Google Ads Benchmarks 2026 — Not From Platform Attribution

Updated 22 October 2026 by Gabriele Franco, founder of Cassandra

Google Ads Benchmarks 2026: What $383M in Spend Actually Returns

Key Findings: Google Ads ROI Benchmarks by Campaign Type and Vertical

Platform metrics such as click-through rate, cost per click, conversion rate and platform-attributed ROAS describe what the ad platform recorded and credited. They do not say how much of that revenue the spend caused.

This report is different. We analyzed 253 Marketing Mix Models across 59 advertisers — covering $383M in total media spend — to measure the actual incremental ROI of each Google Ads campaign type, isolated from cross-channel halo effects, seasonality, and attribution overlap.

According to this analysis, the median incremental ROI for Google Search Non-Brand campaigns is 5.21x, while Performance Max delivers 4.64x and Search Brand sits at 4.14x. These numbers measure what the spend caused, not what the platform attributed to it, so they differ from platform-reported ROAS: usually lower, but not for every advertiser (see below).

Key findings:

• Search Non-Brand has a median ROI of 5.21x

• Search Brand delivers a median incremental ROI of 4.14x across all sectors

• Performance Max has a median of 4.64x and the narrowest 95% interval around its median (3.11x to 6.41x) of the three formats with the most models, so its median is the most precisely estimated of the three.

• Video / YouTube has a median ROI of 2.70x — competitive with Meta and TikTok Performance while reaching entirely different inventory. On 14 models and $3.6M of spend, this points to an underpriced channel, but the sample is small.

• Demand Gen / Discovery shows a median of 2.22x with a confidence interval that requires scrutiny. Most advertisers should validate measurement methodology closely

• Vertical matters more than format: Media & Entertainment advertisers see a median Search Brand ROI of 10.92x. Retail & Ecommerce sees 3.19x on the same campaign type

Why Platform-Reported ROAS Is Not Incremental ROI

Platform benchmarks for click-through rate, conversion rate and cost per click are easy to find, and we summarize one public source below. They are useful for judging ad and landing-page efficiency. On their own, they cannot tell you where to move budget.

Here's why. Platform-reported ROAS is not incremental ROAS. When a platform report shows 12x ROAS for a Search campaign, it means that users who clicked on your ad went on to generate 12x your spend in attributed revenue. Attributed revenue alone does not show how much of it would have happened anyway: organic search intercepted by your own brand keyword, or retargeting of a user who would have converted in three days regardless.

The gap between platform-reported ROAS and incremental ROI can be large, and it is not a fixed factor. Across 449 week-matched channel-model pairs from 87 models and 22 advertisers, the middle half of advertisers had an incremental return between 0.27 and 1.23 times what Google reported. The platform figure was higher for 72% of advertisers on Google, and for some advertisers the gap ran the other way (full comparison of MMM and multi-touch attribution). Where the platform figure is higher, you may be paying for credit on revenue you were always going to get.

Marketing Mix Models, by contrast, measure incremental contribution. They isolate the marginal revenue generated by a dollar of spend after accounting for baseline, seasonality, organic trends, and the simultaneous effects of every other channel. This is not a perfect measurement — no measurement is — but it is structurally closer to the question that matters for budget decisions: what would I lose if I cut this channel?

That is the question these benchmarks answer.

To run the same check on your own account, you can compare platform ROAS with real incremental return in Cassandra, campaign type by campaign type.

Google Ads CPC, CTR and conversion rate benchmarks

This report measures incremental ROI. For the platform metrics, the most recent public source we found is LocaliQ's 2026 Search Advertising Benchmarks (produced with WordStream, last updated 1 June 2026), built from thousands of LocaliQ and WordStream customer campaigns on Google Ads and Microsoft Ads. Across all industries, it reports these 2026 averages for search advertising:

• Cost per click (CPC): $5.42

• Click-through rate (CTR): 6.64%

• Conversion rate: 8.18% (LocaliQ counts leads divided by clicks)

• Cost per lead: $66.69

For the LocaliQ industries closest to the verticals in this report:

Source: LocaliQ, 2026 Search Advertising Benchmarks, last updated 1 June 2026, figures as published. LocaliQ's industry categories do not map exactly to the verticals in this report.

These metrics tell you how efficiently ads turn impressions into clicks and clicks into leads. They do not tell you how many of those leads would have arrived without the ad. A Beauty & Personal Care advertiser can beat the 10.35% conversion rate and still see the modest incremental return we measured for Beauty Non-Brand Search (1.70x median) if the ads mostly convert people who were already going to buy. Use the platform metrics to improve ads and landing pages, and incremental ROI to decide where budget goes.

