Five years and 7.5x the spend, through a deliberate ROAS compression and back to a best-ever 7.35x.

Scaled 7.5x in spend, ROAS rebuilt to its best - five years of scaling a sleep brand without giving up the return.

A North American sleep brand brought us in to run their Google Ads in 2021. What followed was not a single campaign or a short sprint of fixes. It became a five-year engagement that is still running today.

The arc has two distinct chapters. In the first, we scaled spend aggressively, from the 2021 baseline to roughly 7.5x that figure by 2025, while managing the ROAS compression that comes when a brand pushes hard into thinner demand. In the second, we rebuilt efficiency without retreating on scale, recovering ROAS from its compressed floor to an all-time high of 7.35x in 2025.

The starting point was already strong. In 2021 the account returned 7.24x. The headline of the five years that followed is that we scaled spend roughly 7.5x and still finished at a higher return than we started, 7.35x, the best in the engagement. More volume and a better return at the end is the thesis this story proves year by year.

The path was not linear. ROAS compressed sharply as spend scaled through 2022 and 2023, dropping to a floor of 4.38x. The recovery in 2024 and 2025 took a different kind of work than the early push. Both chapters are here because the compression was not a failure, it was the predictable cost of a deliberate scaling decision, and the recovery is what separates a managed scale from an uncontrolled one.

2026 is an in-progress year and is not compared year on year here. The current pace continues at a strong ROAS of 6.78x on a larger base than any prior year.

Spend scaled roughly 7.5x from 2021 to 2025, and ROAS still finished higher than it started, at a best-ever 7.35x.

ROAS recovered +68% from its 2023 floor of 4.38x to the 2025 peak - a managed scale, not an uncontrolled one.

A strong 7.24x was the starting point - the mandate was to scale spend many times over without letting that return collapse.

When we took this account on in 2021, it was already generating a ROAS of 7.24x. That is not a broken account. It is a strong return for a direct-to-consumer brand in a competitive category. The problem was never the account itself. It was what the brand needed it to do next.

The ask was volume. The brand had ambitions that required paid search to punch well above its current weight: more conversions, more revenue, and a spend base that would need to grow many times over to match. The question was whether efficiency could survive that scale, or whether growth would simply buy volume at the cost of return.

That tension is the oldest problem in paid search. Scale and efficiency pull in opposite directions. Every dollar added to a mature account tends to chase weaker demand: lower-intent queries, less competitive placements, broader audiences. A brand that doubles spend rarely doubles revenue. ROAS almost always compresses, and the only question is by how much and for how long.

The mandate we accepted was to grow spend roughly 7.5x over the following years while defending, and ideally improving, the 7.24x we started from. That is a structurally difficult objective. It requires a long-term architecture, not a short-term campaign, deliberate tradeoffs between efficiency and volume at each stage, and the willingness to accept temporary compression when scale demands it, as long as the account has a credible path back to stronger returns.

The five years that followed tested every part of that thesis. There were years where efficiency fell sharply as spend accelerated, and years where the account recovered more than what was lost. The record across the full arc is what the data shows: a starting ROAS of 7.24x in 2021 and a best-ever 7.35x in 2025, with roughly 7.5x the spend behind it.

Starting ROAS of 7.24x in 2021 - the strong return the account had to defend at scale.

A roughly 7.5x spend mandate: the structural challenge that defined the engagement.

We took the account on in 2021 at a strong 7.24x - the baseline every later year would answer to.

2021 is where our tenure begins. We took on an account that was already converting well, and it closed the year at a ROAS of 7.24x. That is the reference point, what we index as 1x spend, the baseline every subsequent year is measured against.

A 7.24x return on a direct-to-consumer mattress account is a strong starting position, not a struggling one. The job was never to rescue this account. It was to scale it hard, many times over in spend, without surrendering the efficiency it already had.

The first work on taking over was making the account legible at the scale that was coming: clean, attributable tracking, a campaign structure split cleanly by brand and non-brand intent, and bidding foundations that could carry far more budget than the account had run before. Knowing exactly what the account was doing and why is what made the aggressive scaling of the next two years a controlled decision rather than a gamble.

The question 2021 set up was simple. The account converted well at this budget. How much further could spend go before efficiency gave way, and could it be rebuilt if it did? The next four years answered both.

A strong 7.24x ROAS baseline in 2021 - the reference point for the scale-up.

