Skip to content
← Articles & field notesArticleCase study15 min readRevised

How one account scaled monthly Google Ads spend from $21K to $336K

A retrospective on contribution economics, intent pockets and account structure; the spend endpoints are one account's experience, not a growth guarantee.

The endpoints measure monthly Google Ads spend over eight months. The contribution comparison is illustrative arithmetic, and the readiness score is an author-defined checklist, not a validated predictor of profitable scale.

We watched a collagen brand go from $21K per month to $336K per month in Google Ads spend over eight months. These are reported ad-spend endpoints for one account, not revenue or a transferable growth promise.

Several operating changes accompanied that increase. They stopped treating Google as a support channel and started running prospecting campaigns with strict ROAS targets.

That's the entire story in one sentence. The rest of this article is about why that's so hard to actually do, what breaks when you try, and the exact architecture that makes it work.

Investigate the Constraint Before Expanding Spend

Inspect the actual share of reported brand and non-brand activity, with the query definitions and attribution window retained. A high brand share describes capture; it does not prove those customers would all have purchased without an ad.

Branded search. Some remarketing. Maybe a Performance Max campaign that's mostly capturing branded queries. That's it.

A branded query indicates familiarity or navigational intent. The incremental effect still depends on the account, competitors and a defensible comparison.

Meanwhile, Meta is doing all the heavy lifting for new customer acquisition. And the brand is convinced Google "doesn't work" for prospecting because they tried it once, set a target ROAS that was too aggressive, watched it underperform for two weeks, and turned it off.

Different accounts can stall for different reasons: demand, contribution, data quality, offer or execution. Use the framework to investigate rather than assume one recurring diagnosis.

The author reports the collagen account beginning at $21K monthly Google Ads spend with Google contributing a small share of attributed revenue. That is the account narrative, not the default condition of every ecommerce account.

The ROAS Trap

Here's the math that changes everything.

These two scenarios are illustrative arithmetic at a 50% gross margin. “Profit after ads” below is contribution after COGS and ads; it excludes other operating, fulfillment, return, tax and retention costs. They are not the collagen brand's measured results.

Scenario A: 10x ROAS at $10K spend

  • Revenue: $100,000
  • Ad spend: $10,000
  • Gross margin at 50%: $50,000
  • Profit after ads: $40,000

Scenario B: 4x ROAS at $80K spend

  • Revenue: $320,000
  • Ad spend: $80,000
  • Gross margin at 50%: $160,000
  • Profit after ads: $80,000

Scenario B generates twice the profit. At what most marketers would call "worse" ROAS.

This is the trap. Efficiency percentages feel good. They look good in reports. But they don't pay for anything.

Under the stated 50% gross-margin assumptions, Scenario B has $40,000 more contribution after COGS and ads. It also spends more and excludes several costs. That model does not establish unrealized profit or available demand for the reader's account.

ROAS describes attributed revenue relative to ad spend. Evaluate contribution after the relevant product, fulfillment, return, fee, retention and operating costs, with the incremental question kept separate. Changing the chosen metric does not itself create profitable demand.

A competitor's spend and reported ROAS cannot establish its profit without the underlying costs, definitions and attribution. Use the actual business contribution model rather than infer a competitor's margin.

Lower ROAS. Twice the contribution after ads.

The article’s two illustrative scenarios, using a 50% gross margin. These are examples, separate from the brand’s reported results.

RevenueContribution after ads
Scenario A · 10× ROAS$10,000 ad spend · 50% gross margin
Revenue: $100,000
Contribution after ads: $40,000
Scenario B · 4× ROAS$80,000 ad spend · 50% gross margin
Revenue: $320,000
Contribution after ads: $80,000

0–320,000 · USD per month · both scenarios on one scale

Source: Published article · this section

The arithmetic behind the comparison
ScenarioROASAd spendAfter ads
Scenario A
10×$10,000$40,000
Scenario B
4×$80,000$80,000

The Four-Pocket Architecture

The four-pocket framework can make campaign roles easier to inspect. It does not predict scalability from a brief look; demand, economics, creative, measurement and execution remain account-specific.

The four pockets:

Separate demand creation from demand capture.

Conceptual relationship from the article. Each pocket keeps its own bidding strategy and measurement.

Prospecting SearchFind new customers through search
Prospecting Shopping and PMaxIntroduce products to cold traffic
Branded SearchCapture brand searches
Branded ShoppingCapture branded product demand

Source: Published article · this section

People searching the brand may be navigating, comparing or returning. Define what the campaign should cover and measure its actual cost and incremental contribution rather than assume a typical ROAS or CPC.

