Transparency · Data

How Our Benchmarks
Are Built

Every benchmark on this site has a source, a confidence level, and a known limitation. This page explains all three — so you can use the data correctly and know when to trust it.

Last reviewed May 2026
On this page
  1. Primary data sources
  2. How benchmarks are compiled
  3. Confidence levels by metric
  4. Known limitations
  5. How to use benchmarks correctly
  6. Update changelog
  7. Corrections & feedback

Primary Data Sources

Calc4Marketers benchmarks are compiled from a combination of publicly available industry research, platform transparency reports, and aggregated ad intelligence data. No primary data collection is conducted — all figures represent synthesis of third-party published sources.

Ad Intelligence
WordStream / LocaliQ
Google Ads and Meta benchmark reports covering CPC, CTR, CPA, and CVR across 20+ industries. Published annually with large advertiser sample sizes.
CPM Tracking
Gupta Media CPM Tracker
Cross-platform CPM data updated quarterly. Strong coverage of Meta, TikTok, and YouTube with historical trend data going back to 2019.
Creative Analytics
Lebesgue
E-commerce focused Meta and Google benchmark data. Particularly strong on industry-level CPM and CPA variance across DTC categories.
Social Media Intelligence
Rival IQ / Socialinsider
CTR and engagement rate benchmarks across social platforms. Used primarily for Meta CTR and LinkedIn engagement benchmarks.
Platform Reports
Google Ads Benchmarks
Google's own published advertiser benchmarks, including Performance Max and Shopping data. Used for Google Search CPC and Display CPM baselines.
B2B Intelligence
Demandbase / Metadata.io
LinkedIn and B2B-focused CPC and CPM data. Used for LinkedIn platform benchmarks and SaaS/enterprise CPA figures.
Programmatic
Pixis / eMarketer
Programmatic display and cross-platform CPM data. Used for Google Display Network benchmarks and programmatic industry figures.
Country Data
Regional Ad Market Reports
Country-specific benchmarks draw from regional market reports, local agency surveys, and geo-filtered platform data where available. Confidence is lower for smaller markets.
Source availability

Most primary sources are publicly available but some require registration or paid access. Where a source is behind a paywall, the benchmark reflects the published summary figures rather than raw data. If you need the underlying source for a specific figure, use the contact form and we'll point you to it directly.

How Benchmarks Are Compiled

Step 1 — Source collection

For each metric (CPM, CPC, CPA, CTR, ROAS) and each platform or industry segment, we identify all available published benchmarks from the sources listed above. For well-covered segments (Google Search CPC by industry, Meta CPM by placement), this typically yields 4–6 independent data points. For less-covered segments (country-specific CTR, niche industry CPM), sometimes only 1–2 sources are available.

Step 2 — Range construction

When multiple sources are available, we construct a range that reflects the central tendency of the data — not a simple average. Outliers (figures that diverge from the majority by more than 40%) are noted but typically excluded from the published range unless a specific explanation exists (e.g., Q4 seasonality, market-specific anomaly). The "typical" range published on benchmark pages reflects where approximately 60–70% of advertisers in that category would expect to see their results.

Step 3 — Practitioner validation

Benchmark figures are cross-referenced against practitioner experience in active campaigns. Where published figures conflict with observed campaign data in similar categories, the discrepancy is investigated: the source with the larger sample and more recent data is weighted higher. This step catches systematic biases in published reports — for example, some vendor reports skew toward their own high-performing client base.

Step 4 — Contextual framing

Raw benchmark figures without context are misleading. Every benchmark page includes: the platform and targeting assumptions the figure reflects, the campaign objective it's most relevant to, the key variables that move the number up or down, and — critically — the conditions under which the benchmark doesn't apply. A $45 ecommerce CPA benchmark is meaningless without knowing it reflects Google Search, broad match, conversion-optimised campaigns with median audience targeting.

Step 5 — Annual review cycle

All benchmarks are reviewed annually. Platform CPMs in particular change year-over-year (Meta CPMs have risen 8–12% annually for the past three years). Pages are labeled with the year they reflect. When a significant mid-year shift occurs (a platform policy change, a major auction shift), affected pages are updated and noted in the changelog below.

Confidence Levels by Metric & Platform

Not all benchmarks are equally reliable. This table shows our confidence level for each major benchmark category — reflecting source quantity, source quality, and how stable the figures are over time.

Metric & Segment Confidence Sources Key caveat
Google Search CPC — US, major industries High WordStream, Google, eMarketer Varies 3–5× by keyword competitiveness within industry
Meta CPM — US, major placements High Gupta Media, Lebesgue, Rival IQ Q4 figures 40–60% above annual average
Google Search CPA — by industry High WordStream, Google Ads reports Reflects median campaign, not top or bottom quartile
Google Search CTR — by industry High WordStream, Search Engine Land Position-dependent — assumes avg position 1–3
YouTube CPM — by format High Gupta Media, Google Ads Geographic variation is significant (US 30–50% above global)
TikTok CPM — global Medium Gupta Media, agency reports Rising rapidly — 2026 figures may lag current market by 1–2 quarters
LinkedIn CPM & CPC Medium Demandbase, Metadata.io, agency data Audience size dependency makes single figures unreliable; ranges are wide
Google Display CPM Medium Pixis, eMarketer, Google Targeting type (retargeting vs cold) creates 3–5× internal variance
ROAS — by industry Medium WordStream, Northbeam, agency surveys Margin structure makes ROAS uncomparable across businesses in same industry
CPC — India, UAE, Brazil Medium Regional reports, Meta transparency Fewer sources; figures less stable than US/UK equivalents
CPM — Australia, Germany, France Medium Regional agency data, Gupta Media Smaller sample sizes; ranges are wider than US/UK figures
CTR — by country Lower Limited regional sources Country-level CTR varies enormously by industry — country average has limited value
Programmatic CPM — open exchange Lower eMarketer, DSP reports Inventory quality varies 10× within "open exchange" — figure is highly aggregated

