# Building Ourselves — The Story So Far, and What's Next

**Audience:** ExCo, CFO, Commercial Director, Operations Director
**Saved:** 2026-06-29 by Plex / Jeff
**Status:** Straw man for review

---

## 1. The proof — we've already done it

OMX has built two production-grade internal platforms in the last 18 months. Neither was a packaged SaaS. Both are live, both are paid for, both work.

### Lens — replaces Sisense
- **What it is:** OMX's own BI tool. Same Snowflake, same dbt models, same data — better UX, better fit, no vendor licence.
- **Status:** 181 imported Sisense reports stay as the regression test set; new build is on the forward-state design language (`lens\design-language\`).
- **Why it matters:** We took an industry SaaS, asked it to fit OMX, found it couldn't, and shipped our own.
- **Lesson:** When the data is in Snowflake and the question is OMX-shaped, the fit-for-OMX build wins on cost, time-to-value, and team capability.

### WhoOnSite — frontline identity, NFC + Gallagher
- **What it is:** OMX-built worker identity stack. NFC tap, Gallagher integration, real-time on-site presence.
- **Status:** In production. Listed as a confirmed input to multiple downstream decks (DC Optimisation, Mobile Workforce identity, Desk Booking).
- **Why it matters:** We solved a frontline-people problem with internal engineering — not a global HRIS vendor.
- **Lesson:** When the workflow is OMX-specific (DC frontline + AD-only workers without SF licences), fit-for-OMX is the only honest answer.

**These two prove the pattern.** Snowflake-native data + thin web UI on the Lens design language + Postgres for state + Anthropic-grade AI where it earns its keep = OMX can ship internal tools that beat the packaged-SaaS alternative on every dimension that matters.

---

## 2. Where we are — the 25-deck library tonight

The pitch deck library at `pitchdecks.plex.nz` now holds **25 forward-bet ideas**, all aligned to one north star: **serving the unmanaged 540k NZ small businesses, plus running OMX better at the same time.**

The library is organised into five stacks. Each stack pays off differently.

### Stack A — Direct savings (~$710k/yr baseline)

These ideas reclaim spend from packaged SaaS or sales-team time. The numbers are conservative and confirmed.

| # | Deck | Reclaim / yr |
|---|---|---|
| 02 | PPSS — Snowflake-native promo + rebate + demand | **$350k** (Enable $120k + Anaplan $200k + storage $30k) |
| 14 | IBP Full Programme — replaces Anaplan, adds Supply + ASN + Put-away | **$230k** Year-1 (subset of PPSS; sequencing matters) |
| 23 | Contract Vault — lifecycle management | **$70k** |
| 24 | Desk Booking | **$30k** |
| 25 | Project Management (EPMO tier-aware) | **$30k** |
| 22 | Territory Management (replaces walked-away SF custom code) | **$25k+** (sunk) — plus future change cost avoided |

**Note:** PPSS + IBP overlap on the Anaplan reclaim (counted once). Total clean baseline ~**$510-710k/yr** depending on sequencing; the high end assumes Vault + Desk + PM ship in parallel.

### Stack B — Strategic revenue + experience (multi-year)

These ideas don't reclaim spend; they grow the top line and the brand. Multi-year horizon, multi-deck dependencies.

| # | Deck | Outcome |
|---|---|---|
| 01 | Ask Max (live, walkable) | Conversation as the next interface for the 540k |
| 04 | Self-Service Quoting + BTS-27 | Quote turnaround days → <1hr |
| 05 | Global Catalog (PIM) | Catalog as a competitive moat |
| 08 | Stage 2 — Web Optimisation | Search + UX + conversion lift |
| 09 | Stage 3 — Switching | "Match my old supplier" → first order |
| 10 | Automated Comms + Socials | Triggered + lifecycle + GenAI social |
| 15 | Curated Marketplace (Farmlands-shape) | Brand-extension without inventory; 7x Farmlands base addressable |
| 16 | Product Data Enhancement (PDX → Snowflake) | $1.8-4.8M off-range margin recovery |
| 20 | Product Matching (cost / sub / environmental) | Win-win switching + sustainability |

### Stack C — Operations + innovation

These ideas modernise the operating model — many headed for Innovation Garage Oct 2026 demo.

| # | Deck | Status |
|---|---|---|
| 03 | DC Optimisation (IoT + Robotics + Twin) | $750k saving modelled |
| 11 | Direct Ship Supplier Visibility (QR + photo) | **Garage Oct 2026** target |
| 12 | Local Owner-Driver (high-volume zones) | $110k/yr per zone × 6 zones = $660k |

### Stack D — Commercial + intelligence

These ideas codify what we already do informally — turn artisan into engine.

| # | Deck | Outcome |
|---|---|---|
| 13 | Competitor Intelligence (Plex-CI productised) | $625k/yr margin recapture potential |
| 17 | AI Semantic Conversations (search-to-sales) | $900k/yr QBR capacity reclaim |
| 18 | Company Research + QBR (Briefing Room) | Same engine, two surfaces |
| 19 | RFP Response Platform | 7-12% win-rate lift industry benchmark |
| 21 | Strategic Deep Dives | Codify the standing dive types |

### Stack E — Internal capability

These ideas back the pattern — fit-for-OMX tools for OMX.

