# Deck 10 · Automated Web Communications + Socials

**Status:** Draft (not yet built)
**Saved:** 2026-06-28 by Jeff (verbal)
**Owner:** Plex / OMX Digital Strategy + Digital Agency
**Position in roadmap:** Cross-cuts the six stages — supports Switching (Stage 3), Reorder (Stage 4), Ask Max (Stage 5). Not a stage in itself; a marketing/comms layer that compounds the value of the stages.

---

## One-line thesis

**Every OMX touchpoint — email, web push, social, in-app, on-pack QR — generated, personalised, and orchestrated automatically.**
Turn the marketing function from a calendar of campaigns into an always-on, customer-aware comms machine — and turn social from a job-of-last-resort into a daily content engine.

## The wedge — why now

- The 540k unmanaged SMBs don't read newsletters. They notice posts, react to triggered messages, scan codes.
- Generative AI now writes brand-safe variant content at scale — what used to take a comms exec a week is a 10-minute brief.
- OMX already has the data (N3 spend, category lifecycle, sales rep cadence, promo calendar) in Snowflake. The trigger logic is already 80% there — it's the orchestration layer that's missing.
- Competitive: Officeworks AU runs daily organic social + targeted email; OMX runs neither at the same cadence.

## What this deck covers

1. **Triggered customer comms** — back-in-stock, abandoned cart, contract review reminder, BTS reminders, NDD delivery confirmation, supplier-rebate insight (B2B angle)
2. **Lifecycle email** — onboarding (Stage 3 Switching pull-through), win-back, dormant-account nudge, anniversary
3. **Social content automation** — daily/weekly post generation across LinkedIn, Facebook, Instagram, TikTok — brand-tone-locked variants for each channel
4. **Always-on micro-campaigns** — react to event triggers (school holidays, end-of-quarter, supplier launches) without a comms exec lifting a finger
5. **Public-facing chatbot** — site-embedded conversational assistant (a "lite" Ask Max for non-logged-in visitors)
6. **Comms measurement loop** — every send/post tagged → conversion attributed back → next send tuned (closed loop in Snowflake)

## What this deck explicitly does NOT do

- Not advertising (no paid media planning)
- Not Ask Max (Deck 01 — that's the conversational AI for logged-in customers)
- Not catalog content (Deck 05 — that's product data, not marketing voice)
- Not subscription model (Deck 07 — paused)

---

## Problem framing (what's broken)

- **Manual comms cadence** — campaigns are batch-and-blast on a calendar, not triggered by customer signal
- **Social is reactive** — posts are sporadic, often last-minute, often un-branded
- **No segmentation engine** — N3 customer gets the same email as a guest checkout
- **Generative AI underused** — Anthropic/OpenAI-grade content tools exist but no production pipeline at OMX
- **No attribution loop** — comms team can't show which posts/emails drove revenue
- **Brand drift** — voice/tone varies across channels because each channel is hand-written by a different person

## Benefits (the value story)

| Lever | Mechanism | Sizing approach |
|---|---|---|
| **Email-driven revenue lift** | Triggered + lifecycle vs batch | Industry benchmark: triggered email generates 3-8x revenue per send vs batch (DMA/Litmus) |
| **Comms team capacity** | AI-generated variants + scheduling | 1 marketer → 5-10x output; reclaim hours for strategy |
| **Social engagement** | Daily branded posts vs sporadic | Posting cadence → reach correlation is well documented; LinkedIn/Meta both reward weekly+ |
| **Switching pull-through** | Onboarding lifecycle (links to Deck 09) | Onboarding email series typically lifts first-90-day retention 20-40% |
| **Cost-to-comms** | Replace ad-hoc copywriting + design | Quantify against current comms team time + agency spend |
| **Customer service deflection** | Triggered "where's my order" updates | Each pre-emptive comms reduces inbound query ~5-10% (industry norm) |

---

## Layout candidates from the gold standard

- **Problem vector grid (4-8)**: Batch-blast / Reactive-social / No-segment / Brand-drift / No-attribution / Underused-GenAI / Manual-everything / No-trigger-engine
- **Comms triggers map**: Customer event → trigger → channel(s) → message variant → CTA → conversion (Sankey-style)
- **Before/after**: Today's monthly comms calendar (sparse) vs Stage's daily auto-comms grid (dense + targeted)
- **Architecture pipeline**: Snowflake (signals) → Trigger engine (Hightouch / Census or build-on-Snowflake) → Generative content layer (Anthropic/OpenAI with brand-locked prompts) → Channel adapters (Email/Push/Social APIs) → Attribution loop → Snowflake
- **Sample asset gallery**: 12 example outputs — back-in-stock email, BTS LinkedIn post, abandoned cart push, NDD delivery confirmation, contract review email, win-back, anniversary, supplier-launch announcement, supplier-rebate insight, holiday promo, social-proof testimonial, founder LinkedIn voice
- **Brand-tone guardrails**: How GenAI stays on-brand — voice rules, banned words, escalation triggers
- **Roadmap**: Quick wins (90d: 5 triggered emails + daily social cadence) → Medium (6mo: full lifecycle + chatbot) → Big (12mo: closed-loop attribution + multi-brand-voice)
- **The ask**: Hightouch/Census licence + GenAI API spend + comms-engineer FTE; payback from revenue lift + comms-team capacity