Tracking and structure set for a budget many times the size the account had run.

Spend jumped 2.3x year on year - and ROAS paid the price.

By 2022 the account had a clean baseline year behind it, seasoned audiences, and a bidding engine that understood the brand's buyer. That foundation created an opening: how far could spend scale before efficiency broke down in a way that could not be recovered?

The answer came from pushing spend to 2.3x the 2021 base, a 2.3x jump in a single year. That is not steady growth. It is a step-change in the account's ambition.

Volume responded. Clicks grew +91% year on year. Conversions grew +76%. Revenue grew +49%. The machine was producing at a scale the brand had never operated at before.

But the efficiency math shifted. When spend scales faster than revenue, ROAS compresses, and it did. ROAS fell from 7.24x to 4.66x, a -36% drop. Value per conversion fell -15%, meaning the extra buyers brought in at this scale were transacting at lower average order values than the core base.

This is the basic tension in aggressive scaling: the bidding algorithms, pushed to find more volume, reach beyond the high-intent core and pull in audiences with lower purchase values. Spend grows, revenue grows, but the ratio between them narrows. That is not a failure of execution. It's the expected shape of pushing hard into new demand.

The decision to absorb a -36% ROAS compression in one year was a deliberate trade. The brand chose volume over efficiency at this stage. What the account gained was a dramatically larger footprint in the auction, a far larger conversion dataset to learn from, and a new floor to optimize from.

Clicks up +91%, conversions up +76% - the largest single-year volume expansion in the account's history.

ROAS compressed to 4.66x (-36%) - the cost of a deliberate 2.3x spend jump, paid in efficiency.

Spend reached 3.85x the 2021 base - and the volume thesis held underneath the numbers.

By 2023, the account had been running at meaningful scale for two years. The pattern from 2022 continued: spend climbed again, this time to 3.85x the 2021 baseline, a +66% increase over the prior year. ROAS slipped a further -6% to 4.38x. On paper, that looks like a second year of declining efficiency. Read alongside the volume figures, the picture is different.

Conversions grew +68% year over year. Revenue value grew +56%. Clicks expanded +60%. These are not the numbers of a campaign in trouble. They are the numbers of one absorbing a big spend increase and still delivering. The ROAS compression is the cost of buying a harder audience, not a collapse in creative or account structure.

This is the efficiency trough in the arc. Spend had now grown faster than ROAS for two years running, and value-per-conversion had declined -7%, a sign that the marginal customer at this scale is worth a little less than the core buyer. That is just what pushing volume looks like in any maturing channel, and the right response is not to retreat.

The read at this point in the engagement was straightforward: the volume thesis was still working. Each percentage of spend added was returning conversions at a rate that justified the outlay. A ROAS of 4.38x at 3.85x the baseline spend is a very different thing from a 4.38x on a flat budget. The absolute return to the brand was multiples larger than at the baseline.

Holding a floor at this scale is not passive. Campaign structure, bidding signals, and feed quality all need active management as volume increases, because the algorithms are now working across a much wider auction footprint than they were at baseline. The account work in this period was about keeping signal quality high, so the efficiency recovery that showed up the following year had a clean foundation to build from.

Conversions grew +68% and value grew +56% even as spend scaled to 3.85x the 2021 base - the volume thesis held through the trough.

ROAS compressed a further -6% to 4.38x, the lowest point in the engagement - setting the floor before the +38% recovery that followed in 2024.

Spend grew again, and this time ROAS grew with it - a combination the prior two years had not delivered.

2024 broke the pattern that had defined 2022 and 2023. In both of those years, scaling spend had come at a cost to ROAS. In 2024, spend grew +39% and ROAS grew +38% alongside it, to 6.04x, up from the floor of 4.38x the year before. That is not a small correction. It is the strongest single-year ROAS recovery the account had seen.

What made 2024 different was not just the ROAS recovery. It was that conversion value grew far faster than conversions. Value grew +92% year on year while conversion volume grew +60%. The gap between those two figures is the tell: value-per-conversion rose +20%. Buyers were spending more per order, not just buying more often.

This matters because the prior two years had shown the opposite. In 2022 and 2023, value-per-conversion had declined -15% and -7% respectively, a sign that scale was being bought partly by pulling in lower-value transactions. In 2024, both moved the right way at once: more buyers, higher baskets, more efficient spend.

