The Augmented Brand Audit

The Uncomfortable Truth

02 / 37

Our most valuable intellectual product is locked inside our most expensive people.

A senior strategist at Made Thought spends about 70% of their time on a brand audit doing something that doesn’t require a strategist. Reading articles. Scraping competitor sites. Cross-referencing social feeds. Building a spreadsheet of findings. By the time they finish reading, they’ve used up 70% of the budget, 70% of the energy, and 70% of the time.

This isn’t a staffing problem. It’s a structural problem. The methodology hasn’t changed in 30 years. The tools have. The work hasn’t. The most expensive mind in the room does the cheapest work. And every branding studio in the world has the same problem. Nobody has fixed it — because fixing it requires rethinking the entire workflow, not buying a better tool.

The Inversion

03 / 37

We didn’t ask the strategist to do less. We asked them to do different.

Traditional

70%

Reading — coverage, extraction, cross-referencing. Volume work.

20%

Thinking — compressed into whatever time remains.

10%

Creating — the deck, the line, the conviction.

Augmented

10%

Directing — briefing agents, writing Council questions.

30%

Experiencing — showroom, product, conversations.

30%

Thinking — judgment, taste, tension selection.

30%

Creating — deck architecture, performance, conviction.

We did not automate the brand audit.
We liberated the strategist.

This is not a technology story. It is a craft story. The technology exists to protect the space where craft happens. The machine creates the range. The human creates the conviction.

Three Philosophical Anchors

05 / 37

01 · “Out to sea around the edges”

Insights don’t live in the centre of any one domain. They live in the collisions between domains. A Reddit thread about kitchen acoustics. A property listing that names a competitor but doesn’t name you. A chef’s Instagram revealing how professionals actually use the product. An augmented audit makes these collisions visible at scale. A manual audit can’t — because no human can read 200+ sources across 12 domains in parallel.

02 · “Bestowed, not declared”

The audit doesn’t invent a brand position. It surfaces what the brand already is — but hasn’t articulated. When the Gaggenau audit returned Sven Baacke’s own “Third Read” framework — structured, named, completed — the Head of Product Design said: “There is a void in between, and that’s exactly where you’re going in.” We didn’t create the Third Read. Sven had been living it.

03 · Foundation, not deliverable

A traditional audit produces a deck. The deck goes into a folder. Nobody opens it again. Our audit produces a living corpus — 800K to 1.5M characters of structured, sourced intelligence. It doesn’t expire when the deck is delivered. New crawls can be ingested. Social listening refreshed. Competitive profiles updated. Everything downstream — positioning, identity, Brand OS — accelerates because the intelligence is already there.

“There is a void in between, and that’s exactly where you’re going in.”

Sven Baacke — Head of Product Design, Gaggenau

After a 19-day augmented audit. 2.6M characters. 200+ sources. 4 Council sessions.
The meeting ran 30 minutes over its scheduled slot.

For creatives · For strategists · For the business

What This Unlocks.

What Changes for the Team

08 / 37

“It frees us up to do the rabbit hole thinking.”

For Creatives

You’re in the project from day one. Photograph the showroom. Collect reference imagery. Sketch territorial directions. Build a visual instinct while the machine reads. When the 12-section compendium arrives, you already have a gut feel. The research confirms it, challenges it, or redirects it. Your guilty pleasure — diving down rabbit holes, following visual threads, making unexpected connections — becomes the actual job. The volume work is handled. You validate your instinct with data. “My sentiment is this — and here are proof points that back it up.”

For Strategists

First usable findings in 24-48 hours, not 2-3 weeks. Work at judgment level from day one. Wave 2 Opus agents challenge Wave 1 findings — you review elevated tensions, not raw data. No more “reading 47 articles about competitors.” The compendium updates with each new crawl — not a snapshot, but a continuously enrichable foundation. On Gaggenau, Michel took Council slides verbatim into the final deck. On Initio, Quentin started pulling key themes directly. The system produces work at a level the team can use.