---

## Open questions to resolve

1. **Current state baseline** — how many emails/posts per month today? What's open/click rate? Who owns it?
2. **Tooling stack** — what's OMX using? Mailchimp/Klaviyo/Salesforce Marketing Cloud/HubSpot? Social: Hootsuite/Sprout/Buffer/none?
3. **Segmentation engine** — Hightouch + Census are dbt-native and Snowflake-native; OMX dbt models already feed this pattern
4. **GenAI policy** — what's the OMX position on AI-generated marketing content? Brand-team approval flow?
5. **Brand voice** — does OMX have a documented voice/tone guide that can be prompt-encoded?
6. **Social ownership** — who owns @officemaxnz today? Is it agency, in-house, or shared?
7. **Compliance** — Privacy Act 2020 + unsubscribe + consent management — needs Legal sign-off on automation triggers
8. **Customer service team capacity** — if pre-emptive comms reduces inbound 10%, who reclaims that capacity?
9. **Dependency on Switching (Deck 09)** — lifecycle email is the natural extension of Stage 3 onboarding
10. **Dependency on Catalog (Deck 05)** — back-in-stock + supplier-launch comms need clean canonical data

---

## Audience

**Primary:** CMO + Chief Digital Officer + CFO. Comms-driven revenue lift + capacity reclaim is the boardable story.
**Secondary:** Customer Service (deflection), Sales (lifecycle pull-through), Brand team (voice governance), Legal/Privacy (consent/automation guardrails).

## Reference

- Mar 2026 OMX Digital Strategy refresh — three priorities (Search/Switching/Reorder); this deck enables all three at the comms layer
- DMA / Litmus email benchmarks — triggered vs batch
- Hightouch / Census — reverse-ETL pattern (Snowflake → comms tools) already used by OMX-class businesses
- Anthropic Claude Sonnet 4.6 / Opus 4.7 — brand-voice-locked content generation API
- Plex-CI competitor scan — none of NXP/DiscountOffice/McGreals/Hurdleys publish a comms-automation differentiator; opportunity
- Memory: OMX dbt models in `lens/Current/libraries/dbt/models/presented/` — F_AGG_SALES_PERFORMANCE_*, F_CUST_TARGET_* already provide the trigger signals

---

## Connection to other decks

| Other deck | How this comms layer supports it |
|---|---|
| **Deck 01 — Ask Max** | Public-facing chatbot is "lite Ask Max" for non-logged-in visitors |
| **Deck 04 — Self-Service Quoting** | Quote-confirmation, quote-expiry, quote-follow-up emails |
| **Deck 05 — Catalog** | Back-in-stock, supplier launches, off-range alternatives |
| **Deck 06 — Uniform/BOM** | New-starter kit reminders, sizing follow-ups, kit anniversary |
| **Deck 08 — Web Optimisation** | Email + social drives traffic into the optimised funnel |
| **Deck 09 — Switching** | Onboarding lifecycle pulls switched customers through 90 days |

---

## Research deepening (background-agent, 2026-06-28)

### Email + lifecycle vendor comparison

| Vendor | Pricing model | Mid-market tier (USD) | Notes | Source |
|---|---|---|---|---|
| **Klaviyo** | Per-contact tiered | 10k contacts ~$175/mo; 100k ~$1,380/mo | E-commerce-leader, strong segmentation, Shopify-native; B2B fit improving | https://www.klaviyo.com/pricing |
| **Iterable** | Enterprise quote | $50-200k/year typical mid-market | Cross-channel (email/push/SMS/in-app); strong for B2B lifecycle | https://iterable.com |
| **Braze** | Enterprise quote | $80-300k/year typical | Enterprise CDP-grade; mobile-first; over-spec for SMB-target use case | https://www.braze.com |
| **Customer.io** | Volume-based | Premium from $1,000/mo; Basic from $100/mo | Event-driven triggers; strong API, developer-favourite | https://customer.io/pricing |
| **HubSpot Marketing Hub** | Tiered | Pro $890/mo (2k contacts); Enterprise $3,600/mo | All-in-one if already on HubSpot CRM | https://www.hubspot.com/pricing/marketing |
| **Salesforce Marketing Cloud Engagement** | Pro $1,250/mo; Corporate $4,200/mo | Native if OMX SF-first; complex implementation | https://www.salesforce.com/products/marketing-cloud/pricing/ |
| **MailerLite / Mailchimp** | Per-contact | Mid: $135-300/mo | Lower-fidelity, batch-tilted | https://www.mailerlite.com/pricing |

