{"id":1430,"date":"2026-11-14T19:40:00","date_gmt":"2026-11-14T19:40:00","guid":{"rendered":"https:\/\/cotocus.cn\/blog\/?p=1430"},"modified":"2026-11-15T18:40:00","modified_gmt":"2026-11-15T18:40:00","slug":"identityops-faq-engineering-clients","status":"publish","type":"post","link":"https:\/\/cotocus.cn\/blog\/identityops-faq-engineering-clients\/","title":{"rendered":"IdentityOps FAQ: Questions Engineering Clients Keep Asking"},"content":{"rendered":"<p>Clients who hire engineering consultants increasingly bring the same cluster of questions: who actually has accounts on our platform, how do we know a departure is complete, and what should we measure? The cluster has acquired a name, IdentityOps, though the term is young and not everyone uses it the same way. What follows are the recurring queries from our calls, each answered with reference to what the Ops Atlas has written about the topic and what widely cited standards say. Where the evidence is thin, we say so.<\/p>\n<h2>What is IdentityOps, and is it a recognised standard?<\/h2>\n<p>It is not a standard. The Ops Atlas states that the label is emerging, that no founding document exists and that vendors scope it differently; one describes it as the daily running of a deployed identity system, another as a meeting point of access management, security operations and IT automation. The Atlas offers its own editorial boundary: IdentityOps owns the roster of who exists plus the proof they present, whereas permission design and access approvals sit in sibling practices. The complete argument is in the <a href=\"https:\/\/www.xopsschool.com\/identityops\/\">Atlas article on the discipline<\/a>, and it is worth reading with that caveat in mind.<\/p>\n<h2>What belongs inside the scope?<\/h2>\n<p>According to the Atlas, the in-scope list includes HR-driven joiner, mover and leaver handling, directories and identity providers, single sign-on over SAML or OpenID Connect, automated provisioning through SCIM, enrolment and recovery of second factors, governance of birthright access, machine identities, and signals that point to identity-based attacks. Out of scope are permission models, privileged-access workflows, the HR process itself, customer-facing identity and detection engineering.<\/p>\n<p>For a consulting engagement that boundary is helpful, because it prevents the scope creep that happens when an identity project quietly swallows an access-management redesign.<\/p>\n<h2>Which public standards should an engagement cite?<\/h2>\n<p>The Atlas names a handful. SAML 2.0 dates from 2005 and still underpins many enterprise logins. SCIM 2.0, published as two RFCs in 2015, standardises user provisioning. A 2025 OAuth security-practice RFC tells implementers that the implicit grant should not be used and that the password grant must not be. The US standards body published the fourth revision of its digital identity guidelines in July 2025, and its authenticator volume demands that the middle assurance tier offer a phishing-resistant choice. Government agencies add a zero-trust maturity model and a fact sheet on phishing-resistant authentication, and OWASP maintains a top-ten list for non-human identities.<\/p>\n<h2>Why does disabling an account not finish a leaver process?<\/h2>\n<p>Because several things outlive the account. Access tokens already issued stay valid until they expire; the Atlas gives roughly one hour as a default range for one major platform, and up to around a day when continuous access evaluation extends lifetimes. Session revocation calls can lag a few minutes behind. Acting at the identity provider does not necessarily end sessions inside connected applications. And applications interpret a provisioning deactivation signal in their own way, since the SCIM specification leaves the consequence of an inactive flag to the application.<\/p>\n<p>The Atlas also records an ordering trap in one vendor&#8217;s on-demand push: if you remove the user&#8217;s assignment first and push afterwards, the downstream account can remain enabled until the last push completes.<\/p>\n<h2>How should leaver performance be measured?<\/h2>\n<p>End to end. The Atlas&#8217;s worked example takes forty departures. Thirty-six met the service target at the directory, yet three of that group kept a live login inside a minor application. Judged by the directory alone the score is 90.0 percent; judged by the applications it falls to 82.5. The same logic applies to second-factor coverage: 1,190 of 1,400 users had a phishing-resistant method enrolled (85.0 percent), but 41,000 of 62,000 sign-ins used one (66.1 percent), which signals that policy still lets weaker methods through.<\/p>\n<p>The takeaway for a consulting report is to quote both denominators and explain the gap.<\/p>\n<h2>What service targets are reasonable?<\/h2>\n<p>The Atlas lists illustrative figures, not mandates: four hours to disable a leaver at the identity provider, twenty-four hours to remove access in every application, ninety days for the dormant-account cut-off, and a ninety-day ceiling for the age of a machine credential. Treat them as starting points to adapt to the client&#8217;s risk appetite and to what their applications can actually support.<\/p>\n<h2>How do machine identities fit in?<\/h2>\n<p>They are the part most often missing from client inventories. The Atlas prescribes a lifecycle: find them, record who owns each one and why it exists, set an expiry and a review frequency, issue credentials with short lives, rotate, and retire with proof. It also cautions that a standard cloud-provider credential report looks at just two access keys per user and omits roles, so an inventory built from it alone will have holes. Our earlier note on <a href=\"https:\/\/cotocus.cn\/blog\/protecting-revenue-and-trust-through-strong-web-application-security\/\">safeguarding revenue and customer trust in web application security<\/a> touches on why credentials embedded in applications deserve this scrutiny.