Lesson 1 of 16 · Automotive SPICE, the vital 20 %
The two axes of a rating
"What ASPICE level is this company?" has no answer. By the end of this lesson you will know what to ask instead, and you will be able to turn a row of ratings into a capability level in your head.
Throughout the course we follow one invented project: the Seat-Heat team, eight engineers at a Tier-1 supplier writing the control software for a seat-heating ECU. They work in Jira, Git, pull requests and CI. Their OEM customer has announced an ASPICE assessment.
A grid, not a score
An Automotive SPICE assessment does not give an organisation one number. It fills in a grid with two axes[PAM §3, p. 13]:
- The process dimension: which processes are being looked at. These come from the process reference model (PRM), which has 32 named processes such as
MAN.3 Project Management,SUP.8 Configuration ManagementorSWE.1 Software Requirements Analysis. Each one has a purpose and a list of outcomes, the observable results of achieving that purpose[p. 14]. - The capability dimension: how well each process is run, expressed as a capability level from 0 to 5. Each level is made up of process attributes (PAs), and each PA is rated on its own[§3.2.1, p. 17].
The process assessment model (PAM) is the PRM plus indicators: practices and information items that an assessor looks for, gathers objective evidence on, and uses to justify each rating[§3.3, p. 23]. The output is a capability profile: one level per assessed process.
So the question changes. "Is Seat-Heat level 2?" cannot be answered. What can be answered is "what level does Seat-Heat's project management reach? And its configuration management?" One team can come out at level 2 on one process and level 0 on another.
The PAM is the WHAT
The PAM describes three levels of abstraction for the word "process"[§3.3.3, Fig. 3, p. 26]:
| Level | What it holds | At Seat-Heat |
|---|---|---|
| WHAT | Goals of the process, why they matter, dependencies. This is where the PAM sits. | "Progress of the project is monitored and reported" (MAN.3 outcome 6) |
| HOW | Methods, tools, templates, workflows, and who may do what. | Their Scrum setup, the Jira workflow, the branch policy, the PR template |
| DOING | Performing the tasks with the method on real work. | Sprint 14's board, PR #812 and its review, last Tuesday's release |
The PAM deliberately prescribes no lifecycle, no order and no document structure[§3.3.4, pp. 26–27]. The VDA Guidelines say the same about agile methods: they sit at the HOW, so ASPICE and agile "cannot, by definition, contradict each other". The only question is whether a given implementation satisfies the principles[Guidelines §2.5.2, p. 76]. The assessor's job is to map the HOW and the DOING onto the WHAT: read the defined method, interview the people doing the work, inspect the repositories, then rate.
Six levels, nine attributes
| Level | Name | Its process attributes |
|---|---|---|
| 0 | Incomplete: the purpose is not achieved | none |
| 1 | Performed: the purpose is achieved | PA 1.1 Process performance |
| 2 | Managed: planned, monitored and adjusted, with work products controlled | PA 2.1 Performance management PA 2.2 Work product management |
| 3 | Established: run from a defined, organisation-wide process | PA 3.1 Process definition PA 3.2 Process deployment |
| 4 | Predictable: operates within defined limits | PA 4.1, PA 4.2 |
| 5 | Innovating: continually improved | PA 5.1, PA 5.2 |
Source: Tables 3 and 4, pp. 17–18. The VDA Guidelines give rating rules only for levels 1 to 3, "because this covers most of the Automotive SPICE assessments"[§1.3.3, p. 31], and so does this course.
Each attribute is rated N, P, L or F
| Rating | Achievement | Meaning |
|---|---|---|
| N | 0–15 % | little or no evidence |
| P | >15–50 % | some approach, some achievement; parts unpredictable |
| L | >50–85 % | systematic approach, significant achievement; some weaknesses |
| F | >85–100 % | complete and systematic; no significant weaknesses |
Source: Tables 5 and 6, pp. 18–19. P and L can be refined into P−/P+ and L−/L+.
The one rule that turns ratings into a level
Level n is reached when every attribute of level n is rated L or F, and every attribute below level n is rated F.[§3.2.4, Table 9, p. 22]
"Largely" is good enough for the top level you claim but not for the levels underneath it. A process with PA 1.1 at L can never be level 2, however well it is managed, because its foundation is still missing some outcomes.
Practice
A. Ratings in, level out
PA 1.1 L · PA 2.1 F · PA 2.2 F
PA 1.1 F · PA 2.1 L · PA 2.2 L · PA 3.1 F · PA 3.2 F
PA 1.1 P · PA 2.1 L · PA 2.2 L
PA 1.1 F · PA 2.1 F · PA 2.2 F · PA 3.1 L · PA 3.2 L
PA 1.1 F · PA 2.1 F · PA 2.2 P
An assessor judges PA 2.1 to be about 40 % achieved. What is the rating?
B. WHAT, HOW or DOING?
"Progress of the project is monitored and reported."
Seat-Heat's two-week Scrum sprints with a release every third sprint.
Pull request #812, with its two approvals and green CI run.
MAN.3.BP2: "Define the project lifecycle."
Branch policy: every merge to main needs two approvals.
Release 2.3.0, tagged on 2026-09-16 with its release note.
C. From memory
Without scrolling up, state the rule that turns PA ratings into a capability level.
Level n requires every PA of level n rated L or F, and every PA of the levels below rated F.
Seat-Heat's manager asks, "Which ASPICE level are we?" In two sentences, what do you answer?
ASPICE rates processes (say MAN.3 or SUP.8), not companies, and each one gets its own level. An assessor maps our methods and the evidence our work left behind onto each process's indicators, so we'll get a profile, likely with different levels for different processes.
Your win: you can read a capability profile and check that it adds up. Next: open one process and see what "the purpose is achieved" actually means, which is PA 1.1, the foundation every higher level depends on.
To keep it: import this lesson's Anki deck (how).
Read next
The primary source: Automotive SPICE PAM 4.1 (Preview), chapter 3, pp. 13–27. It is fifteen pages and it is the whole PAM in miniature. Figures 1 and 4 are the ones worth studying.
Anything unclear? Ask your teacher in the chat. Disagreeing with where something sits on WHAT/HOW/DOING is a good thing to bring up.