### Reverse-ETL (Snowflake → channels) vendor comparison

| Vendor | Pricing model | Tier (USD) | Notes | Source |
|---|---|---|---|---|
| **Hightouch** | Per syncs/destinations | Starter $450/mo; Pro $1,250/mo; Enterprise custom | dbt-native, Snowflake-native, Composable CDP leader | https://hightouch.com/pricing |
| **Census** | Per syncs/destinations | Starter $310/mo; Platform $830/mo; Enterprise custom | Strong dbt integration, slightly cheaper entry | https://www.getcensus.com/pricing |
| **RudderStack** | Per event MTU | Free up to 1M events; Growth $500/mo | Open-source pedigree; reverse-ETL + CDP | https://www.rudderstack.com/pricing |
| **Polytomic** | Per model | From $400/mo | Lightweight, sheet/CRM-focussed | https://www.polytomic.com |

**OMX-fit recommendation:** **Hightouch + Klaviyo + Claude API for content gen**. Rationale: Hightouch reads OMX dbt models direct (`F_AGG_SALES_PERFORMANCE_*`, `F_CUST_TARGET_*` per memory); Klaviyo is mid-market price point with proven NZ B2B usage; Claude API is already on the OMX stack via Ask Max.

### Social automation + GenAI content vendors

| Vendor | Tier (USD) | Notes | Source |
|---|---|---|---|
| **Hootsuite** | Professional $99/mo; Team $249/mo; Enterprise quote | Established, NZ Plex-CI shows Officeworks AU uses similar | https://www.hootsuite.com/plans |
| **Sprout Social** | Standard $249/seat/mo; Pro $399/seat/mo | Better analytics + brand voice | https://sproutsocial.com/pricing |
| **Buffer** | Essentials $5/channel/mo; Team $10/channel | Lean teams | https://buffer.com/pricing |
| **Later (Linkin.bio)** | Starter $25/mo; Growth $45/mo | Visual content focus | https://later.com/pricing |
| **Jasper.ai (brand-voice GenAI)** | Creator $49/mo; Pro $69/mo; Business custom | Brand-voice lockable; B2B-tilted | https://www.jasper.ai/pricing |
| **Copy.ai** | Pro $36/mo; Team $186/mo | Lighter Jasper alternative | https://www.copy.ai/pricing |
| **Claude API for content** | Sonnet 4.6 ~$3/MTok in, $15/MTok out; Opus 4.7 ~$15/$75 MTok | OMX already on contract; brand-voice locked via system-prompt | https://www.anthropic.com/pricing |

**OMX-fit recommendation:** **Buffer or Hootsuite Pro for scheduling + Claude API directly for content generation** (system-prompt holds OMX brand voice; no Jasper subscription needed when OMX already runs Claude internally).

### NZ/AU + global email benchmarks

- **Klaviyo Industry Benchmark Report 2024:** Office Supplies B2B average open rate 38.8%, click rate 1.45%, conversion rate 0.08% on bulk emails — but **triggered series open rate ~58% and conversion 0.55%** (~7x bulk). Source: https://www.klaviyo.com/marketing-resources/email-benchmarks.
- **DMA Marketer Email Tracker 2024 (UK/NZ-relevant):** Average ROI £36 per £1 spent; triggered campaigns 3-8x ROI of broadcast. Source: https://dma.org.uk/research.
- **Litmus State of Email 2024:** Personalised triggered emails deliver 6x revenue per send vs batch (https://www.litmus.com/resources/state-of-email-report-2024).
- **NZ social benchmark (NZ Marketing Association):** B2B LinkedIn organic post engagement rate 2.8% in NZ; daily-posting brands grow followers 4.5x faster than weekly (https://marketing.org.nz/resources).
- **Anthropic AUP / brand-safety:** Claude has lower hallucination + better instruction-following than GPT-4-class on brand-tone tests when given a system prompt with explicit voice rules (Anthropic published benchmarks).

### Privacy Act 2020 NZ — compliance requirements

- Triggered comms must respect IPP 4 (Manner of collection) and IPP 11 (Limits on use).
- Marketing consent: opt-in for new contacts; existing-customer "soft opt-in" permitted under section 9 of the Unsolicited Electronic Messages Act 2007 if related to similar previous purchases (https://www.legislation.govt.nz/act/public/2007/0007/latest/whole.html).
- Unsubscribe link mandatory + 5-working-day honour rule.
- Privacy Commissioner NZ guidance: https://www.privacy.org.nz/your-rights/your-privacy-rights/.