Click growth slowed sharply, to +16%, from +60% the year before. The account was not just buying more traffic. It was converting a larger share of that traffic into higher-value purchases. The efficiency gain came from better return on the traffic it had, not from flooding the funnel.

By the end of 2024, total spend stood at roughly 5.4x the 2021 base, and the account was returning 6.04x on every dollar of it. The account had not cleared 6x since 2021 itself, when spend was a fraction of the 2024 level. Delivering 6.04x at more than five times the starting scale is a much harder result than delivering it at the baseline budget.

• ROAS +38% to 6.04x - strongest recovery year in the account
• Value-per-conversion +20% - basket quality and volume inflected together
• Spend 5.4x the 2021 base, efficiency recovering hard

ROAS recovered +38% to 6.04x while spend grew +39% - scale and efficiency moving together for the first time since 2021.

Conversion value grew +92% as value-per-conversion rose +20% - the account was winning on basket quality, not just volume.

For the first time, scale and efficiency reached the top of the range together.

2025 was the year the tension between scale and efficiency finally broke our way, and not by pulling back. We pushed through it. Spend reached roughly 7.5x the 2021 baseline, a +40% increase over 2024 alone, while ROAS climbed +22% to 7.35x, the best return in our five years on the account.

That combination is unusual. Scaling spend +40% in a single year typically puts downward pressure on ROAS as campaigns reach further into less-qualified demand. In 2022, a spend jump of 2.3x in one year compressed ROAS by -36%. In 2023, continued scaling pushed ROAS down a further -6%. The 2024 recovery, +38% ROAS on +39% spend, showed the account structure had been rebuilt to handle volume. 2025 confirmed it.

Conversion volume grew +74% year over year, and total conversion value grew +71%. The near-parity between those two figures means the average order value held essentially flat, down only -2% per conversion, a steady sign that the extra volume was not coming at the cost of ticket size.

Clicks grew +60%, matching the conversion growth rate closely. That alignment between click volume and conversion volume points to steady conversion rates across a much larger traffic base: the funnel held its shape as it scaled.

• ROAS 7.35x - best in the engagement, +22% over 2024
• Spend at ~7.5x the 2021 baseline, +40% in a single year
• Conversions +74%, conversion value +71% - scale without ticket-size erosion

The cumulative picture across five years is what gives 2025 its weight. This brand came to us returning 7.24x in 2021. We scaled it hard, absorbed two years of intentional compression as spend multiplied, rebuilt efficiency in 2024, and now, at roughly 7.5x the 2021 spend level, it is returning more per dollar than at any point in the engagement. That is the outcome the 2022 and 2023 scaling was building toward.

ROAS reached 7.35x in 2025 - the best in the engagement, achieved at roughly 7.5x the 2021 spend level.

Conversion volume grew +74% while value per conversion held within -2% - scale without quality erosion.

Half a year in, the brand is running at a 6.78x ROAS pace with the full year still ahead.

2026 is an incomplete year. The figures below reflect only the months elapsed so far, and comparing a partial year to a full calendar year would misrepresent the trajectory. The right frame is pace, and at 6.78x the brand is running well within the efficiency range it set in 2024 and 2025.

A partial year naturally carries timing noise. Spend cadence, seasonal demand windows, and promotional calendars do not spread evenly across twelve months, so a mid-year ROAS read reflects which months have run, not a shift in the account structure or bidding underneath.

The position the account holds entering 2026 is the strongest it has been across this engagement. The efficiency recovery of 2024, which brought ROAS up +38% year-on-year, and the 2025 peak of 7.35x at roughly 7.5x the 2021 baseline spend, were not one-year flukes. They came from changes in how the account is managed at scale.

The work underway in 2026 continues on that foundation: holding the efficiency gains made during 2024 and 2025 while sustaining the spend levels that make those ROAS figures meaningful in absolute terms.

6.78x ROAS pace through the first half of 2026, running on the strongest structural base of the engagement

Spend up 7.5x, ROAS rebuilt to its best: the arc that got there.

Five years in, the numbers tell a clear story: spend scaled roughly 7.5x from the 2021 baseline to 2025, and ROAS finished higher than it started, at a best-ever 7.35x, up from 7.24x in 2021. Most accounts see ROAS compress as budgets grow this much. This one did not, across the full arc.

