The Modular Bottle
Google Ads eCommerce case study: How an account started from scratch reached 3.58x, against a 2x target
3.58x
ROAS across the whole period
179% of the 2.0x target, calculated from the client's margin
99.7%
Of the budget, above the 2x threshold
the remaining 0.3% was stopped
$422,985
Attributed revenue
from 4,809 orders, at an $87.96 average
2.74x
ROAS without brand campaigns
the figure that measures new demand
This case study was produced by the performance marketing agency BigConvert, based on data managed for our partner The Modular Bottle. The data was checked and validated by The Modular Bottle team.
The Google Ads account was ours from day one. There was no takeover, no history and no campaign that had run before, so no baseline to optimize against. There was a single benchmark, given by the client: every dollar spent has to bring back at least two.
The decision that looks wrong in a table
The best ratio in the account got the smallest share of the budget. The channel with the weakest ratio among the big ones got almost two thirds.
If you chased the ratio
13.62x
Search
The best ratio in the account. An account built only around it would have looked spectacular and stayed small.
7.7% of the budget
Because we chased the threshold
2.73x
Performance Max
Below the account average, but above the threshold - and the only channel that could take on volume without dropping below it.
60.9% of the budget
A threshold isn't optimized the same way as an average. You can't buy more than people are searching for, however good the ratio is.
Day zero
What we didn't know on day one
In an account you take over, the first weeks are a reading. In an account started from scratch they are a bet, and the only thing that keeps the bet in check is a threshold set in advance.
Zero campaigns had run before us, so there was no baseline to optimize against. We didn't know which product sells, at what price it converts, how much a click costs in this category in the US market or how long it takes for an algorithm to gather enough data to stabilize.
The only benchmark came from the client, and it was a single number: 2.0x. Every dollar spent has to bring back at least two.
The threshold was 2x. Not because it's ambitious, but because it's enough.
The client sells high-margin products, and at that kind of margin 2x covers the cost of goods and the cost of advertising and leaves room for profit. It is the exact opposite of a thin-margin store, which needs 8x or 10x just to avoid losing money. The right number for a target isn't compared across accounts, it's calculated from each one's margin.
This case study can't prove the margin and doesn't publish it. It explains why the target was 2x, nothing more.
The decision
Budget goes where the threshold holds, not where the ratio is highest
Search was delivering 13.62x, but on low volume: search demand for a new brand is limited. Performance Max was delivering 2.73x, but it could absorb budget without dropping below the threshold. In an account run on a threshold, the second property is worth more than the first.
Each campaign type against the 2x threshold
Google · June 26, 2025 - August 13, 2026 · a single period, no year-over-year comparison
That's what the account delivered across the whole period, against the 2.0x target set at the start - i.e. 179% of the benchmark we were given. The target has since been raised to 3.0x and 3.5x on the main campaigns, which is the only confirmation that matters: the threshold from the start is no longer the threshold now.
-
Concentration
60.9%
Of the budget, in the channel that could take on volume
Performance Max ended the period at 2.73x, below the account average and above the threshold, with 46.4% of total revenue. A channel that holds up at volume is worth more than a ratio that can't be scaled.
-
Discipline
0.3%
Of the budget went to campaigns that never reached the threshold
Two campaigns never got to 2x and were stopped. The remaining 99.7% ended the period at or above the threshold. It is also the only cost of learning we can measure.
The result
The numbers, over a single time window
There is no previous period to compare with, because there was no account before. The table below contains absolute values, not changes: no growth percentage appears, or can appear, in this case study.
Everything that follows comes from a single campaign-level export from Google Ads. There is no other channel in this account, so no number on the page mixes platforms.
| Campaign type | ROAS | Budget share | Attributed revenue | Revenue share |
|---|---|---|---|---|
| Performance Max | 2.73x | 60.9% | $196,196 | 46.4% |
| Demand Gen | 2.52x | 17.8% | $53,054 | 12.5% |
| Shopping | 3.11x | 13.5% | $49,569 | 11.7% |
| Search | 13.62x | 7.7% | $123,707 | 29.2% |
| Display | 2.41x | 0.2% | $459 | 0.1% |
Google Ads · account screenshot
The brand
How much is demand we created and how much is demand we captured
In a new account, brand is the easiest part and the most misleading. A single brand campaign produced 29.2% of all revenue, from 7.5% of the budget. People who search for a store's name have already decided somewhere else; the campaign's job is not to lose them.