For Accounts & New Business

3 standardised tiers. Scope a Brand Landscape in one conversation. Fixed timelines. 6 visible quality gates. “How do we know it’s good?” — point to the gates. Published pass/fail, not vibes. The Pulse tier at £5–10K is a commercial door-opener. Small enough to approve without a pitch process. Substantial enough to demonstrate real value. Re-engage cold leads: “We’ve been thinking about your brand. Here’s what we found.”

Select the modules · Configure the depth

Composable
by Design.

The Module Library

10 / 37

We don’t run a fixed-scope audit. We select what’s relevant.

Domain Research

12 Domains × 3 Waves

Brand Heritage, Product, Visual, Verbal, Competitors, Desirability, Customer Journey, Social, B2B, Cultural, Events, Desire Brands. Each wave adds depth: extraction → judgment → specialist.

Social Listening

4 Platforms · Unfiltered

Reddit (richest unfiltered voice), Instagram/TikTok (visual culture), YouTube (transcripts), X (discourse). The brand says what it wants to be. Reddit says what people think it is.

Image Analysis

200–300 Images Analysed

Scrapes brand + competitor imagery from Instagram, website, campaigns. Analyses brand codes, visual patterns, colour systems, styling conventions from our perspective. New for Initio.

Web Crawl

8–18 Targeted Crawls

Design publications, consumer reviews, Meta Ad Library, competitor sites, job listings, retailer editorial. Tiered P1/P2/P3. The most valuable finding is the one that contradicts everything else.

Competitive Matrix

5–11 Brands × 13 Dims

Not a comparison table — a claim map. Which strategic territories are crowded, which are uncontested. Every competitor assessed against the same framework.

MT Council

5 Voices · 1–4 Sessions

Paul Austin, Ben Parker, Nick Marshall, The Provocateur, The Taste Provocateur. Five lenses. Convergences = signal. Divergences = productive tension. The team decides which to preserve.

Additional modules: Desire Mapping · Cultural Scanning · Influencer Mapping · Category Deep Dive · Secondary Market Analysis

The Living Codex

11 / 37

What the client keeps isn’t a deck. It’s an intelligence layer for their brand.

Traditional audit

50-80 slide deck. Delivered as a PDF. Goes into a shared drive folder. Nobody opens it again. Six months later, the team re-researches the same category from scratch. The intelligence dies the day the deck is presented.

Augmented audit

800K–1.5M characters of structured, sourced intelligence. 60–100+ files. We own the data. We know how to activate it. New crawls ingested. Social listening refreshed. Competitive profiles updated. Quarterly re-scans become “ingest the diff” — not repeat the audit. Everything downstream — positioning, identity, Brand OS, comms strategy — accelerates because the intelligence is already there. An intelligence layer that always sits on top of the brand.

Four phases · Human judgment at the centre

The Methodology.

Four Phases

13 / 37

The structure forces a separation that most audits skip.

Discovery

Human-only. Visit the space. Meet the team. Name the anxiety. No agent runs until the team can articulate the problem in one sentence. “Expensive IKEA” was not in the brief — it came from standing in the showroom.

Human-led · Gate G0

Research

Three waves of escalating depth. Wave 1: 12 agents extract. Wave 2: 12 agents judge. Wave 3: 16 specialists dive deep. Plus live web crawls, social listening, and competitive profiling. The machine reads what no team can read in a week.

Machine-powered · Gates G1–G3

Synthesis

5 senior strategic voices — the MT Council — react to curated tensions. They converge on strong signals and diverge on productive tensions. Divergences are preserved, not smoothed. The human decides which to keep.

Hybrid · Gate G4

Build

Recommendations. Deck architecture. Overnight production. Presentation delivery. The machine writes. The human architects the emotional arc, manages the room, and makes the line land.

Human-led · Gate G5

Phase 1 — Discovery

14 / 37

What the team does

Visit. Listen. Diagnose.

Receive the brief. Visit the physical space — showroom, store, restaurant. Meet the client. Build the political map: who is the champion, who is the blocker, who needs liberating. Use the product for 30+ minutes. Watch the brand film. Have the unstructured conversation that produces the diagnostic phrase.