<\/p>\n<h2>What do the public incidents teach?<\/h2>\n<p>The Atlas assembles several. A password-spray attack on a legacy test tenant with no second factor, followed by abuse of a test application with elevated rights. A support-system compromise in late 2023 in which a service credential sat in an employee&#8217;s personal cloud storage and session tokens in support files were hijacked, affecting five customers; an affected customer admitted to leaving one token unrotated, along with three service accounts assumed idle. A 2024 campaign against a data-warehouse platform where customers lacked second factors, had credentials unrotated for up to four years and enforced no network allow lists, leading to roughly 165 organisations being notified. And a social-engineering group that exploited help-desk resets, SIM swaps and approval fatigue.<\/p>\n<p>Its conclusion is the evidential heart of the matter: each case defeated a process, not a cryptographic primitive.<\/p>\n<h2>Is single sign-on itself a risk to avoid?<\/h2>\n<p>The Atlas rejects that framing. Central authentication concentrates control, and the identity provider should be treated as the most sensitive tier in the estate. The related warning is about provisioning at first login only: with that approach there is no channel to remove an account later, so the Atlas, citing a container vendor&#8217;s recommendation, prefers SCIM with the just-in-time option off.<\/p>\n<h2>How mature is our programme?<\/h2>\n<p>The Atlas offers a five-step ladder, which it labels as its own proposal calibrated against a government zero-trust maturity model. Level zero is ad hoc. Level one is centralised, with a live HR feed and a measured leaver target. Level two is automated, with leaver compliance of at least 90 percent. Level three is governed, with phishing-resistant usage of at least 80 percent and every machine identity owned. Level four is continuous, with shared-signal events, short-lived machine credentials and a sustained leaver figure of 95 percent or better.<\/p>\n<p>For a client, ask which level the evidence supports, not which level they feel they are at.<\/p>\n<h2>What should the first ninety days contain?<\/h2>\n<ol>\n<li><strong>Days 1 to 30:<\/strong> owners named, application register built, administrator security keys issued, recovery script written.<\/li>\n<li><strong>Days 31 to 60:<\/strong> HR-driven provisioning for one population, SCIM for the top ten applications, a machine-identity inventory.<\/li>\n<li><strong>Days 61 to 90:<\/strong> session revocation in the leaver flow, a first drill, alerts on credential resets and federation changes.<\/li>\n<\/ol>\n<p>Before an engagement starts, a companion checklist for the application side lives in our <a href=\"https:\/\/cotocus.cn\/blog\/questions-to-ask-before-building-a-custom-web-application-the-complete-guide\/\">pre-build interview guide for bespoke web apps<\/a>.<\/p>\n<h2>Who should own the programme inside the client organisation?<\/h2>\n<p>The Atlas does not prescribe an org chart, but its practices imply a few ownership lines. Someone in the people function owns the quality of the HR feed, since every automated flow inherits its mistakes. Someone in infrastructure owns the identity platform and treats it as a tier-zero system. Application owners are responsible for confirming that deactivation messages really remove access. Our recommendation is to write those three names on one page and review them whenever the organisation restructures.<\/p>\n<h2>When is it better not to build a separate practice?<\/h2>\n<p>The Atlas advises that a small organisation may not need one, and that governance suites should wait until HR data is trustworthy. It also notes licensing dependencies: HR-driven provisioning on one major platform needs a paid directory tier, lifecycle workflows need a governance edition, and several SaaS vendors gate SCIM behind enterprise plans. Confirm entitlements before designing anything.<\/p>\n<p>The honest summary is that IdentityOps is a useful organising idea, backed by concrete public guidance and a cautionary body of incidents, but still short of a settled body of practice. Cite it as such.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Twelve questions clients ask about IdentityOps scope, standards, leaver metrics and maturity, answered with citations to the Ops Atlas and public guidance.<\/p>\n","protected":false},"author":4,"featured_media":1431,"comment_status":"","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[331],"tags":[],"class_list":["post-1430","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-devsecops"],"_links":{"self":[{"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/posts\/1430","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/comments?post=1430"}],"version-history":[{"count":1,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/posts\/1430\/revisions"}],"predecessor-version":[{"id":1533,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/posts\/1430\/revisions\/1533"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/media\/1431"}],"wp:attachment":[{"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/media?parent=1430"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/categories?post=1430"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cotocus.cn\/blog\/wp-json\/wp\/v2\/tags?post=1430"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}