How much budget each part gets and how much revenue it produces
The bars are on the same scale: a revenue bar longer than the budget bar means captured demand, not created demand
With brand · 14.04x
Almost four times more revenue than budget. It's the ratio that lifts the account average and that, taken alone, would make any account look exceptional.
Without brand · 2.74x
This is where almost all the budget sits, and this is what measures new demand. It stays above the 2x threshold given at the start.
The difference between 3.58x and 2.74x is exactly the brand's contribution to the account average. The figure we consider relevant to our work is the second one, not the first.
In all honesty
What we don't take credit for
An account with no history has fewer things to prove than it seems, precisely because it has nothing to compare itself with.
-
What we can't show
Growth
There is no previous period
All figures are absolute values over a single time window. Any growth percentage would be made up.
-
Would have come anyway
29.2%
Of revenue came from brand
A single campaign, on 7.5% of the budget, at 14.04x. Without it, the account runs at 2.74x - that is the honest number.
-
What the numbers are not
Revenue, not profit
The account runs on classic ROAS
All amounts are revenue reported by the platform. The high margin explains the 2x threshold, but it isn't measured anywhere.
See the full report
The totals for the period analyzed, the brand versus non-brand breakdown and all the methodology notes.
Google Ads · totals for the period analyzed
| Metric | Value |
|---|---|
| Attributed revenue | $422,985 |
| ROAS | 3.58x |
| Target at the start of the collaboration | 2.00x |
| Orders | 4,809 |
| Average order value | $87.96 |
| Clicks | 144,584 |
| Impressions | 15,624,003 |
| Conversion rate, relative to clicks | 3.33% |
Brand versus non-brand
| Segment | ROAS | Budget share | Revenue share |
|---|---|---|---|
| Brand campaign | 14.04x | 7.5% | 29.2% |
| Non-brand | 2.74x | 92.5% | 70.8% |
Methodology and data notes
- Period analyzed: June 26, 2025 - August 13, 2026, as per the export header.
- No year over year: we have managed the account since it was created. There is no previous period and no year-over-year comparison. All figures are absolute values over this single window, and this case study does not and cannot contain growth percentages.
- Source: campaign-level export from Google Ads. Totals are calculated exclusively from campaign rows, after removing 7 Total rows. The campaign sums match the Total: Account row on budget, revenue, orders, clicks and impressions.
- Currency and market: the US dollar. The main market is the United States, which accounts for roughly two thirds of the budget, with the rest split between campaigns with no market tag and a small share in other markets.
- Revenue: the conversion value column attributed at conversion date, the one Google also uses for the value-per-cost ratio shown in the account, 3.59. The export also contains an all-conversions value column, roughly eight times higher because it includes secondary conversion actions; it isn't used anywhere.
- ROAS, not POAS: the account runs on classic ROAS, so conversion value is sales value and all amounts are revenue reported by the platform. The export also contains two profit columns, but they are filled in for fewer than a third of the campaigns and are not used. No claim is made about margin or profit.
- The threshold: the 2x target from the start of the collaboration and the products' high margin are information received from the client. They explain the threshold, but are not measured in this document.
- Conversion rate: calculated relative to clicks. The platform interface shows a rate calculated on interactions, which also includes click-less formats, so the two figures don't match.
- Cost of learning: in a new account, the first months are more expensive than the last, but a campaign-level export doesn't show that. The only measurable part is the budget put into campaigns that never reached the threshold: 0.3% of the total.
- What we don't claim: the choice of market and products is the client's decision. What gets sold and at what price is outside this account.
- Limitation: the results reflect one specific account, category and period. They are not a promise of performance.
The system
What can be repeated
An account started from scratch doesn't need more ideas, it needs a threshold and the patience not to move it. The number doesn't repeat; the order does.
-
Phase 01
The threshold is calculated from margin
It isn't borrowed from another account
A high-margin store can work profitably at 2x, a thin-margin one can't.
-
Phase 02
Volume goes where the threshold holds
The channel with the best ratio is usually also the smallest
Budget goes where there is still demand to buy without dropping below the threshold.
-
Phase 03
The threshold is raised once it holds
The benchmark at the start doesn't stay the final benchmark
Here it went up from 2.0x to 3.0x and 3.5x on the main campaigns.
A good ROAS isn't a big number, it's a number above your threshold.