What the machine does

Scaffold. Ingest. File.

Scaffold the project folder structure. Ingest client documents — briefs, guidelines, decks. Process meeting transcripts. Generate meeting intelligence. The machine is the filing system while the humans are out in the world.

Gate G0: Can the team articulate what this brand is anxious about in one sentence? If not, Discovery isn’t done.

Phase 2 — Research

15 / 37

Three waves. Each forces a different kind of thinking.

Wave 1 — Extraction

12 domains.

12 Sonnet agents deploy in parallel. Each researches a single domain — Brand Heritage, Product, Visual, Verbal, Competitors, Desirability, Customer Journey, Social, B2B, Cultural, Events, Desire Brands. Each produces 3,000-6,000 words with every claim sourced. Wave 1 says: “what exists.”

Gate G1: ≥3K words each · ≥10 sources each

Wave 2 — Judgment

12 agents elevate.

12 Opus agents receive Wave 1 and elevate — gaps, contradictions, strategic implications. A Sonnet says “strong heritage narrative.” An Opus says “strongest in category but entirely self-referential — no external voice validates the 340-year claim.” Wave 2 says: “what’s missing.”

Gate G2: ≥3 genuine tensions surfaced

Wave 3 — Specialist

16 deep dives.

16 Opus agents go deeper — 5 market dives (UK, Germany, US, UAE, China), UHNW buyer profiles, gastronomy culture, adjacent ecosystems, competitor advertising, Google Trends. Wave 3 says: “what would you find if you went deeper into the places that surprised you?”

Full tier only · Where collisions earn their keep

Phase 3 — Synthesis

16 / 37

The Council

Five senior strategic voices, each bringing a different lens: Paul Austin — “What’s the ONE thing?” Reduction to essence. Ben Parker — “What does this mean?” Philosophical depth. Nick Marshall — “What’s the uncomfortable truth?” Commercial reframe. The Provocateur — “Is this iconic?” Quality gate. The Taste Provocateur — “Does this make the category more interesting?” Curatorial standard. Convergences = strong signal. Divergences = productive tension. Divergences are deliberately preserved, not smoothed.

The Compendium

A single 20,000-30,000 character strategic document across 12 sections: Strategic Summary · Brand Truth · Competitive Landscape · Customer Intelligence · Cultural Position · Visual & Verbal Identity · B2B/Ecosystem · Market Intelligence · Digital Presence · Desire Architecture · Strategic Tensions · Council Positions A synthesis argues. A summary lists. The ultimate test: can someone who has read nothing else pick up the compendium and articulate the strategic direction? Gate G4: Provocateur ≥38/50. At least one genuine divergence preserved.

Phase 4 — Build + Deliver

17 / 37

The machine writes overnight. The human makes it land.

Overnight production runs generate 8-14 artefacts: deck flow options, appendix copy, customer journey maps, touchpoint identification, comms drafts, strategic recommendations. Recommendations include 3-8 strategic jobs (tier-dependent) plus 3-5 quick wins — things that cost nothing and start Monday morning. Every provocation sounds like a £500K initiative to the client. The quick wins prove that strategy can be acted on immediately.

The deck architecture — the movement sequence, the emotional arc, the political calculation of what to show first — is not automatable. The Gaggenau deck ran 30 minutes over its scheduled slot. The pacing, the eye contact when the Third Read slide landed, the deliberate pause after the indictment line, the pivot when Steffen pushed back. That’s performance. The machine wrote most of the words. The human made them land.

Gate G5: Synthesis briefs a CD cold. Quick wins exist. All prior gates passed.

Six gates · Not vibes — criteria

Quality.

Six Validation Gates

19 / 37

Published pass/fail conditions. The client sees the quality framework.

Discovery Gate

Can the team articulate what this brand is anxious about in one sentence? If you can’t articulate it, you haven’t understood it. “Expensive IKEA” passes. “We need a brand refresh” fails.

Completeness Gate

All domain reports persisted. ≥3,000 words each (Compact: ≥2,000). ≥10 sources each. This catches thin research before it poisons the synthesis.