Methodology: 253 MMM Models, 59 Advertisers, $383M in Spend

Dataset: 253 revenue-outcome MMM models from 59 unique advertisers, spanning 2023–2025. All spend and revenue values converted to USD at time-of-model exchange rates.

Campaign classification: Channel variable names were parsed and classified into seven campaign types — Search Brand, Search Non-Brand / Generic, Performance Max, Shopping, Display / GDN, Demand Gen / Discovery, and Video / YouTube — using consistent logic across all naming conventions. Each record represents one channel from one model run.

Vertical classification: Industry labels were backfilled for all records using marketer-facing vertical categories matched to client business models. Sporting sector excluded from the full dataset.

Statistics reported: For each campaign type, we report the P10–P75 ROI range, median ROI, and a 95% bootstrap confidence interval on the median (3,000 resamples). That interval shows how precisely the median is estimated; it is not the range of results individual advertisers see, which the P10 to P75 range describes. Records flagged as n<5 should be treated as directional only.

Google Ads Benchmarks by Campaign Type: Search, PMax, Shopping, Video, Display

Search Ads ROI: Brand vs Non-Brand Benchmarks

Search Brand and Search Non-Brand represent the two largest spend categories in our dataset — $87M and $95M respectively. Search captures demand that already exists. You are not creating intent; you are intercepting it.

What is worth noting is that:

• Search Non-Brand delivers higher median ROI (5.21x vs 4.14x) because it captures category-level intent across a broader set of queries. Users searching for "houses in Stockholm" have high purchase intent but have not yet chosen a solution — the incremental value of appearing at that moment is significant.

• Search Brand has a narrower spend ceiling — you cannot always spend more on your own brand name, while non-brand keywords offer near-unlimited scale (by increasing budget and keywords). This makes Non-Brand the higher-ROI and higher-scale format.

• The 95% interval around the median is wider for Search Non-Brand: more variance in keyword quality, landing page relevance, and competitive auction dynamics create a wider range of outcomes.

• Search Brand ROI reflects defensive value: we read part of the 4.14x median as revenue that competitors bidding on your brand keyword would otherwise capture.

• The 95% interval around the Search Brand median is 1.95x to 7.78x. That is the range where the true median very likely sits, not the range covering 95% of advertisers: individual results spread wider, with P10 at 1.31x. Total spend is capped by brand search volume, and the ROI depends heavily on how much your competitors bid on your brand keywords.

Performance Max Benchmarks: 4.64x Median, Precisely Estimated

Performance Max automates creative, bidding, audience, and placement across Google inventory in a single campaign. Here is what the incremental data shows.

At 4.64x median ROI, Performance Max performs well. It ranks second among Google campaign types, behind Search Non-Brand (5.21x) and ahead of Search Brand (4.14x). The 95% interval around its median is 3.11x to 6.41x: the true median very likely sits in that range. It does not mean most advertisers get returns within it.

What Performance Max does offer is a precise middle estimate. Its interval around the median is tighter than those for Display/GDN and Non-Brand Search, so 4.64x is a more reliable planning number than their medians. This makes it a reasonable default for advertisers who want predictable performance without the complexity of managing separate search, shopping, and display campaigns.

The concern is opacity. Our data includes $66M in PMax spend, and in most cases the advertisers themselves could not decompose what portion was search, shopping, display, or YouTube. You are buying a bundle and trusting the automated bidding to optimize within it.

Performance Max at 4.64x median is a reasonable baseline expectation. If your PMax ROAS is significantly higher than this, verify whether your measurement methodology is capturing true incrementality or platform-attributed conversions. The two often diverge substantially on automation-heavy formats.

Want to know your PMax incremental ROI? A calibrated Marketing Mix Model isolates PMax's true contribution from the bundle. See how it works.

YouTube Ads ROI: Underpriced in This Dataset, on a Small Sample

This is the finding that would change many media plans, if it holds on a larger sample.

Video / YouTube has a median incremental ROI of 2.70x. That makes it competitive with Meta and TikTok Performance advertising while reaching entirely different inventory, different audiences at different stages of attention, and different creative contexts. In this dataset, advertisers with little or no YouTube spend may be leaving low-cost reach on the table.