Choose brand coverage and bid controls from current supported settings, competitive conditions and the account's economics. A fixed coverage target is a planning choice, not a universal requirement.

The key insight: this pocket tells you how much demand exists for your brand, but it doesn't grow your brand. If your entire Google Ads strategy lives here, your Google revenue is capped by whatever Meta, email, and word of mouth generate in brand searches.

Define this campaign role, actual query/product eligibility and conversion objective. Expected return, click cost and scale depend on this account's economics and evidence.

Choose the bidding strategy from current supported requirements and actual account signal, delay and economics. No universal fifty-conversion threshold or mandatory initial Maximize Clicks phase is established here.

This is the pocket most brands never build. They think Google is "just for branded." This is where scale lives. This is where that collagen brand found most of its growth.

Pocket 3: Branded Shopping

Branded Shopping is a possible product-discovery or capture role where inventory is eligible. Inspect actual queries, placement, product economics and the business's goals before choosing it.

Pocket 4: Prospecting Shopping and PMax

Choose the campaign mix from the account's actual growth question and contribution limits. A brand/prospecting share does not establish either an appropriate spend level or incremental return.

How They Work Together

Think of it as a funnel. Prospecting pockets (2 and 4) feed the top. Branded pockets (1 and 3) capture the bottom. You need both. Scaling means pushing more money into prospecting while maintaining branded efficiency.

Separate the question being measured. Keep clear brand and prospecting definitions and inspect actual query routing. Their conversion rates vary; a blended ratio cannot answer every acquisition question.

Choose brand controls for the campaign's actual measurement and buying role. For eligible Performance Max campaigns, Google's brand-exclusion guidance distinguishes exclusions from literal negative-keyword matching and limits the control to supported inventory. Inspect actual query routing and coverage rather than apply account-wide negatives blindly or assume complete separation.

Foundation Phase: What to Fix Before Spending Another Dollar

Review measurement, product eligibility, destination quality and operating capacity before adding spend. The checklist helps investigate readiness; it is not a complete explanation of every failure.

The author’s foundation checklist.

The reported initial score against the article’s unvalidated rubric. Its upper band does not establish profitable scalability, measurement quality or permission to act.

Reported initial checklist score
54/100

0–100 · author-defined checklist points · Middle rubric band: 60 · Upper rubric band: 80

Below 60 · lower band60–79 · middle band80–100 · upper band

Source: Published article · this section

Conversion Tracking

Your conversion tracking needs to pass five checks before it's trustworthy:

Your purchase event fires exactly once per transaction. Not on page reload. Not on back-button. Test it: complete a test purchase, reload the confirmation page five times, check your Google Ads conversions. If you see five conversions, you have duplicate counting.

Define the revenue basis and reconcile representative transactions with the business ledger. Document differences from taxes, shipping, discounts, refunds, timing, attribution and currency rather than require an unexplained universal match percentage.

Inspect the primary and secondary conversion actions, counting settings and transaction deduplication for the business outcome. Multiple action labels do not automatically imply duplicate optimization, and one label does not guarantee correct counting.

For eligible enhanced-conversion use, verify the actual consent, data handling and supported configuration. Hashing and an enabled setting do not establish valid permission, working coverage or a guaranteed attribution recovery percentage.

Your attribution window matches your actual sales cycle. A 7-day consideration product on a 1-day click window is missing credit for conversions you actually drove.

Merchant Center Quality

Review Merchant Center eligibility, required product facts and the actual destination experience. Better shipping and return information can help a buyer, but this article does not establish a deterministic store-score-to-auction rule.

Quick wins: add expedited shipping options, display clear return policies on product pages, enable guest checkout, improve mobile page speed. None of this is Google Ads work, but all of it determines whether your Google Ads campaigns can compete.

Customer Lists

First-party data is your competitive advantage. Upload four lists: all purchasers from the last two years, high-value customers (top 20% by LTV), email subscribers who haven't purchased, and engaged website visitors.

Use first-party customer data only within the applicable permission, current targeting capabilities and account configuration. Retain its source, freshness, intended use and actual performance; a list does not guarantee new-customer learning.

The Readiness Score

We score every element 0-10 before scaling. Conversion tracking accuracy, conversion action hygiene, Merchant Center quality, customer list freshness, campaign segmentation, negative keyword coverage, feed attribute completeness, landing page quality, historical conversion volume, reporting clarity.

The author's checklist uses bands of 80–100, 60–79 and below 60 to prioritize unresolved checks. These are unvalidated rubric bands; a high score does not establish profitable scalability, complete measurement or authorization to change an account.

