Skip to content
Start with Identity
Analysis

186 identity security statistics, each traced to the publisher that measured it

Start with Identity now publishes eight statistics pages: passkeys, MFA, credential attacks, identity breaches, non-human identity, CIAM, digital identity and the identity market. Every figure has its publisher, release date, method, source link and a citation you can copy.

By SWI Community TeamNov 21, 2026

Start with Identity now publishes identity security statistics on eight topic pages, with 186 figures from 49 primary publishers. Each figure is one sentence with the number in bold, the publisher, the release date, how the number was produced, a link to the publisher's own page and the date we last checked it. Every card has a citation you can copy, and each page adds figures computed live from our own datasets.

Why we built it

Search for "passkey adoption statistics 2026" or "mfa survey" and most of the results are roundup pages. They quote a number, link another roundup, and the trail ends at a figure with no year, no sample and no publisher. Journalists end up citing a 2019 number as current. Vendors put a statistic in a deck that its original author never published.

So we set three rules. A statistic qualifies only if it is primary (we link the report, press release or official page of the organization that measured it), dated (we show when the publisher released it) and comes with a stated method (survey sample, telemetry or official count). We read every number on its source page. Anything we could not find there, or that had no date or no method, was dropped. That cut a lot of numbers that circulate widely, including several from our own older research page.

What the pages show

Passkeys. The FIDO Alliance estimates that 7 billion passkeys were in use worldwide in October 2026. Across the 15 consumer services in its Passkey Index 2026, 36% of successful sign-ins used a passkey, and passkey sign-ins succeeded 93% of the time against 78% for passwords and traditional MFA.

MFA. Okta found that 70% of workforce users at its Workforce Identity customers signed in with MFA as of January 2025, and only 14.0% used phishing-resistant authenticators. The UK government's 2025/26 survey found 47% of businesses have some two-factor requirement. Microsoft says phishing-resistant MFA can block more than 99% of identity-based attacks.

Credential attacks. Phishing was the first step in 16% of breaches in Verizon's 2026 DBIR. APWG counted 1,069,681 phishing attacks in the second quarter of 2026 alone. Flashpoint recorded 1.7 billion credentials stolen by infostealers in the first half of 2026.

Identity breaches. Credential abuse was the initial access vector in 13% of breaches in the 2026 DBIR, behind vulnerability exploitation at 31%, but it appeared at some point in 39% of breaches, the most of any action. IBM puts the global average cost of a breach at $4.99M. CrowdStrike measured an average eCrime breakout time of 29 minutes.

Non-human identity and AI agents. In Palo Alto Networks' 2026 Identity Security Landscape survey of 2,930 security leaders, machine identities outnumber humans 109:1, including 79 AI agent identities per human. GitGuardian detected 28.65 million new hardcoded secrets in public GitHub commits in 2025.

Customer identity. In Auth0's 2026 Customer Identity Trends Report, 74% of consumers say they would abandon a site that asks them to create an account before showing key information. In the Thales Digital Trust Index 2026, 68% abandoned or switched providers over slow or complicated sign-up.

Digital identity. The European Commission reports that 95% of EU citizens had access to an electronic ID in 2025. Japan's My Number Card is held by 84.6% of the population. Italy's CIE was used for 295.4 million logins in the 12 months to October 2026.

Identity market. Gartner forecasts $240 billion in worldwide information security spending in 2026. Palo Alto Networks agreed to acquire CyberArk at an equity value of about $25 billion, and Okta reported $2.919 billion in revenue for fiscal 2026.

Our own numbers, updated on every build

Each page ends with a block called "From SWI's own datasets". Those figures are computed when the site is built from data we collect ourselves: the Identity Breach Database, the identity CVE catalog, the Phishing-Resistant MFA Support Matrix, the national digital ID directory, our funding and acquisitions datasets and our pricing data. At launch, 59% of the 123 incidents in the breach database began with a credential or session attack. Only 13% of 95 business SaaS apps let an admin require phishing-resistant MFA. And 49% of the 99 identity equity rounds we tracked since January 2025 pitched AI agents or non-human identities. When the datasets change, these numbers change with them.

How to use and cite them

Journalists and analysts. Every card links the original source. Cite the publisher, then link the statistics page as where you found it. The "How to cite" line on each card is ready to paste, and each statistic has its own anchor, for example /statistics/passkeys/#passkeys-fido-index-success-rate.

Vendors and marketers. Check the method line before you quote a number. Many figures are one vendor's survey of its own respondents or telemetry from its own customers, and the card says so. A 70% MFA rate at Okta customers is not the MFA rate of all companies.

Researchers. The full compilation is free under CC BY 4.0 as CSV or JSON, with publisher, title, release date, URL, method and check date for every row.

What is missing

Some numbers are too new for our rules. The pages do not yet have a primary-sourced survey of decentralized identity or verifiable credential adoption: what we found was either an analyst figure behind a paywall or a vendor quoting someone else. We also could not get dated official counts of US mobile driver's license use. When a publisher releases one with a stated method, we will add it. Spot a newer edition, an error or a primary source we missed? Tell us.

Last reviewed By SWI Community TeamSuggest a correctionHow we research
Independent analysis. No vendor sponsorship. Editorial policy · Report a correction