The structural case for Video is upper-funnel economics: reach bought at a low cost per impression, though this dataset does not measure CPMs. You are reaching users earlier in the decision journey: before the competitor's brand keyword ad, before the shopping comparison, before the price negotiation. Investment in Video can build brand equity that later shows up in Search Brand ROI. Our hypothesis is that the brands seeing 10x+ on Search Brand in this dataset built the brand before they captured the demand; the models do not test that link directly.

At 2.70x, Video's median incremental ROI is lower than Search — that is expected and correct for an upper-funnel channel. The relevant comparison is not "is this better than Search Brand?" It is "what is this reach (expansion) worth relative to its cost?" On this dataset's numbers, the answer is: more than its current share of spend suggests.

If YouTube is underinvested relative to Search, its auction may also be less saturated, and that kind of advantage does not last forever.

Opportunity: The 95% interval around the Video/YouTube median is 1.50x to 4.20x, so the true median very likely sits in that range. The spread across advertisers is wider: P10 is 0.50x and P75 is 4.50x, so a quarter of the Video models returned 4.50x or more, close to Performance Max's median of 4.64x. We expect those to be advertisers with strong creative, clear brand positioning, and measurement frameworks that account for delayed brand-to-purchase effects; the dataset is too small to confirm it. The difference between the median and the top quarter is often the difference between treating YouTube as a direct-response channel and treating it as low-cost reach at scale.

Because of the lower spend volume on youtube compared to other channels: 3.6M vs +30M, take this assessment with a grain of salt.

Video and Demand Gen belong in your media plan as upper-funnel investments. Measure them using incrementality measurements against a financial metric. Leaving YouTube out because platform ROAS looks weak is a reactive choice. If Video builds brand equity as we expect, the brands investing there now should see it in their Search Brand ROI over time.

Google Ads Benchmarks by Industry: Why Vertical Matters More Than Format

The aggregate benchmarks matter. But they obscure the most important finding in this dataset: where you operate determines what Google can deliver for you.

Fashion & Luxury vs Retail & Ecommerce: Two Different Stories

Fashion & Luxury and Retail & Ecommerce are both retail-adjacent — but they play differently on Google.

Fashion & Luxury sees exceptional Search ROI — brand (5.19x) and non-brand (8.07x) — because these advertisers are competing on desirability, not price. When a user searches for a specific luxury brand, the intent is high and the competitive alternatives are limited. Non-brand search in Fashion & Luxury captures aspirational queries that convert at high margins.

Retail & Ecommerce tells a different story. Search Brand sits at 3.19x — below the all-vertical median — because retail brands frequently have high organic presence. Shopping (2.96x) and PMax (4.00x) hold up better as product-intent channels. For Retail & Ecommerce advertisers, the portfolio should skew toward Shopping and PMax rather than brand search alone.

Beauty & Personal Care: Brand Search Leads at 3.71x

Beauty & Personal Care shows a distinct pattern in this dataset. Search Brand leads at 3.71x — below the all-vertical median (4.14x) — while Search Non-Brand sits at 1.70x, with a wide 95% interval around the median [0.01x to 15.53x] that reflects high variance across advertisers. The typical beauty brand sees modest incremental returns from non-brand search; a small number of outliers see exceptional results. Do not use the Non-Brand median as a planning assumption without validating your own incrementality first.

PMax at 2.37x falls between Brand and Non-Brand. For Beauty & Personal Care advertisers, Search Brand is the highest-ROI and most reliable format. PMax can extend reach, but should be validated against Brand Search before receiving significant incremental budget.

Travel & Hospitality: The Most Balanced Google Ads Portfolio

Travel & Hospitality presents the most balanced vertical in this dataset. The spread between campaign types is relatively narrow, suggesting that Google's full portfolio — brand search, non-brand, and PMax — all contribute meaningfully. The lower Search Brand ROI (4.00x) compared to Media & Entertainment or Healthcare reflects the higher competitive intensity and longer consideration cycles in travel.

Google Ads ROI Benchmarks by Vertical: Summary Table

Cells marked — indicate insufficient sample size (n<5) for reliable median estimation. Treat as directional only.

How to Use These Google Ads Benchmarks for Budget Decisions

These numbers are not targets. They are reference points for asking better questions.

1. Audit your channel ROI against vertical-appropriate benchmarks, not all-vertical averages. If you are a Retail & Ecommerce advertiser with Search Brand ROI of 3.0x, you are near the vertical median (3.19x) — not underperforming. If you are in Healthcare with Search Brand at 3.0x, you are significantly below median (9.32x). The same number tells a different story in each context.

