The Digital Brand Tree. Part 2. June 2026

This is part two, of three companion briefs to the thesis “The Digital Brand Tree.” 
See also 1: Introducing the opportunity  ·  3: The beneficiaries of an antifragile digital brand.

Welcome to the 2^nd part of my thesis/proposal around the digital brand tree. In part 1 (the introduction) I set out the tree: the divisions of an organisation are the roots that produce, the customer connections (channels/touchpoints) are branches that engage, and the trunk (made of governance and technology) connects and makes it a tree.

I’ll go inside the trunk, looking at how governance and technology together carry production out to the channels and carry performance back. This will detail how and why it’s so important as a stewarded framework, and in part 3 will build on the benefits and future steps.

Because the trunk centres the tree, it’s particularly sensitive to the fractures of ownership and decision-making – in other words, whether the tree flourishes, or is stunted. Growth starts by building a strong foundation on the organisational semantics that define why the branches and roots exist, and why their existence in the tree has meaning. These include tangible schema such as revenue and customer targets, strategic OKRs (including growth, risk, ROTE), and intangible schema such as NPS, trust, customer engagement (including partners, investors, regulators) as well as values and culture.

Governance is the essential ‘why’ logic that decides what is allowed to flow, in which direction, at what velocity. Most divisions of organisations have a complicated relationship with governance – particularly the balance of speed vs quality of decision.

All parts of an org run some level of governance: prioritisation, curation, funding. The natural discernment of work is a key step in governance, with semantics determining commonality across the organisation. And it does this hand in hand with technology as the enabler of efficiency and antifragility. Technology is essential to the connection and efficiencies, storage and retention of the noise and signal records, operational accountability. It forms the ‘how’ that cannot be separated from getting a signal record from roots to branches and back, just as governance forms the ‘why’ that the signal record exists in the first place in the framework.

With a reliance on semantic meaning, the digital brand tree framework is seen as stewardship – it does not seek to own, or override. All the work that is recorded in the model can continue without the tree (albeit much more fractured). Stewardship cares that the noise is filtered into the important signals. Stewardship also brings to focus the antifragile ideal state, by nurturing the semantic meaning from theoretical into operational.

Stewardship also brings to focus the antifragile ideal state, by nurturing the semantic meaning from theoretical into operational.

In practice, the signal record that is key to this framework, is a capture of an event – something that stands out in the noise because it has hit a level of semantic importance. This is the governance. The signal record is able to get from branches to roots and back, because of the technology.

The customers for this proposal are many. I’ll go into this in more detail in part 3, so the main point I’d like to make here is that it is easy, but limiting, to believe that this is a purely outward facing concept. While customers are the main data creators for the branches, the roots are also a rich source of data. In reality, as the tree embeds itself and grows, prediction will become one of the key benefits for solid signal records, and antifragility. The more signal records are collected, decisions made and implemented, the more data is collected about which decisions resulted in a better outcome. Each signal record builds additional records. This lends itself not only to efficiency, but being a proverbial prediction growth engine.

Mapping the branches: ownership and classification

Before the trunk can govern a branch, the organisation has to be able to describe it. A branch is any channel or touchpoint where the brand meets a customer, and describing one well means holding two different things apart: who owns it, and why does it exist.

Ownership of a branch is based on ownership of the customer of that branch. This can be marketing, technology, risk, customer service, etc Ownership tends to determine the decision making behaviours and prioritisation of the signal record, but this is not always correlated. A division can legally own a channel but not be the responsible operator (or decision maker).

Once up and running, a signal record resulting from an event in a branch, goes to the owner of the branch’s customers. There should be no surprises there, and any out of the ordinary signal records for a branch would naturally be flagged with a higher priority. Within semantic meaning, the signal record will identify if the legal owner needs to get involved (for example a very serious breach on social media may need legal owner escalation to resolve).

I like to classify branches into three buckets: owned, earned and paid. Owned is the directly controlled platforms (such as account portals, main website), paid being mainly events, advertising or membership platforms where a direct cost exchange results in how the brand is visible on that platform. Earned is the somewhat gray area in between, where a brand is shown or interpreted without an exchange of money or value. This includes most social media posts, news and review websites. But it’s also regulators, media, fraud, where the brand is part of the exchange in value, but where the organisation has less/no control over how it appears.

A branch can span multiple buckets – Meta/Facebook will span both paid and earned. The signal record will be able to identify and flag that a paid campaign is prioritised for marketing, while a series of complaint comments or a cluster of impersonation accounts, also prioritises that record for risk, legal or customer-service signals.

Governance classifications

Mapping governance should target explicit, repeatable sets of decision rights, thresholds and records that turn intent into consistent action. For the digital brand tree, a workable governance process has five defined parts.

The RACI (who is responsible, accountable, consulted and informed).  Most digital fracture is a RACI failure wearing other clothes. The social account that is owned by a junior marketer who has since left. The domain registered to whichever team happened to be closest at the time. The app-store account no one can now access. The campaign page that outlived the campaign. The RACI is the instrument that identifies ownership (accountability) and decision-rights cascading through responsibility to informed. The RACI is a key governance tool for stakeholders and their decision-making to legal-provisioned prioritisation.

Decision rights (who decides). While ownership is essential for accountability, the decision rights are the most essential operational classification that needs to be embedded in the signal record schema. Who needs to receive this output because they are the most important person for making a decision on a signal record.

