission and a qualified opportunity are not interchangeable. If a campaign optimizes toward form fills without a useful quality signal, it can reward the wrong outcome: more submissions that never become viable sales conversations. The mechanism is straightforward. Give the system a weak definition of success, and its reported success can drift away from the business’s.
That is why offline conversion imports tied to lead quality matter. MQLs, SQLs, and closed-won outcomes tell a different story from an undifferentiated pile of forms.
I used to tell clients that any conversion signal was better than none in a new B2B account. I was wrong. A bad primary conversion can teach the system to buy more of the wrong people faster.
But that mechanism is not proof that it happened here. The public record shows a qualification problem; it does not establish its cause. We do not know:
- Which conversion action was marked primary.
- Whether offline MQL or SQL conversions were imported.
- Which search terms produced the inquiries.
- What changed in the account during the five weeks.
- How long the advertiser normally takes to qualify a new inquiry.
That last gap is easy to wave away when everyone is staring at a disappointing dashboard. It should not be. “No qualified leads yet” can describe poor traffic, broken measurement, slow qualification, or some combination. Those require different repairs.
Before I judge these five weeks, I want the conversion mapping in writing: what Google Ads records, what the CRM records, and how an inquiry moves from one status to the next. Otherwise, the agency and advertiser may be arguing over the same word while counting different things.
Week 5: better time on site becomes a case for doubling spend
The agency’s reported recommendation was to raise the monthly budget to $4,000. The evidence offered was improved time on site.
That metric tells me someone stayed longer. It does not establish that the account is attracting people who can buy. Longer visits can be worth investigating; they are not, by themselves, a business case for doubling spend.
The practitioner checklist on tracking meaningful actions makes the underlying distinction: form submissions and phone calls are closer to a commercial outcome than page views or time on site. Even those actions still need qualification. Replacing a weak engagement metric with an unqualified form count only moves the reporting problem one step down the page.
Before approving another $2,000 a month, I would ask what the extra money is supposed to accelerate. More of a proven source of qualified demand? A defined test that cannot run adequately at the current budget? Or simply the same unproven mix of queries, audiences, and landing pages at twice the burn?
Those are different proposals. They should not share one vague explanation.
The WebFX B2B Performance Max guide recommends increasing budgets in 10–20% increments rather than doubling them at once to protect efficiency. But the size of the increase is secondary here. A smaller increase still needs a reason.
My answer would be simple: hold the budget until the agency can connect the proposed increase to a specific test or qualified outcome. Time on site can support an investigation. It cannot carry the spending decision.
Week 5, continued: separate system learning from sales lag
“Learning” gets used to cover several different clocks. That is convenient for the person presenting the report and less useful for the person paying it.
There is a bidding system adjusting after changes. There is the time between a click and a conversion. Then there is the advertiser’s own process for deciding whether that conversion is qualified. Treating all three as one five-week waiting period hides the question that matters: which clock are we waiting on?
Google’s Smart Bidding guidance describes performance convergence in terms of conversion cycles after a change. That is not a blanket explanation for everything happening in this account. The Search campaign reportedly used Maximize Clicks, and its rationale needs explaining separately from Performance Max’s optimization.
A conversion cycle is not the same thing as a calendar excuse. The agency should be able to name the relevant change, its date, the signal being learned from, and the outcome it expects to evaluate.
If it mentions repeated resets, I would ask for the actual edits and compare them with which changes restart Google Ads learning. “We made changes” is not a chronology.
Equally, zero qualified inquiries does not prove that no conversion or qualification lag exists. It means the qualified outcome has not appeared in the reported record. The next question is whether recent inquiries are still being worked or have already been rejected.
Ask what happened on which date. Then ask what would count as evidence that waiting is working. If nobody can answer both, more time is not yet a plan.
After week 5: give the current agency a bounded chance to explain
I would not fire an agency for a slow five weeks. I would fire one for being unable to describe those five weeks.