2. Test Video before your competitors do. At 2.70x median ROI, Video / YouTube is competitive on an incremental basis in this dataset, on a small sample (14 models). The brand equity built through Video can show up in Search Brand ROI over time. Test it with your own incrementality measurement before scaling it.

3. Distinguish Performance Max from Performance Max. A PMax campaign in Retail & Ecommerce has a median of 3.59x. Healthcare PMax data has insufficient sample size for benchmarking — treat directionally only. Automated bidding optimizes toward the conversion signals you give it, not toward your vertical's economics. Set vertical expectations accordingly.

4. Read each median together with its spread. The P10 to P75 range for Search Non-Brand spans 0.06x to 11.93x. The same campaign type, in the same vertical, can deliver near-zero or exceptional returns depending on execution, competitive dynamics, and landing page quality. The median tells you the central tendency, the 95% interval around it tells you how precisely that median is estimated, and the P10 to P75 range tells you the stakes.

5. Use MMM, not platform attribution, to evaluate these formats. Every benchmark in this report is MMM-attributed. Platform-attributed figures will often differ: in our comparison of 449 week-matched pairs, the platform figure was higher for 72% of advertisers on Google, though not for all. We expect the differences to be largest for Search Brand (where organic intercept is common) and Video (where view-through windows add credit). If you are evaluating campaign ROI against platform ROAS alone, you are comparing against a different dataset. A calibrated Marketing Mix Model is the method we used here to isolate incremental contribution for budget allocation decisions. Search Brand is the format most exposed to over-crediting, because a click on a brand ad can take credit for a purchase the organic result would have delivered.

Google Ads Benchmarks FAQ

Why should I still invest in Search Brand if Non-Brand has higher ROI?

Defensive necessity. If you don't bid on your own brand keywords, competitors can. When a competitor bids on your brand name, they can intercept users who already intended to buy from you. A user typing your brand name into Google has usually completed the consideration and decision stages. Where competitors bid on your brand, not investing in brand search doesn't save budget; it hands revenue to them. The 4.14x median is lower than Non-Brand (5.21x), but Search Brand serves a different function: it protects revenue you have already earned through brand building. The ROI floor is high (P10 = 1.31x, still positive) and the spend ceiling is low (limited by brand search volume), making it a low-risk, defensive allocation in this dataset.

Why is Search Brand ROI sometimes lower (P10 = 1.31x)?

Several factors can drive lower incremental ROI from Search Brand campaigns. The most common is organic cannibalization: users who search your brand name would have clicked the organic result below your paid ad if the ad weren't there. In high-organic-share categories, brand keyword spend captures revenue that was never at risk of being lost. MMMs that properly account for organic baseline will assign lower incremental value to brand search in these cases. This is not a failure of the channel — it is accurate measurement. Even at P10, the 1.31x floor is positive.

Should I always prioritize Search over Performance Max?

Not necessarily. The comparison depends on scale and objectives. Search requires active keyword management and is limited by query volume. PMax can reach demand that Search cannot access, and at 4.64x median ROI it delivers positive returns. The right portfolio typically uses Search for high-intent, high-value queries and PMax to extend reach where explicit keyword targeting is impractical.

Why do Media & Entertainment and Healthcare have such high ROI across all formats?

Two structural factors: customer lifetime value and search intent quality. In both verticals, the revenue per conversion is typically high (subscriptions, healthcare services) and the search queries are high-intent (people in active decision mode). Google Ads has higher incremental value when it intercepts qualified, high-intent demand. Lower-ticket, lower-intent categories naturally see lower incremental returns from the same investment.

Is 2.70x a good ROI for Video / YouTube?

Yes, when you account for what Video is actually doing. Video is an upper-funnel channel. The 2.70x median represents MMM-attributed incremental revenue, which means it already accounts for the delayed and indirect effects the model can capture. At 2.70x, Video is competitive with Meta and TikTok Performance advertising on an incremental basis, while reaching different audiences at different points in the funnel. The sample is small (14 models, $3.6M of spend), so treat this as directional. If YouTube also builds brand presence that lifts Search Brand ROI later, the 2.70x may understate its full-funnel value.

Are Google Ads worth it in 2026?

