If DIBCAC reached out tomorrow, would your SPRS score hold up?
With CMMC Phase II on pause, it's tempting to treat cybersecurity compliance as something that can wait. It can't. The pause didn't touch your NIST SP 800-171 obligations, and the government still has a team whose job is to check whether the score you posted in SPRS is real.
In our recent webinar, Lead Certified CMMC Assessor Nick Graning and Information Security & Risk Manager Charles Van Horn walked through how DIBCAC assessments work, how to run a self-assessment that holds up, and what we learned going through a DIBCAC assessment of our own. Here are the highlights.
Why This Matters During the CMMC Pause
At the time of the webinar, the Department of War (DoW) had advised contracting offices not to include CMMC certification requirements in new contracts while it reviews the program. The rules in 32 CFR and 48 CFR remain on the books, which leaves contractors in what Nick called a legal gray area: the requirements exist, but third-party certification isn't being enforced.
What is still required is everything under DFARS 252.204-7012:
- All 110 NIST SP 800-171 requirements — must be implemented
- A current score in SPRS — whether that's 110, 70, or −203
- DIBCAC verification — organizations are still receiving Medium and High assessments
The question isn't whether you have a score in SPRS. It's whether that score would survive someone checking it.
What Is DIBCAC, and What Triggers an Assessment?
The Defense Industrial Base Cybersecurity Assessment Center (DIBCAC) is a team of government assessors within DCMA. Their job is to verify that what you reported in SPRS matches what is actually implemented, and their results are posted to SPRS right alongside your self-assessment.
Notice typically comes by email or phone, often with only two weeks to a month before the assessment. If you've posted a 110 but you're really closer to zero, that's not much time to close the gap.
Common Triggers
- Program priority — your contracts support critical programs or technologies
- Your SPRS score itself — a sudden jump, say from −203 to 110 in three months, can raise a flag
- A government request — a contracting officer or program office asks for a review
- A cyber incident — a DFARS 7012 incident report can prompt a follow-up
And sometimes, as Nick put it, your number simply gets pulled.
Basic vs. Medium vs. High Assessments
There are three levels of assessment under the DoD Assessment Methodology. In short: a Medium checks your paperwork; a High checks your environment.
| Basic | Medium | High | |
|---|---|---|---|
| Performed by | You (self-assessment) | DIBCAC | DIBCAC |
| What happens | You score yourself using the DoD Assessment Methodology | DIBCAC reviews your score, SSP, and supporting documents | Medium review, plus examining, interviewing, and testing on-site or virtually |
| Confidence | Low | Medium | High |
| Score posted by | You, in SPRS | DIBCAC, in SPRS | DIBCAC, in SPRS |
In practice, Mediums often blend in a few phone calls and emailed questions, and Highs are usually conducted on-site. Medium and High results are valid for up to three years.
That makes a Medium deceptively hard. Your documentation has to speak for itself, because you won't be in the room to add context.
Conducting an Accurate Self-Assessment
The best self-assessment follows the same process a third-party or DIBCAC assessor would. That means assessing all 320 assessment objectives, not just the 110 requirements.
Why it matters: a requirement only counts as MET when every one of its objectives is met, with evidence to prove it. Some objectives aren't obvious from the requirement's title. Take MFA (3.5.3): objective [a] is simply that you've identified your privileged accounts. It's easy to skip, and missing it fails the whole requirement.
NIST SP 800-171A (or the CMMC Assessment Guide) lists every objective along with what an assessor expects to examine and test, so use it as your checklist.
How SPRS Scoring Works
You start at 110 and subtract points for every requirement that isn't fully met:
- −5 — failure could lead to network exploitation or CUI exfiltration
- −3 — failure has a specific, confined effect on CUI security
- −1 — limited or indirect effect on CUI security
Rules That Catch People Out
- The range is −203 to 110 — it's possible to post a deeply negative score
- No SSP, no score — without a System Security Plan (3.12.4), you can't assess at all
- Partial credit exists for only two requirements — MFA (3.5.3) and FIPS-validated encryption (3.13.11)
- Open POA&M items still subtract points — a plan to fix something isn't the same as having fixed it
- Some one-pointers can't go on a POA&M — 3.1.20, 3.1.22, 3.10.3 through 3.10.5, and 3.12.4
So if you fail one 5-point, one 3-point, and one 1-point requirement, your score is 110 − 5 − 3 − 1 = 101.
Where Self-Assessments Fall Apart
Here's the number that should get your attention. According to DCMA's Nick DelRosso at a 2022 Cyber AB Town Hall, contractors' scores dropped by 113 points on average between their self-assessed SPRS score and their DIBCAC Medium assessment score.