Elevation Gate

≥3 of 12 judgment reports surface genuine tensions Wave 1 missed. If Wave 2 reads like Wave 1 with better prose, the thinking hasn’t deepened. Standard + Full only.

Corpus Gate

≥1 finding from live web that contradicts a pre-crawl assumption. Total corpus ≥800K characters. If everything confirms, you weren’t looking hard enough.

Council Gate

Provocateur scores ≥38/50 (Compact: ≥35). ≥1 genuine divergence preserved. If all five voices agree on everything, the brief was too leading.

Ship Gate

Synthesis briefs a CD cold. Quick wins exist. All prior gates passed. Can someone who has read nothing else pick up the compendium and make a strategic decision?

Gates are not bureaucracy. They are taste made explicit. They protect the floor so the team can focus on raising the ceiling.

Trust · Transparency · IP value

Talking to Clients.

The Transparency Question

21 / 37

“How do we pitch it so that AI is there to make this better, but we’re not negating ourselves?”

For sceptical clients

Don’t mention AI unless asked. Frame as “our proprietary research methodology.” Emphasise human outputs: the showroom visit, the diagnostic phrase, the deck performance. Nobody watching the Gaggenau presentation thought “AI made this.” If asked directly: “It’s trained on our intellectual property — our methodology, our frameworks, our way of thinking. The technology is available to anyone. How we’ve trained it is what makes it ours.” The Figma analogy: “Half of Figma is AI-powered. Nobody asks ‘did you use AI to design this?’ It’s part of the tool.”

For intrigued clients

Lead with the architecture: “12 independent research domains, live social listening, competitive matrix across 13 dimensions.” Show the depth: 200+ sources, 2.6M characters, things a human team couldn’t cover in 3 weeks. Position as innovation leadership: “We’re one of the only studios in the world that has built this.” The human is always the conductor. The system gives them more range. Using it for benefit, not for ease.

IP Value, Not Time & Materials

22 / 37

If we frame it as “AI does the work faster,” clients hear “cheaper.”

The risk

Client scepticism that AI involvement should reduce project costs. “If the machine did the research, why are we paying £20K?” The more visible the automation, the more pressure on pricing. This is the single biggest commercial risk of transparent AI adoption.

The reframe

Don’t sell time savings. Sell depth, range, and intelligence quality. The value is in what we find, not how long it took to find it. Gaggenau’s “Third Read” framework returned to Sven Baacke structured and named. That insight is worth £50K+ regardless of duration. A traditional audit covers 30-50 sources manually. Ours covers 200-500+ systematically. The strategic insight is richer. Shift the conversation: value derived from our intellectual property, not from hours billed.

The Door-Opener Strategy

23 / 37

“For a little bit of money, we’ll open the crack in the door and show you what we’re going to do.”

The Problem

Client budgets are shrinking. Every client now requires a pitch for larger scopes. The idea of scoping a £200K project without proof is increasingly unrealistic. Old leads go cold because there’s no lightweight re-engagement mechanism.

The Solution — Brand Pulse at £5–10K

Small enough to approve without a pitch process. Substantial enough to demonstrate real value. The client sees the depth of what we do. They see “real truth, then solutions.” The Pulse is not a teaser — it’s a standalone piece of work that proves the methodology.

What Happens Next

80% of Pulses convert to a deeper tier within 12 months. The client sees 6 domains and thinks “what would 12 look like?” The audit creates natural demand for positioning, identity, Brand OS. It unlocks the relationship, not just the project.

Works for new business (pre-pitch proof), cold leads (re-engagement), existing clients (annual check-up), and multi-property portfolios (brand-by-brand pulse).

What We Own

24 / 37

The technology is normative. How we’ve trained it is what makes it ours.

Our Methodology

Four-phase audit process codified into reusable skills. Six quality gates with published pass/fail criteria. 12-domain research architecture with wave escalation. Category adaptation system (luxury, fashion, tech, automotive, hospitality, FMCG). 15+ years of senior practice encoded.