The collagen brand scored 54 on their first assessment. We spent two weeks on foundation work before touching their budget. The work preceded the budget increase; the counterfactual claim of three months of avoided waste is not measured here.

Structure Phase: Building What Supports Scale

With the foundation clean, the collagen brand moved into building campaign structure. This is where most brands rush and fail.

Feed Quality Is Your Keyword List

In Shopping and PMax, you don't bid on keywords. Google matches your products to searches based on your feed. Your product feed is your keyword list.

A product with a generic title like "Collagen Powder" matches generic searches and competes with everyone. A product with an optimized title like "Marine Collagen Peptides Powder - Hydrolyzed Type I and III - Unflavored 300g" matches specific, high-intent searches with less competition.

The original feed anecdote reported an impression-share change after title work. Its auction eligibility, window and concurrent changes are not established here, so it cannot identify title optimization's independent effect.

The front-loading principle: the first 70 characters of your title show in Shopping results. Put your highest-volume keywords there. Everything after character 70 contributes to matching but doesn't display prominently.

Bidding Progression

The click-based progression below is a historical Search example, not a PMax option. Current Google PMax guidance, reviewed November 4, 2026, uses conversion- or value-based Smart Bidding. Choose supported settings for the actual campaign; the example thresholds are not universal eligibility rules.

New campaigns need data before automated bidding can work. The sequence:

Start with Maximize Clicks plus a bid cap. This gives Google freedom to find impressions while capping your cost per click. Run for two to four weeks until you hit 30-50 conversions.

Set a target from actual contribution, strategy behavior, conversion delay and risk. A fixed fraction below earlier ROAS does not establish a profitability floor.

Choose target and budget changes from actual contribution, conversion delay, volume and risk. A percentage step and fixed two-week schedule are planning heuristics, not universal platform rules.

Document the bid change, expected conversion delay and review window. Avoid interpreting an immature sample, but do not assume every change resets a universal clock or guarantees a performance dip.

The Four-Stage Deployment

You don't launch all four pockets at once. Deploy in stages with strict advancement criteria.

Stage 1 (Days 1-14): Bottom-of-funnel only. Branded search, branded shopping, high-intent prospecting search. Hit 20 conversions in 14 days with clean tracking before advancing.

Stage 2 (Days 15-42): Add mid-funnel. 70% bottom-of-funnel, 30% consideration-stage prospecting. Mid-funnel campaigns need to reach ROAS within 20% of your target with less than 25% week-over-week variance for two consecutive weeks before advancing.

Stage 3 planning: add only the channel roles the account can operate and measure. Inspect brand-query routing and actual exposure; a fixed mix or leakage threshold cannot establish incremental prospecting.

Stage 4 (Day 71+): Scale. 45% bottom, 30% mid, 25% top. From here, increase total budget 10-20% weekly while maintaining these ratios.

The collagen retrospective describes an earlier attempt with overlapping branded traffic, three weeks of stronger reported ROAS and a later decline. That account narrative does not establish that the overlap caused the decline or that Google requires this sequence of stages. The staged plan is an operating approach for reviewing risk and measurement before adding work.

The Brand Spillover Effect

Prospecting can introduce a brand that someone searches for later. That is a plausible path to investigate; branded-search correlation alone cannot show which purchases prospecting caused.

The original retrospective proposed reviewing branded impressions two to four weeks after a prospecting increase of 20% or more, and reported 5–15% impression changes in other accounts. Those are author-reported observations and timing heuristics, not a controlled lift estimate or an expected result for another account.

For the collagen brand, the author reported $80K in monthly prospecting Shopping spend and a 25% increase in branded search volume over the following six weeks. The author also associated roughly $18K in monthly branded revenue with that activity. These figures use different windows and do not establish that the associated revenue was incremental or caused by prospecting.

The reported direct attributed ROAS was 3.2×. The author estimated a return closer to 4.5× after assigning additional branded credit. That is an attribution estimate, not observed incremental return. The stated $18K added to 3.2× on $80K spend would yield about 3.425×, not 4.5×; the supplied inputs therefore do not reproduce the larger estimate. Reconcile the spend/revenue windows, additional credit, order identity and attribution rules before using either estimate in a budget decision.

A causal conclusion needs a defensible holdout or other suitable comparison that addresses seasonality, offers, other channels and prior demand. Attribution reconciliation prevents double-counting; it does not by itself supply that counterfactual.

