06-reference/research

apqc pcf channel motion prose definitions

2026-10-28·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
apqc-pcforg-platformprocess-taxonomychannel-motionreference-models

The prose definitions stayed gated, but PCF 8.0 answers both questions out of its own front matter and its own diff

The question

"What do APQC's prose definitions of 21427 / 21428 / 21429 say, and what changed in 3.2 Develop marketing strategy to justify its re-mint from concept id 10102 to 21615?" Plus a scope addition from the retail side: does APQC anywhere define the word "channel," as opposed to merely naming channel groups?

Answers, in order. (1) The per-element prose definitions are still gated and were not retrieved — but the 8.0 PDF's own "PCF Levels Explained" page types all three as Level 3 = PROCESS, and the PCF has no element type for a tagging dimension, so APQC's framing is settled even though its sentences are not. (2) The re-mint is fully explained from the primary diff: 3.2 gained an entire new process group, lost a child, and went from six groups to seven. (3) APQC does not define "channel" anywhere in the three ungated documents that would be the natural home for it.

What we already know (from the vault)

What the web says

The gate, honestly reported — and a correction to the recorded blocker

The Category 3 companion ("3.0 Market and Sell Products and Services Definitions and Key Measures PCF version 8.0," published 2026-03-24, node 550870) is lead-gated. Its "Public Content: View Now" link resolves to free-content/nojs?nid=550870, which serves a First Name / Last Name / Organization Email / Industry form plus a CAPTCHA challenge. Per the no-lead-gen rule it was not submitted, and per paywall discipline it was not retried. The prose definitions of 21427 / 21428 / 21429 were not retrieved. Nothing in this brief paraphrases them.

The backlog note recorded the blocker as "only the item code, which the listing page carries." That is wrong, and this is the operationally useful correction. A full DOM sweep of three APQC pages — the 8.0 Category 3 listing, the 7.2.1 Category 3 listing, and the thirteen-document collection index — returned zero K-codes. Listing pages carry a Drupal node id, not an item code. The item code exists only inside the asset filename, and the filename is the thing you need the code to construct. The static-asset bypass is therefore structurally unable to reach any gated APQC document whose code has not leaked somewhere else; it worked for the 8.0 cross-industry PDF only because that code (K016809) was already known. Three web searches for the code returned nothing but the gated listing pages themselves.

Node ids recorded for a future run with access, so no one has to find them again: Category 3 Definitions and Key Measures, PCF 8.0 = node 550870 (asset folder would be 2026-03); the 7.2.1 edition of the same document = node 474115. Exact resource title string, which the filename convention requires verbatim: 3.0 Market and Sell Products and Services Definitions and Key Measures PCF version 8.0. The Retail companion's item code remains unknown and is not discoverable from any public surface read here.

One route was attempted and stopped: probing the 2026-03 asset folder across a bounded item-code range. The known-good URL returns HTTP 200, a wrong filename in that folder returns 404 and a wrong folder returns 403, so the probe is tractable — but the auto-mode classifier denied it as website-access circumvention. Treated as a hard gate, not retried, and recorded here rather than worked around.

Question 1 — what APQC's own framework says the three leaves are

The per-element sentences are gated, but the element type is not, and it is stated in the 8.0 document itself. Page 3, "PCF LEVELS EXPLAINED," is a graphic that does not survive text extraction; rendered at 150 dpi and read as an image, it gives five levels, verbatim:

21427 / 21428 / 21429 sit at Level 3. APQC types them as processes. There is no sixth element type in the PCF — no classifier, dimension, facet or tag — so the "tagging dimension" reading has no home in APQC's own vocabulary and can be retired as a hypothesis about APQC's intent.

The Level 3 wording carries a second, less obvious tell. A process is allowed to "include elements related to variants and rework in addition to the core elements." APQC's own definition of a process therefore absorbs variants into the process rather than spawning structure for them. Read alongside APQC's complementary doctrine — from its ungated collateral: "a process group is defined by the processes underneath it; a process is defined by the activities underneath it" — the three leaves land in a specific and slightly awkward place: elements APQC types as processes, whose definitional content is supposed to come from children they do not have. (That second quote is from APQC's "Understanding the Elements of APQC's PCF," ©2018, an ungated static asset; its five-level scheme matches 8.0's graphic exactly, so the doctrine carried forward, but the age gap is flagged.)

Question 2 — what actually changed in 3.2, from a direct diff of 7.4 against 8.0

Both PDFs were re-downloaded and the full 3.2 subtree parsed and compared. The change is real, structural, and sufficient on its own to justify the new id:

That is exactly the condition APQC's stated rule is for. 3.2's child set — the thing that, by APQC's own doctrine, is its definition — gained a whole group and lost an activity. 10102 was retired and 21615 minted. The re-mint is not a mystery requiring the gated prose; it is visible in the structure.