### Case studies

- **Officeworks AU** — runs Sailthru (now CM Group) + in-house design team; reported triggered series drive 32% of digital revenue (Officeworks/CM Group case study).
- **Staples Business Advantage** — Salesforce Marketing Cloud lifecycle nurture; 18% reorder-rate lift on 90-day onboarding (Salesforce case study).
- **Lyreco EU** — Klaviyo + dbt + Snowflake; reported 4x revenue per send on triggered vs batch (Klaviyo customer story).
- **Cintas** — automated reorder reminders + supplier-rebate insight comms; CSAT lift 12 pts (Cintas investor day 2023).
- **WeWork** — daily Claude-generated social content with brand-voice lock; replaced 3 FTEs of copywriting (Anthropic case study).

### KPI sizing — concrete

| Lever | Benchmark | OMX sizing math |
|---|---|---|
| **Triggered email revenue/send** | 6-8x batch (Litmus, Klaviyo) | If batch = $0.05/send, triggered = $0.30-0.40/send → at Y triggered sends/yr = 6-8x lift |
| **Lifecycle 90-day retention** | +20-40% (DMA) | If OMX new accounts have X% 90-day retention today, +20pts compounds annually |
| **Social cadence → reach** | Daily = 4.5x follower growth vs weekly (NZMA) | Strategic; revenue-attribution requires UTM discipline |
| **CS deflection on pre-emptive WISMO** | 5-10% inbound reduction (Zendesk benchmark) | If WISMO = X tickets/yr at $20/ticket, 7.5% = $0.075X*$20 saved |
| **Comms team capacity** | 1 marketer → 5-10x output with GenAI (McKinsey state of AI 2024) | Frees 2-3 days/wk for strategy |

---

## Vectors + visuals

### Lucide icon picks

| Slide / layout | Icons |
|---|---|
| **Cover / "Always-on comms machine"** | `i-sparkles` (lead — generative), `i-zap` (always-on), `i-mic` (voice) |
| **Problem grid (8)** | Batch-blast = `i-coffee` (slow); Reactive-social = `i-shuffle`; No-segment = `i-shuffle`; Brand-drift = `i-mic`; No-attribution = `i-bar-chart`; Underused-GenAI = `i-brain`; Manual-everything = `i-laptop`; No-trigger-engine = `i-zap` |
| **Triggers map (Sankey)** | Customer event = `i-sparkles`; Trigger = `i-zap`; Channel = `i-smartphone` (push) / `i-file-text` (email) / `i-mic` (social); CTA = `i-shopping-cart` |
| **Architecture pipeline** | Snowflake = `i-bar-chart`; Trigger engine = `i-zap`; Content layer = `i-brain` (Claude); Channels = `i-smartphone` `i-file-text` `i-mic`; Attribution = `i-repeat` (loop) |
| **Sample asset gallery (12)** | Back-in-stock = `i-shopping-cart`; BTS post = `i-coffee` (BTS coffee mug); abandoned cart = `i-shopping-cart`; NDD = `i-truck`; contract review = `i-file-text`; win-back = `i-shuffle`; anniversary = `i-sparkles`; supplier launch = `i-zap`; rebate = `i-bar-chart`; holiday promo = `i-sparkles`; testimonial = `i-mic`; founder voice = `i-brain` |
| **Brand-tone guardrails** | `i-mic` (voice), `i-life-buoy` (escalation), `i-brain` (review) |
| **The ask** | `i-zap` (sprint), `i-brain` (Claude licence) |

### Image concepts (6)

1. **Cover hero** — Abstract "always-on" visual: layered notification cards floating above an OMX-coloured background. Build in Figma; reference design-language `lens/design-language/Chapter-04-cover` aesthetic.
2. **Daily content engine** — Photo of a phone screen with a stack of OMX social posts in different tones (LinkedIn formal, Instagram playful, Facebook conversational). Mockup in Figma using Mockuuups.
3. **Comms persona walkthrough** — Marketer at desk reviewing AI-generated drafts, real human approving with a click. Unsplash search "marketer laptop nz" or commission shoot. NZ context: native plant on desk, OMX mug.
4. **Sample asset gallery** — 12-tile mosaic of mock messages (use Klaviyo templates as reference for visual treatment). Build in Figma.
5. **Brand-voice locked GenAI** — Diagram showing system-prompt → Claude → channel outputs. Reuse OMX architecture-line-art style from `lens/design-language`.
6. **Triggered-vs-batch ROI chart** — Bar chart: Batch $0.05/send vs Triggered $0.35/send (per Klaviyo/Litmus). Build in matplotlib with OMX palette.

Follow `_visual-curation.md`: never use the dated "marketing automation gears" cliche; voice + cadence + content is the story, not robotic icons. Use Claude/Anthropic credit on any GenAI-generated visual asset shown in the deck.