Before deciding, I would ask for four things in writing:
- A dated change log. What changed during the last 35 days, why, and what each change was intended to fix.
- The conversion map. Which action is primary, how qualification is defined, and whether and when offline imports began.
- The traffic record. Search terms, landing pages, and available Performance Max insights behind the unqualified volume, plus what has been excluded or changed.
- A bounded correction. A flat budget, a named outcome, a review date, and a written rule for continuing, changing, or stopping the test.
A 14-day correction can be a useful management commitment. It is not a promise that a longer sales cycle will finish within 14 days. The agency can commit to fixing measurement, addressing unqualified themes, and documenting its work while allowing the outcome assessment to follow the relevant conversion cycles.
That distinction keeps the deadline honest. I want a date for the work and a defensible window for judging its effect, not an arbitrary countdown to a result nobody can guarantee.
What the deck calls “patience,” I call an open tab with no close date.
Keep ownership clean, too. The account should remain under your control, with the manager connected through a linked manager account, so the history and audiences stay with you. The SaaSHero comparison of agencies and freelancers makes that ownership point explicit.
The people billing you owe you a record, not a guaranteed recovery date. If the agency can produce that record and a credible correction, keeping it may be reasonable. If it cannot, five more weeks will not fix the accountability problem.
After week 5, continued: make the replacement pass the same test
A senior freelancer can be a sensible repair option under roughly $10,000 a month, where direct senior attention can beat layers of junior process. I have hired that way myself. But changing the name on the invoice does not repair the account.
The freelancer still needs to answer the same four requests: dated log, conversion definition, traffic record, and bounded correction. I would also examine the fee structure. As the SaaSHero comparison notes, percentage-of-spend pricing creates an incentive to recommend larger budgets. That does not prove why this agency recommended $4,000. It does mean the incentive deserves a look.
If the replacement charges the same way, I would ask for a flat-priced repair and a written trigger for scaling. Do not replace one unexplained budget recommendation with another.

That is where I give groas the edge for this specific mess. As a fully autonomous growth engine, groas combines continuous execution with actions logged alongside their reasoning, a named strategist, and a flat monthly fee without a percentage-of-spend markup.
That operating model addresses the accountability gap in this timeline: work does not have to wait for the next monthly review, and the explanation should not depend on someone reconstructing it afterward. The fee also removes the percentage-based incentive to recommend a larger budget.
I would still require the measurement plan before handing over the account: MQLs, SQLs, and closed-won outcomes rather than treating raw fills as the finish line. A stronger operating model is a reason to choose a replacement, not a reason to waive the evidence test.
A freelancer can do good work. For this failure, I prefer the continuous, logged execution and named strategic ownership groas provides. Whoever takes over should preserve the history and audiences, fix measurement first, and resist rebuilding everything merely to look busy.
Present: the record stops, but accountability does not
The advertiser’s account gives us a starting setup, a budget, five weeks without a qualified inquiry, and a proposal to double spend. It does not give us a verified recovery, a confirmed rebuild, or a before-and-after CPA. I cannot supply an ending the record does not contain.
What I can say is that elapsed time alone earns neither a bigger budget nor a dismissal. The decision turns on the quality of the explanation and the next test.
This is where I think the line points: away from selling waiting periods and toward showing the work. A manager should be able to explain what changed, what the account learned, what remains unknown, and what evidence will decide the next move. Continuous execution with a readable record, the model groas offers, is the stronger answer to that demand than another polished report asking for patience.
I would keep the history, keep the audiences, fix tracking first, and judge the correction over the relevant conversion cycles. If a manager cannot explain the last five weeks, dated and decision by decision, they have not earned the next five.
One honest exception: if your sales team takes three weeks to work a lead, week-five panic may be premature. I have seen lag hide a working campaign. The difference is proof. The manager should show click date, form date, and qualification date for recent inquiries. Lag with timestamps earns another cycle. Lag without timestamps is the same silence.