Companies that self-assess overestimate their cyber maturity. When you're deciding whether a control is met, ask yourself: if I handed this evidence to someone who knows nothing about our organization, would it answer the question?
Most findings come from claims the evidence can't back up. The six most common:
- The SSP doesn't match reality — it was written once and never updated as the environment changed, or it contradicts your policies; if your policy says one thing and your SSP says another, that's a failed control
- Scoping gaps — CUI flows and assets are left out of the boundary; categorize every asset, because if you hand DIBCAC an uncategorized inventory, they'll assume everything is in scope
- "Partial" scored as met — missing one objective fails the whole requirement; if you're not fully meeting the intent of a control, mark it not met
- Cloud and MSP inheritance — without FedRAMP Moderate (or equivalency) and a customer responsibility matrix, you can't show which controls your provider covers and which are yours
- No evidence — "we do that" without an artifact to show it won't hold up, especially in a Medium, where the evidence is all the assessor sees
- Non-validated encryption — encryption is on, but the module isn't FIPS validated, or the device is capable of FIPS mode but it was never enabled; check your modules against NIST's Cryptographic Module Validation Program (CMVP)
Nick's FIPS example: in Cisco Meraki's cloud portal, compatible devices don't run in FIPS mode until you turn on compliance mode, and every device should show green. If a provider doesn't have a responsibility matrix, build one with them and have their representative verify it.
Avoid Inflating Your Score
Charlie shared the habits he's seen inflate scores during internal gap assessments, and what to do instead.
| What inflates a score | What holds up under review |
|---|---|
| Scoring POA&M items as met | Score each objective with evidence in hand |
| Counting planned or in-progress work | Subtract for every open POA&M item |
| Taking credit for a tool you never configured | Have someone uninvolved re-score it |
| Scoring only the systems you know about | Update SPRS when your posture changes |
Done Is Binary
"We're working on it" becomes "met" on paper. Then the work slips behind another priority, and nobody circles back. Treat every control as binary: it's done or it isn't. And "yeah, that's done" on a call isn't sufficient. Ask to see the screenshot with the host, the IP address, and the setting.
Buying a Tool Isn't Configuring It
A vulnerability scanner that isn't running authenticated scans of the right hosts, on the right cadence, with remediation tracked against an SLA, doesn't meet the control. The same goes for FIPS on Linux: the capability is there, but if /proc/sys/crypto/fips_enabled doesn't return 1, it's off.
A lower, accurate score is safer than a higher one you can't defend.
One note from our live poll: most attendees said they were least confident in their evidence for every assessment objective. The good news is that you don't need a separate artifact for each objective. One screenshot that shows the setting and its implementation can cover several objectives, and your policy and SSP count as evidence too.
The best time to find the gaps in your self-assessment is before DIBCAC does. Speak with a consultant about an independent review of your SPRS score, scored objective by objective, with evidence in hand.
Talk to a CMMC ConsultantLessons From Our Own DIBCAC Assessment
Charlie led 112 Cyber through its own DIBCAC assessment, and he was candid that the process felt daunting at first. Three things made it hard:
- Heavy documentation — policies, procedures, plans, and evidence for 110 requirements; what you write is what you're tested on
- Dense jargon — NIST, DFARS, and DoD methodology language has to be translated into your environment
- Many right answers — technology offers many ways to meet one control, and you have to prove yours works
What made it manageable was the right resources and a system to stay organized.
Reliable Resources
Find people you can bounce questions off who know CMMC, and who can help judge how your unique environment meets each control. Communities like the Cyber AB, Discord groups, and Reddit can help, but ground your answers in the official source documents, not guesswork. Most environments run on the same underlying platforms, so you're rarely the first to face a given question.
One Organized System
Catalog every control's status and artifacts in one place: what's met, what isn't, the evidence, and your notes. Scattered Word and Excel files bring version-control headaches, like someone editing a checked-out copy on a plane and merging it back later.
| Control | Status | Artifact |
|---|---|---|
| 3.1.1 | Met | Access control policy |
| 3.5.3 | Met | MFA configuration export |
| 3.13.11 | POA&M | FIPS module upgrade |
Overcoming the IKEA Effect
We overvalue what we build ourselves. When you built the controls, it's easy to score them generously. The fix is to have someone who wasn't involved review your work. Fresh eyes catch your assumptions. Ask for evidence, not explanations, and invite them to challenge you. The goal is to find the gaps before DIBCAC does.
Know Your SSP Cold
"If you can't explain it clearly, you don't know it."