| # | Deck | Status |
|---|---|---|
| 06 | Uniform & BOM | Kit builder + sizing + logo |
| 07 | Subscriptions | Paused; angle clarification needed |
| 22 | Territory Management | Replaces walked-away SF code |
| 23 | Contract Vault | $70k/yr |
| 24 | Desk Booking | $30k/yr |
| 25 | Project Management | $30k/yr |

---

## 3. The roadmap — sequenced for proof + payback

The roadmap follows three rules:
1. **Ship the small things first to prove the pattern compounds.** Internal tools (Territory, Vault, Desk, PM) are 3-6 months each; they pay back fast and build muscle.
2. **Land the big things on top of proven foundations.** PPSS + IBP are 12-24 month bets — but only feasible because Lens + WhoOnSite proved we can ship.
3. **Garage every quarter.** Innovation Garage is the public commitment device — DS Visibility Oct, then the next.

### Phase 1 — Q3 2026 to year-end (ship + prove)
- **Internal tools sweep:** Territory · Vault · Desk · PM — all four shipping in parallel; design language reuse, Postgres + thin UI pattern
- **PPSS WS1** in flight (already cased in Deck 02) — research → POC
- **DS Visibility** — Garage Oct demo (Deck 11); hybrid Option A feed + Option B sticker
- **Web Optimisation** Sprint 1 — search + checkout quick wins (Deck 08)

**By year-end:** ~$160k of savings landed (Vault + Desk + PM + Territory baseline); two Garage demos in market; Web Opt baseline measured.

### Phase 2 — H1 2027 (data foundations + revenue)
- **IBP WS2 Demand Planning** — replace Anaplan core workflow
- **Catalog + PDX (Deck 05 + 16)** — data quality programme + platform
- **Switching (Deck 09)** — Stage 3 lands on top of Web Opt
- **Automated Comms (Deck 10)** — triggered + lifecycle + social
- **Plex-CI productisation (Deck 13)** — channels unblock + internal use cases

**By H1 2027 end:** PPSS + Demand reclaim banked (~$350k); first switching cohorts measured; comms running daily.

### Phase 3 — H2 2027 to 2028 (closed-loop + revenue compounding)
- **IBP WS3 Supply + ASN + Auto Put-away** — closed-loop demand→supply→DC
- **AI Semantic Conversations (Deck 17)** — internal first; QBR pre-fill = $900k/yr capacity
- **Product Matching (Deck 20)** — internal 3-use-case; feeds Switching + RFP + Catalog
- **RFP Platform (Deck 19)** — central knowledge base on Vault
- **Briefing Room + QBR (Deck 18)** + **Strategic Deep Dives (Deck 21)** — codified engines
- **Owner-Driver (Deck 12)** — AKL CBD pilot then 6-zone rollout

**By 2028 end:** Full IBP closed loop; ~$1.5M+/yr of confirmed savings + recapture; commercial + intelligence engines codified.

### Phase 4 — 2029+ (long horizon)
- **Marketplace + curated partners (Deck 15)** — Farmlands-shape; OMX as the platform for the unmanaged 540k
- **Marketplace credit facility** — Funding A or B
- **Ask Max as the front door** to the whole platform — same engine, every surface

**By 2030:** OMX is the trusted hub for the unmanaged 540k. Brand-extension without inventory expansion. The Farmlands of small business.

---

## 4. The case to commit

1. **The proof is on the floor.** Lens runs. WhoOnSite runs. The pattern is no longer theoretical.
2. **The savings pay for the build.** ~$700k+/yr of confirmed reclaims funds the engineering capacity that ships the rest.
3. **Vendor lock-in is real.** Anaplan, Enable, Sisense, Robin, Asana, Jira, Salesforce-custom — every one of these tells OMX what to do next. Building ourselves keeps the roadmap in OMX hands.
4. **Capability compounds.** Every internal tool the team ships makes the next one cheaper. Lens enabled Ask Max. Ask Max enables Semantic Conversations. Semantic Conversations enables QBR pre-fill. The library is not 25 independent bets; it's one capability with 25 surfaces.
5. **The unmanaged 540k won't serve themselves.** Somebody is going to build the platform for NZ small business. Our choice is whether that's us — leveraging the data, brand, and infrastructure we already own — or someone else.

---

## 5. What we need from ExCo

| Decision | Owner | Window |
|---|---|---|
| Endorse the build-ourselves pattern as the default | CEO + ExCo | This quarter |
| Approve Phase 1 envelope (internal tools sweep + PPSS + DS Visibility + Web Opt Sprint 1) | CFO + COO | This quarter |
| Confirm Innovation Garage Oct demo set (DS Visibility + 1-2 more) | Commercial + Strategy | August 2026 |
| Resource the EPMO Project Management tool first — every other deck routes through it | EPMO + COO | This quarter |
| Sign the Anaplan exit plan (contractual lock-in review) | CFO + Commercial | Q4 2026 |
| Funding model decision for Marketplace credit (Deck 15 — A vs B) | CFO + Board | 2027 |

---

## Closing line

**We are not a SaaS shop. We are an office products business that has learned to build software when no off-the-shelf product fits. Lens proved it. WhoOnSite proved it. The next 23 ideas in this library are the same pattern, applied at scale.** The choice is whether we sequence them with intent — or watch a competitor do it first.