Known Limitations

These benchmarks reflect typical, not optimal, performance

Published industry benchmarks represent median campaign performance — what a typical advertiser with standard creative, targeting, and bidding sees. Top-quartile campaigns (excellent creative, precise targeting, well-structured accounts) consistently outperform benchmarks by 30–50%. Bottom-quartile campaigns underperform by a similar margin. If your numbers are significantly above or below benchmark, the gap is more likely explained by campaign quality than by benchmark inaccuracy.

US-centric bias

The majority of available benchmark data reflects US market conditions. Non-US figures — particularly for emerging markets like India, Brazil, and UAE — are derived from fewer sources and carry lower confidence. Country-specific pages note where data availability is limited. For campaigns targeting non-US markets specifically, treat the published figures as directional rather than precise.

Benchmarks assume standard campaign setup

All figures assume: broad or phrase match targeting (not exact match only), standard ad formats (not premium placements), conversion-optimised campaigns where applicable, and no exceptional market conditions (no active crisis, no Q4 holiday surge). Campaigns deviating significantly from these assumptions should adjust expectations accordingly. Q4 campaigns should assume CPMs 40–60% above the annual benchmarks published here.

Industry categories are broad

"Finance" encompasses mortgage lenders, insurance brokers, retail banks, and fintech apps — categories with CPAs that differ by 5–10×. "Ecommerce" covers everything from $15 impulse items to $2,000 furniture. The industry-level benchmarks represent the midpoint of these ranges. If your business is at the extreme of an industry category — very high AOV or very low margin — the generic industry benchmark may be a poor reference point. Calculate your own target CPA and ROAS from your unit economics first.

Platform policies and auction dynamics change

CPMs, CPCs, and CTRs shift as platforms change their auction mechanics, introduce new ad formats, or expand their advertiser base. TikTok CPMs have risen 15–25% annually for three years. Meta's Advantage+ migration has changed how placements are priced. These structural shifts can make even recent benchmarks stale within 6–12 months. Pages are labeled by year; if you're reading a 2025-labeled page in 2027, treat the figures with additional skepticism.

The most important limitation

No benchmark tells you what's good for your specific business. Your break-even CPA (derived from your margin and LTV) and your break-even ROAS (1 ÷ gross margin %) are always more important than industry averages. Use benchmarks to diagnose whether you're in a normal range — not as targets to optimize toward.

How to Use These Benchmarks Correctly

Step 1: Calculate your own targets first

Before consulting any benchmark, derive your own targets from unit economics. Max CPC = Target CPA × Conversion Rate. Break-even ROAS = 1 ÷ Gross Margin. These numbers are specific to your business and always take priority over industry averages. Benchmarks help you understand whether your targets are achievable in the market — they don't tell you what your targets should be.

Step 2: Use benchmarks as a diagnostic, not a target

If your CPM is 2× the benchmark for your platform and industry, something is worth investigating: audience size, creative fatigue, Q4 seasonality, or objective mismatch. If your CPM is in range, the benchmark has done its job. Don't optimise toward the benchmark — optimise toward your cost-per-result. A CPM below benchmark with poor CTR and CVR is worse than an above-benchmark CPM with excellent downstream performance.

Step 3: Compare like-for-like

Never compare a blended CPM across campaign types to a single benchmark figure. Separate by platform, objective, audience type, and campaign goal before comparing. A blended Meta CPM of $11 may contain a $7 prospecting campaign and a $17 retargeting campaign — both healthy for their respective benchmarks but appearing inflated when combined.

Step 4: Account for your market

US benchmarks are 30–50% higher than global averages for most metrics. If you're running in India, Southeast Asia, or Latin America, apply the appropriate discount to US-based benchmarks. If you're running global campaigns, segment by market tier before benchmarking — a single global CPM benchmark obscures regional efficiency differences.

Best use case for these benchmarks

New campaigns and account audits. When setting up a new campaign, benchmarks tell you what a reasonable starting expectation looks like. When auditing an existing campaign, benchmarks identify which metrics are significantly out of range and worth investigating. In both cases, the benchmark is a starting point — your actual data always takes over from week 2 onward.

Update Changelog

Significant data updates and methodology changes are logged here. Minor copy improvements and formatting changes are not included.

Corrections & Feedback

If you find a benchmark that conflicts significantly with your campaign data, or a source that contradicts a published figure, we want to know. Corrections improve the data for everyone.

To submit a correction or suggest a missing benchmark: use the contact form on the About page and include the specific page URL, the figure in question, and the source or campaign data you're comparing against. Verified corrections are applied within one update cycle and noted in the changelog.

Benchmark suggestions — new platforms, industries, or countries not currently covered — are also welcome. Pages are prioritised by search demand and data availability.