According to this dataset, every Google Ads campaign type analyzed delivers a positive median incremental ROI, from Demand Gen at the lower end (2.22x) to Search Non-Brand at 5.21x. The question is not whether Google Ads are worth it, but which campaign types deliver the highest incremental return for your specific vertical. A Healthcare advertiser sees 9.32x on Search Brand. A Retail & Ecommerce advertiser sees 3.19x on the same format. The channel works; the returns vary by 3 to 4x depending on what you sell and how you measure.

What is a good conversion rate for Google Ads?

Cassandra's own data in this report covers incremental ROI, not conversion rates. For platform conversion rates, LocaliQ's 2026 Search Advertising Benchmarks put the all-industry average for Google Ads at 8.18%, with wide differences by industry (see the section on Google Ads CPC, CTR and conversion rate benchmarks above). Conversion rates measure platform-reported efficiency (clicks to conversions), not incremental business impact. A campaign with a high conversion rate can still have low incremental ROI if it is converting users who would have purchased organically. Read the two together: conversion rate to improve ads and landing pages, incremental ROI per dollar of spend to set budget.

How reliable are the vertical-level findings given the sample sizes?

Robustness varies by vertical. Fashion & Luxury, Retail & Ecommerce, and Beauty & Personal Care have sufficient sample sizes for reliable conclusions. Media & Entertainment and Healthcare are directionally strong but should be treated with wider confidence intervals given smaller client counts. Any campaign type showing n<5 in the detailed tables should be considered directional only.

Can I use these benchmarks to set budget allocation targets?

Use them as starting points, not conclusions. A benchmark tells you the central tendency of a distribution; your business operates at a specific point within that distribution. Set your own MMM baseline first. Then use these benchmarks to identify where your numbers diverge meaningfully from vertical medians — those divergences are where the most important questions live.

What is a Marketing Mix Model and why does it matter for benchmarks?

A Marketing Mix Model (MMM) is a statistical model that measures the incremental revenue contribution of each marketing channel, isolating the effect of one channel from all others, from seasonality, and from organic baseline. Unlike platform attribution (which counts clicks and assigns credit), MMM answers: "what revenue would I lose if I turned this channel off?" Every benchmark in this report is MMM-attributed, so it estimates incremental contribution rather than attributed credit.

The Benchmark Problem — And What Comes Next

Benchmarks have a measurement problem. Platform ROAS averages and industry CTR reports are built on platform data, which mixes incrementality with attribution. They measure what happened and what was credited. They do not measure what your spending caused. Both kinds of benchmark are useful, for different questions.

This report attempts a different standard. Two hundred and fifty-three Marketing Mix Models. Fifty-nine advertisers. Three hundred and eighty-three million dollars in analyzed spend. The picture that emerges is more nuanced than the platform-facing benchmarks suggest — and in some places, more optimistic.

Search works — both brand and non-brand — across virtually every vertical we analyzed. The incremental evidence for search intent capture is robust. Search Non-Brand leads at 5.21x median, reflecting the high incremental value of capturing category-level intent. Search Brand at 4.14x serves a different but essential function: defending revenue against competitors who will bid on your brand if you don't. Performance Max delivers consistent middle-ground returns, with the important caveat that its opacity makes optimization difficult. Shopping holds up particularly well for retail categories.

Video and Demand Gen carry more nuance than the negative narrative suggests. At 2.70x median incremental ROI, YouTube is competitive with social performance channels in our data while reaching entirely different inventory, on a small sample of 14 models. Brands that skip YouTube may be leaving low-cost reach on the table and ceding brand equity ground to competitors who show up there. If that brand equity carries into search, as we expect, it should show in their Search Brand ROI over time.

Vertical context changes everything. The same campaign type can be your highest-ROI channel or your biggest budget mistake depending on what you sell, who buys it, and how much intent precedes the purchase. No all-vertical benchmark survives contact with a vertical-specific business reality.

Use these numbers as anchors for better questions. Not as substitutes for measurement.

Author: Gabriele Franco, founder of Cassandra

Data sourced from Cassandra's MMM benchmark dataset. Revenue models only. Sporting sector excluded. USD-standardized using time-of-model exchange rates. N=253 models, 59 advertisers, data period 2023–2025.

Measure Your Own Incremental ROI

These benchmarks show what the market looks like. The next question is: where does your portfolio sit?

Cassandra is the AI marketing brain for growth teams: it measures the incremental impact of your media spend with marketing mix modeling and incrementality tests, then plans budgets and does the work in your ad accounts, within your rules. If you want to know your actual incremental ROI by channel, not platform-attributed ROAS — book a demo or start a free trial.