Your SSP is the first thing DIBCAC reads, and it's the story you'll have to tell. Leading up to the assessment, Charlie read 112 Cyber's SSP, several hundred pages of it, at least 20 times, keeping a side spreadsheet of what each control meant internally and who owned it.
As the system owner, you should know the SSP better than anyone and be able to walk an assessor through every control without notes. Picture an assessor with no context and two weeks to learn your system. Would your SSP give them a fair, accurate picture, without sending them through four documents and an obscure vendor knowledge-base article? When you know your own story, confidence replaces fear.
Documentation to Have Ready
Have all of this organized before DIBCAC asks for it:
- System Security Plan — current, approved, and matching reality
- Policies and procedures — covering all 14 control families and consistent with your SSP; put your access control policy side by side with the AC section of your SSP and check that everything, including your organization-defined parameters, matches
- POA&M — open items with owners and dates
- Evidence for each objective — screenshots, configurations, logs, and records; spot-check a sample of scans or logs to confirm they show what you claim
- Network and data flow diagrams — showing the CUI boundary
- Customer responsibility matrix — from your cloud and MSP providers
- Asset inventory — every in-scope asset, categorized
- SPRS score worksheet — how you calculated every point
Then have someone peer review it all. Ask them: if you came back to this in a month, without the knowledge you have from writing it, would it still make sense?
The Risk of an Inaccurate Score
An unsupported SPRS score isn't just a compliance problem. It creates False Claims Act exposure. When you post a score, you're telling the government you've implemented those controls, and every invoice under that contract rests on that claim.
Recent cases, per U.S. Department of Justice settlement announcements:
| Case | What happened | Settlement |
|---|---|---|
| Logzone Inc. (2026) | Self-assessed at 110; DCMA assessed it at −170. Billed the Navy without the required cybersecurity controls in place. | $507K |
| MORSECORP (2025) | Failed to correct an inflated score | $4.6M |
| Honeywell (2026) | Aerospace unit missed NIST 800-171 requirements | $2.0M |
False Claims Act liability can include treble damages plus penalties per claim, and whistleblowers can receive 15–30% of the recovery.
There's no published threshold for how big a gap triggers enforcement. A score of 110 that's really 70 may not, and the cases above likely involved far larger gaps. But it doesn't take long to fall a long way. Nick's example: if you claim monthly vulnerability scans but can only show 10 of the last 12 months, you could fail several requirements at once and lose double-digit points.
Whether you're the IT lead assembling the score or the executive signing off on it, make sure it's a score you'd be comfortable having someone check.
From the Q&A
Does a DIBCAC High assessment with a 110 count as a CMMC Level 2 certification?
No. A DIBCAC assessment is the government verifying that you're doing what you say. It isn't a substitute for Level 2 certification. Nick has even seen DIBCAC assess an organization that already held a third-party certification, checking every control anyway, which verifies both the contractor and the assessor that certified it. He also advised against publicizing that you've received a DIBCAC assessment.
How often does DIBCAC assess?
It varies widely. Some organizations are assessed several times a year, depending on the contracts and information they handle, and others never are. At one of Nick's past employers, DIBCAC visits came every three to four months. A silver lining: you often get the same assessor, so you learn their expectations.
Are subcontractors with no direct DoD contract subject to DIBCAC assessments?
Nick hasn't seen it happen. DIBCAC has limited resources, and primes working directly with the government are far more likely to be assessed than a small subcontractor.
How much detail belongs in an SSP implementation statement?
Keep it simple and straightforward. Answer the control at a high level in the SSP. For example: "We implement CIS Level 2 baselines, deployed and enforced through Intune, with daily compliance checks and reports on devices out of compliance for seven days." Put the granular technical detail in a separate document, such as a security baseline, and reference it.
Most implementation statements should be a paragraph or two. Fifteen hundred words per control is hard for an assessor to read and harder for you to keep current. And don't make the assessor dig through a dozen external links. Your SSP and evidence should give them full context without needing to call you.
Our team works with defense contractors to run accurate self-assessments, build documentation that holds up under review, and prepare for DIBCAC and CMMC assessments. Speak with a consultant to see where your organization stands.
Talk to a CMMC ConsultantThe Bottom Line
In our live poll, about half of attendees said they were only somewhat confident their SPRS score would hold up. That's understandable, but failing even a quarter of your controls can drop a score by 80 to 150 points, depending on their weights. Whether or not CMMC enforcement resumes next month, the path forward is the same: assess every objective, keep evidence in hand, get a second set of eyes on your work, know your SSP, and post the score you can defend.
Have questions about your own self-assessment? Join our biweekly Office Hours, where you can bring your specific questions directly to a certified CMMC assessor.