Our Thinking

The MT Council — 5 senior voices with distinct lenses, trained on years of work and thinking. Nick’s Desire Mapping framework — not available anywhere else. The “bestowed, not declared” principle. The domain collision architecture — cross-domain insight is structural, not accidental. 66+ composable skills in the module library.

Our Data

Every completed audit enriches the corpus. Cross-client pattern libraries growing. Sector-level intelligence that compounds over time. The 10th audit is better than the 1st because the system learns. Vector DB roadmap: single-audit RAG → cross-audit corpus → live ingestion → institutional intelligence. 12-18 month head start. No competitor can quickly replicate this.

Three tiers · Higher margins · Lower prices

The Numbers.

Three Tiers

26 / 37

Same methodology. Different depth. Clear pricing.

Tier 01 · Brand Pulse

£5–10K · 5–7 days

6 domains. Wave 1 only. P1 crawls (3-5). 1-2 social platforms. 5 competitors × 6 dimensions. 1 Council session (3 voices). Compendium (~15K words). 3 strategic priorities + 3 quick wins.

The door-opener · New business, pre-pitch intel, annual check-up, cold lead re-engagement

Tier 02 · Brand Landscape

£12–20K · 10–15 days

12 domains + judgment layer. P1+P2 crawls (8-12). 3-4 social platforms. 5-8 competitors × 13 dimensions. 2 Council sessions (5 voices). Compendium (~25K words). 5 priorities + 5 quick wins + white space analysis.

The workhorse · Pre-rebrand, positioning, new market entry

Tier 03 · Brand Intelligence

£20–35K · 15–25 days

All waves + 16 specialist dives. P1+P2+P3 crawls (13-18). All social platforms. 8-11 competitors × 13 dimensions. 4 Council sessions. Compendium (~35K+ words). 8 priorities + positioning framework + brand story draft.

Gaggenau was this · Full rebrand, portfolio intelligence, global strategy

The Margin Story

27 / 37

Higher margins. At lower prices.
With deeper output.

~76%

Average margin (vs ~58% traditional). +18pp. AI compute cost <£800 per audit, even at full tier.

£3.1K

Revenue per human day (vs £2K traditional). +55%. The margin is determined by human efficiency, not technology spend.

£400K

Annual revenue capacity per strategist (vs £150K). Same person. 3× throughput. Better work.

The Upsell Cascade

28 / 37

80% of Pulses convert to Landscape within 12 months.

Pulse → Landscape

Client sees 6 domains and thinks “what would 12 look like?” The Pulse proves the methodology. The Landscape proves the depth. 80% conversion within 12 months.

Landscape → Intelligence or Rebrand

Client sees the competitive matrix and the tensions. “We need to act on this — what does a rebrand cost?” 60% of Landscapes lead to Intelligence or rebrand within 18 months. Positioning projects (£30-60K). Brand identity (£80-200K+).

Any Tier → Annual Refresh

Point-in-time audit creates demand for ongoing intelligence. Quarterly re-scans (£5-15K/year). The corpus stays alive. Retained relationship, not transaction.

Year 1 projections: Conservative (1 strategist) = £105K / Base (2) = £211K / Ambitious (3 + network) = £360K. All at ~76% margin.

Where the data comes from · How we know it’s good

Data.

Source Triangulation

30 / 37

No single-source claims survive to the synthesis.

What we do

— Every assertion cites a specific external source — Domain agents research independently — no agent sees another’s findings — Wave 2 identifies what is unsupported or contradicted — Live crawls validate or contradict pre-crawl research — Social listening checks brand-authored claims against unfiltered consumer voice — Council members react independently to the same research — All processing on GCP europe-west2 (London). No cross-border data transfers.

What we don’t do

— No self-generated claims. “We think” is not acceptable — “the data shows” is. — No confirmation bias in crawl design. The crawl brief includes explicit “disprove this assumption” targets. — No consensus manufacturing. Council divergences are preserved. — No scraping behind login walls without authorisation — No PII collection — we research brands, not individuals — Client data deletion within 30 days on request. GDPR compliant from infrastructure level.

The Future — Compounding Intelligence

31 / 37

Each completed audit generates the data to build the next one.