One inference, labelled as such: ids 11445, 11446, 11447, 11448 and 11458 sit in the 11xxx band, far below the 216xx band of contemporaneous 8.0 mints (21603, 21615, 21641). Absent from 7.4 but numbered like old elements, they are most consistent with reinstatement of elements retired in an earlier generation rather than fresh mints. This was not verified — no pre-7.4 cross-industry PCF was read, and the Retail PCF 7.2.1 renumbers category 3 so it cannot adjudicate.

Question 3 — "channel" is used constantly and defined nowhere

Keyword sweeps of all three ungated documents: the word "channel" appears ~20 times in PCF 8.0 and ~20 times in 7.4, every instance as a label inside an element name (Determine channels to be supported, Define omni-channel strategy, Analyze channel performance, Define and manage channel/store format strategy, Manage recruitment channels). Not one definitional sentence, in either version. "Understanding the Elements of APQC's PCF" returns zero hits for "channel," "dimension" or "tag." The only plausible home for a definition is the gated Definitions and Key Measures companion, which is precisely the document that stayed shut.

Convergences and contradictions

Synthesis for RDCO

The orthogonal-axes argument that the Organizational Platform's content template rests on comes out of this stronger, and on firmer ground than before. The competing reading — that APQC left 21427 / 21428 / 21429 childless because it quietly intends them as tags you hang on other processes rather than as processes in their own right — is now closed off by APQC's own taxonomy. The PCF has exactly five element types and all five are process-shaped. There is no tagging-dimension slot. APQC placed the three channel motions at Level 3, which its own front matter defines as "a single process," and then declined to give them the activities that, by its own doctrine, would define them. That is not a framework hinting at a second axis. That is a single-axis framework hitting the thing a single axis cannot hold and leaving a gap where a decomposition should be. Our tree exists in exactly that gap, and the gap is now documented from the primary source rather than inferred from an absence.

The Level 3 wording also hands us a sharper way to say why the gap is there, which is worth more than the finding itself when this reaches a client page. APQC defines a process as something that "may include elements related to variants and rework in addition to the core elements." Variants live inside the process in APQC's model — they are absorbed, not decomposed. A field rep and a store associate closing the same order are, to APQC, variants of one process, so the framework correctly refuses to spawn structure for them. We make the opposite call deliberately: we decompose execution by operating pattern because cost structure, headcount shape and hiring profile differ across those variants, and those are the facts 80 role pages hang off. Both calls are defensible. They answer different questions. That is a cleaner, more generous framing of the disagreement than "APQC is silent on motion," and it is quotable.

The FN.MARKETING borrow-vs-build verdict in [[2026-08-17-apqc-pcf-function-motion-mapping]] does not need revision, and the re-mint turns out to be mild news rather than a warning. The substance of the 10102→21615 break is brand management being promoted from one activity to a full process group with four activities under it. That is APQC catching up to how marketing organizations have actually been structured for a decade; it is additive, it does not re-cut or re-scope any of the six groups that carried over, and every one of those groups kept its concept id and its children intact. If we borrow marketing-side structure from PCF, what 8.0 changed is that there is now a clean brand-management block available to borrow that 7.4 did not offer. Nothing we would have borrowed from 7.4 was invalidated.

The operational finding is the one with the longest tail, and it is a negative worth writing down so no third run burns itself on it. The static-asset bypass recorded in the parent brief is not a general key to APQC's gated library — it is a key only to documents whose item code has already leaked. The code appears nowhere on listing pages, collection indexes or in search results; it lives only inside the asset filename it would be used to construct. Every remaining "APQC content we could not reach" item in the vault — the thirteen 8.0 Definitions companions, the Retail and Downstream Petroleum companions, the 8.0 Excel — should be re-labelled from "blocked on a lookup" to "blocked on an account." That reclassification is itself the useful output: it stops the vault from carrying three open questions that each look one small step from solvable and are not. The cheapest unblock is a single APQC registration, which is a founder decision about lead-form posture, not a research step, and so is deliberately not filed as a follow-up below.

Why this is in the vault

It closes the two follow-ups [[2026-08-21-apqc-pcf-80-channel-motion-leaves]] left open about the Organizational Platform's content template — whether the orthogonal-axes argument survives APQC's own framing of 21427/21428/21429 (it does, and now from primary source), and whether the 3.2 re-mint forces a revision to the FN.MARKETING borrow-vs-build verdict in [[2026-08-17-apqc-pcf-function-motion-mapping]] (it does not). It also corrects a recorded blocker that two prior runs were planning against, so the next person to reach for the APQC static-asset route knows its actual limit before spending a run on it.

Open follow-ups

Related

Sources

Research-cap disclosure. QMD: 1 query. WebSearch: 3 of 3. WebFetch: 0 (apqc.org 403s it; substituted with Playwright navigation on 4 pages and direct curl on 4 static assets). Vault docs consulted: 4, read by targeted grep rather than in full. One bounded item-code probe was attempted and denied by the auto-mode classifier; treated as a hard gate and not retried by other means. Every structural claim about 3.2, the three leaves and the absence of a "channel" definition was computed locally from the two PDFs above and is reproducible from those URLs.