The arc was not linear. It moved in three phases: a strong baseline, a deliberate scale-and-compress period, and a recovery that reached the best ROAS of the engagement.

2021 set the baseline. The account came to us returning 7.24x, and we indexed every later year against that spend level, what we call 1x.

2022 opened the aggressive scaling phase. Spend grew 2.3x in a single year, to 2.3x the 2021 base. Conversions rose +76% and clicks grew +91%. ROAS compressed to 4.66x, a -36% drop, expected when volume is pushed this hard, but the absolute conversion growth was substantial.

2023 held the floor. Spend grew another +66% to 3.85x the 2021 base. Conversions grew +68%, value grew +56%, and ROAS settled at 4.38x, a modest -6% drift. At nearly four times the baseline spend, holding a ROAS above 4x was a functional floor, not a collapse.

2024 was the recovery year. Spend grew +39% to 5.4x the base, but the efficiency story flipped: ROAS jumped +38% to 6.04x, conversion value grew +92%, and value-per-conversion rose +20%, the only year in the record where that metric improved clearly. Clicks grew just +16% while conversions grew +60%, pointing to a real conversion-rate gain rather than volume alone.

2025 reached the peak. Spend grew +40% to roughly 7.5x the 2021 base. Conversions grew +74%, value grew +71%, and ROAS hit 7.35x, a +22% gain and the best in the engagement. The return surpassed the 2021 baseline while running at many times the spend.

2026 is in progress. The current pace shows a ROAS of 6.78x, within the strong band established since the 2024 recovery, on a spend base that has kept growing. A partial year does not produce a meaningful year-on-year figure, but the level is consistent with the account's recent operating range.

Spend scaled roughly 7.5x from 2021 to 2025 while ROAS finished higher than it started, at a best-ever 7.35x.

ROAS recovered +68% from its 2023 floor of 4.38x to the 2025 peak.

Five years of proof that scale and efficiency are not opposites. They just need patience.

The arc from 7.24x to a best-ever 7.35x ROAS across a 7.5x spend expansion is not a straight line. It is a compression-and-recovery curve, and understanding that shape is the central takeaway of this engagement.

The first phase of aggressive scaling, 2022 and 2023, compressed ROAS from 7.24x to a floor of 4.38x. That compression was the cost of entering a new spend tier. When volume grows 2.3x in a single year, the bidding absorbs a wider, less-qualified conversion pool before it recalibrates. Defending efficiency through a step-change like that is not possible in the short run. The right move is to hold the position, not retreat.

The recovery in 2024 validated that posture. ROAS climbed +38% year on year, the largest single-year efficiency gain in the engagement, while spend kept growing at +39%. Value-per-conversion rose +20%, the only year in the record where that metric improved meaningfully while volume was still expanding. Given enough time and data at the new spend level, the algorithms found the signal.

2025 extended the recovery to a peak: 7.35x ROAS at roughly 7.5x the 2021 spend, a return higher than where we started. The common assumption is that scaling spend erodes returns. This engagement shows it does, temporarily, and then the system adapts.

• Compression at scale is normal, not a signal to pull back - the floor held, and the recovery arrived
• Efficiency recovery takes a long horizon: two full years between peak compression and peak ROAS
• Value-per-conversion is the leading indicator - it turned positive before overall ROAS peaked

A shorter-horizon client or agency would have read 2022 and 2023 as a campaign failure and cut spend. That would have locked in the compression and abandoned the upside the scale itself was building toward. The willingness to stay through the compression, with spend growing +66% in 2023 even as ROAS slipped a further -6%, is what made 2024 and 2025 possible.

The engagement also shows the relationship between spend and conversion volume is not fixed. Clicks grew +91% in 2022 while conversions grew only +76%: new reach is less efficient at first. By 2024 the ratio had inverted: clicks grew +16% while conversions grew +60%. The same budget was converting at a fundamentally different rate once the system had learned the new scale. That full-stack discipline, the analytics and the media run as one system, is exactly what we build for ecommerce brands.

2026 is a year in progress. The pace to date reflects a mature account operating at full scale, the foundation built across five years is what makes that pace possible.

Spend scaled 7.5x while ROAS finished at a best-ever 7.35x, higher than the 2021 baseline.

The +38% ROAS recovery in 2024 proved the compression phase was temporary, not structural.