How the Author Estimated Branded Spillover

The author was investigating this possible buying path, rather than demonstrating a proven sequence for every customer:

  1. Prospecting spend increases and more people may see the products.
  2. Some exposed people may register the brand without clicking.
  3. Some may later search for the brand directly.
  4. Branded campaigns may capture those searches; the original narrative cited 10–15× attributed ROAS for that capture activity.
  5. The branded report receives credit even where prospecting may have contributed to earlier awareness.
  6. Prospecting and branded reports can consequently tell different parts of the same buying path.
  7. Comparing their dashboard ROAS alone may suggest shifting budget to capture.
  8. A prospecting reduction can then be investigated alongside later branded-volume changes.
  9. Seasonality, offers, other channels and market conditions remain alternative explanations.

The author reported observing this association across repeated account reviews, using 30–60-day views alongside shorter operating reviews. The collagen account used separate branded and prospecting dashboards and plotted prospecting spend against later branded volume. The reported three-to-four-week association was a reason for further investigation, not proof that every prospecting change caused the later movement.

The retrospective reports a 65% prospecting allocation at $336K in monthly ad spend, or roughly $218K for cold traffic. It reports around 3.4× direct attributed ROAS and an author-estimated 4.8× after additional branded credit. The public account does not provide the reconciled order-level attribution or causal comparison needed to establish that estimate as incremental return. At $218K spend, moving from 3.4× to 4.8× would require about $305,200 in additional revenue credit in the same window; that is an implied arithmetic requirement, not a new observed revenue figure.

Before assigning that credit, align the period, conversion and revenue definitions, deduplicate orders already credited elsewhere and test the causal claim separately. A correlated branded increase cannot on its own justify scaling prospecting.

The Seven-Layer Budget Architecture at Scale

Once the collagen brand reached full operating state, their budget had more structure than simple pocket ratios. Each layer has different economics, different bid approaches, different ROAS expectations.

The Seven-Layer Budget Architecture at Scale · table 1
LayerBudget ShareIllustrative ROAS bandPurpose
Branded Search
10-15%
10-15x
Demand capture, brand defense
Branded Shopping
5-10%
8-12x
Product-level brand capture
Prospecting Search (high intent)
15-20%
4-6x
Core new customer acquisition
Prospecting Shopping/PMax
20-25%
3-5x
Cold product discovery
Competitor Campaigns
10-15%
3-8x
Conquest, incremental
Pre-Sell Pages
10-15%
2-4x blended
Top-of-funnel, remarketing pool
Demand Gen
5-10%
1.5-3x blended
Awareness, assisted conversions

This historical planning sketch mixes campaign roles and destinations: a pre-sell page can receive traffic from several of the listed campaigns. Its ranges are not disjoint budget shares or a forecast. Build a non-overlapping allocation that sums to the actual budget before using it operationally.

The brand didn't launch all seven layers at once. The retrospective describes adding work after earlier reviews: competitor campaigns, more pre-sell destinations and then Demand Gen. Reuse the decision process rather than treating those illustrative return bands as expected results.

Review allocation against actual demand, contribution and risk. A fixed monthly rebalance percentage is a planning choice, not an established algorithm-stability rule.

What Made the Difference

Eight months. $21K to $336K per month. Google went from covering less than 10% of revenue to being the primary acquisition engine.

The "secret" was boring. Fix the foundation. Build the structure. Deploy in stages. Measure what actually matters instead of what looks good in a screenshot.

Three things made the difference:

They investigated tracking before scaling. The retrospective describes two weeks of foundation work to improve the purchase signals available to Smart Bidding. Cleaner signals support better-informed bidding; they do not establish that every prospecting dollar reached the right buyer or produced a profitable purchase.

They reported branded and prospecting separately. Four pockets made the intended roles and attributed results easier to inspect. Query routing, attribution overlap and actual business contribution still needed review; separate reports are not complete causal explanations.

They investigated possible brand spillover. Instead of judging prospecting only by its direct attributed ROAS, they tracked its association with later branded volume. That informed the next investigation and budget discussion; it did not establish incremental return.

None of this required proprietary technology. None of it required a massive team. It required discipline - doing things in the right order, measuring the right metrics, and resisting the urge to skip steps because the math on paper looks good enough.

The brands stuck at $10-30K per month aren't stuck because Google doesn't work. They're stuck because they're using Google to capture demand instead of creating it. Changing that role can expose new demand, but the ceiling still depends on account economics, market size and execution.

Put the framework to work

Review the scaling prerequisites

Give your growth
an operating team.

Google Ads, Meta Ads and Klaviyo Email/SMS. Buying, creative, copy, pages, testing and optimization, handled by the team that built the AI systems.

Typically live within 48 hours. The clock starts when we have account access and the assets we need.

Discuss your growth goalCompare services and fees

Keeping execution in your team?
Book a paid working session or explore licensed systems.