Decision thresholds (what needs more scrutiny).  Not every decision deserves the same weight, particularly given the absolute noise and wealth of records that would be generated without filtering signals. Once working, anticipated outcomes should flow through because of these thresholds which should be based on expected target outcomes, strategic value and targets (eg sales), as well as sensitivity, regulatory triggers, complaints. These should cascade from the RACI, so the more that a signal record exceeds the threshold for the Accountable recipient, the more prioritised the signal record for senior or multi-function review. It may make sense to have a certain Accountability and Urgency flag for the most serious signal records to be flagged for C-Suite and Board.

Documented workflows (how the work actually runs).  A workflow is the path a piece of work takes from production, through the trunk, out to a branch – and the path a signal takes on the way back. Writing it down is unglamorous, and that is precisely the point. An undocumented workflow lives in one person’s head and leaves when they do. A documented one can be audited, improved, handed over and, increasingly, executed in part by agents.

The record and the precedent baseline (how the process learns).  Every decision of consequence is captured: the decision itself, its rationale, the approvals behind it and the outcome. Over time those records become a precedent baseline: the organisation’s own evidence of what it decided last time and how that turned out. It is what lets the process improve rather than simply repeat.

The signal record: a single, modular structure

The signal record becomes the durable unit that builds a signal lifecycle. Structurally, it holds the branch the signal came from, that branch’s owner and its classification, the initial detection and score, the enrichment and context, the governance decision and its rationale, the approvals behind that decision and – critically – the outcome. Ownership and classification are not incidental schema here; they are what tell every later step how the signal should be handled.

Importantly, this is also cumulative, and would ideally be a recursive model. The more signals that fit the pattern of an original model, the more the model learns and can alter prioritisation based on previous signal records and decisions made. This essentially becomes a predictive model, that over time can more accurately identify patterns in decisions and build those into the signals for the decision makers.

This will be especially useful where a signal is sensitive or scores highly to be prioritised for the attention of certain higher tier divisions – for example Legal, CEO office, Security, or requires multi-function approval (for example Communications, Legal, Security may be necessary in a data breach situation). In the current model, the reaction time is slowed down by fracture, and reliable process is often lacking – deprioritised in light of the urgency. A signal record becomes an auditable, repeatable practice, and feeds the recursive learning pattern.

. . . . so do their decision rights: what one division may decide alone, another must escalate.

One question this framework has to answer head-on is that no two divisions read a signal record the same way. A single signal around a regulatory change spikes cost exposure for Finance; emerging audit risk for Compliance; complaints strategy for Marketing. Their triage differs, and so do their decision rights: what one division may decide alone, another must escalate. The signal record is built for these moments. Stewardship designed to capture and objectively measure. It is pattern-robust: reliance on semantic schema holds the same record for all, but can still prioritise the flag for the recipient team, so a record raised by marketing is still legible to risk, and a precedent set in one division is still usable in another. This makes the trunk legible to an organisation that will never read it from a single point of view, and essential by pulling together a steward’s eye view of the organisational semantics.

There is a deeper challenge that is raised in practice: divisions do not come to a shared record empty-handed. Risk has its own system of record, legal another, finance and marketing others again. A canonical signal record threatens those systems. The digital brand tree is designed not to replace those systems but to place a framework around them. The steward’s eye view connecting the systems in use, not a new master system they must migrate to. Nor does it relocate any decision: the seams between divisions visible, but each division still decides within the mandate it already holds. In practice, fractures will likely still exist, but will be less foundational and disruptive.

This is where agentic technology earns its place in the model. Agents are well suited to the unglamorous translation work of reading each division’s system in its own format and projecting it into the record schema. The divisions do not change how they work; the agent layer is there to reconcile the difference. Building governance as central to the trunk also means explainability is part of the design. This is a precondition that delivers clarity, reportability and semantic veracity.

Better branches: where antifragility is built

Documentation, defined workflows and disciplined record-keeping are unglamorous, but essential for governance. That is antifragility in practice: improving by being able to survive under stress.

With agentic modelling earning a place by delivering governance, it means that the branches can be governed with more confidence. A branch left ungoverned is fragile. It has no clear owner, no measurement, workflow or lifecycle. When something goes wrong: a compromised account, an impersonation site, a campaign page still quietly collecting data long after the campaign; the failure is discovered late, and fracture has already embedded damage.

Building a foundation of governance framework on the organisational semantics, supported by an explainable by design practice, means the branches can thrive independently, but be made more antifragile. Social media can perform at its own pace, but the signal records will absorb the noise and report how they need to. The framework does not attempt to force the branch into a fixed schema for reporting and definition. Instead it absorbs and reconciles, making it easier to add branches and shed dead ones.

From here, I’ll look at the benefits and stakeholders. Who performs the role of the steward, how the system works in practice, matching to the decision maker and why. Time for Part 3.

---

AI note: This work was created with the assistance of AI. As with Part 1, Claude Opus 4.7 was prompted to ingest multiple drafts and transcripts of my voice notes, as part of an overall project on this topic, then create a single thesis as well as 3 shorter articles (1 for each section of the overall thesis), after which it was prompted to tweak the sections at least three more times. I’d give it a 5 out of 10 for the work it did in this article. The coordination of my drafts was pretty good structurally, but I had to pare back the technical bent of the output, and rewrite quite a large amount.