Q2 2026

V1 — Single-Audit RAG

Embed one audit’s corpus into Vertex AI Vector Search. Test semantic retrieval during Council sessions. The system retrieves the most relevant passages per Council question automatically.

Q3 2026

V2 — Cross-Audit Corpus

Connect 3-5 completed audits (anonymised). Cross-client pattern retrieval. How previous luxury brands handle the heritage/innovation tension becomes searchable institutional knowledge.

Q4 2026

V3 — Live Ingestion

New crawls and social listening feed directly into the vector DB. Delta analysis for re-scans. Quarterly re-scans become “ingest the diff” — not repeat the audit. Cost drops dramatically.

2027

V4 — Institutional Intel

Full corpus across all clients. Sector-level pattern libraries. Automated pre-research briefs from vector similarity. The same pattern recognition a 15-year veteran brings — made systematic and searchable.

The infrastructure is already provisioned on GCP. Vertex AI Vector Search, Gemini embeddings, Cloud Storage, BigQuery, Cloud Run. The gap is the embedding pipeline.

Proof Point 01 — Gaggenau

32 / 37

Ultra-luxury kitchen appliances. Full tier. 19 days.

2.6M

characters of brand intelligence. 200+ sources. 28 subreddits. 520 images analysed. 14 live web crawls.

competitors profiled across 13 dimensions each. Miele, Sub-Zero, BORA, La Cornue, Thermador, Wolf, V-ZUG, Smeg.

Council sessions. 5 voices per session. Progressive depth — each building on the last.

“There is a void in between, and that’s exactly where you’re going in.” — “I feel very confident. Looking very forward to work with you.”

Sven Baacke — Head of Product Design, Gaggenau · The meeting ran 30 minutes over its scheduled slot.

Proof Point 02 + Competitive Moat

33 / 37

Initio Parfums — Niche Luxury Fragrance

Different category. Same methodology.

— Direction locked: “Sacred Molecules” — 30-slide content master · 6 brand codes · 100+ files — Council v4 synthesis complete · Pitching 27 April — Same four phases, same six gates, adapted for fragrance — Two audits, two categories, one methodology

The Competitive Moat

No other studio has this engine.

— Wolff Olins, Pentagram, Landor — they run traditional audits — 150-500+ sources (systematic) vs 30-50 (manual) — Live social listening vs none or outsourced — 5 Council voices, stress-tested vs 1 strategist’s perspective — A living corpus that compounds vs a dead PDF — 12-18 month head start. Methodology + Council + institutional memory.

If it works for kitchen appliances and niche perfume, it works for anything we take on. No competitor can quickly replicate the system.

“It was a type of presentation that I’ve never seen us do before. It perfectly positioned us — real truth on it, then solutions.”

Andy Burden — Account Director, Made Thought

After the Gaggenau presentation. Before seeing the methodology behind it.

The Selling Language

35 / 37

Phrases that land. Use them.

For the team

“We codify the preparation so we can refuse to codify the judgment.” “It frees us up to do the rabbit hole thinking.” “You become the conductor. The system gives you more range.” “Scale desire — augment and scale what we do best, not automate.” “Break new ground.” — our AI philosophy.

For clients

“We read everything written about your brand and bring you what matters.” “Real truth, then solutions.” “It’s trained on our intellectual property — our methodology, our frameworks, our thinking.” “Using it for benefit, not for ease.” “Bestowed, not declared — we surface what the brand already is.”

Next Steps

36 / 37

From this deck to the first client proposal.

Working Session

Andy, Alice, and G sit together next week. Walk through the client, the context, what you want to achieve. We record it. That recording becomes the brief.

Context Ingestion

Send all relevant client materials. Brand guidelines, strategy decks, social plans, performance data. The system ingests everything. The more context, the sharper the module selection.

Module Selection + Proposal

Based on the ingested context, we select the relevant modules and build a proposal. Tier, timeline, deliverables, pricing. Tailored to what the client actually needs.

Costing Model

G and Andy productise the commercial articulation. How we present it. How we price it. How we frame it. Ready for any client conversation.