Strategy

The architecture says how we publish. This page says what wins and why — six lenses: organic · viral · when · grow · pay off · moral. Everything here is governed by the drive (cash flow first) and the brand rules. If a tactic contradicts either, the tactic loses.

The multiplier — create once, repurpose everywhere

The binding constraint is not channel count. It is content capacity: Samia carries 2–3 core pieces a week, Abdout adds occasional technical posts. Five brands × eight channels is impossible if every post is bespoke — and trivial if it isn't:

            1 core piece (story-first, Arabic-first)
                             │
              Claude adapts per platform, per brand
   ┌──────┬──────┬───────┬───────┬────────┬────────┬──────────┐
   ▼      ▼      ▼       ▼       ▼        ▼        ▼          ▼
 IG Reel TikTok FB post X clip LinkedIn Telegram WhatsApp   blog/SEO
 (9:16)  (hook) (parent) (dev)  (B2B)   (owned)  (broadcast) (long-form)

Claude rewrites the hook, length, caption, and hashtags per platform; /higgs re-ratios the media. ~3 core pieces/week → ~15–20 platform-native posts. This is what the automation is for — the pipeline's entire payoff concentrates here.

Rule: never write platform-first. Write story-first, then adapt.

Dormant during the slice (2026-08-06). The multiplier is the payoff of breadth, and breadth is exactly what the one-slice decision defers: hogwarts × Facebook × Sudan until the loop is proven. With one channel live there is nothing to fan out to, so a core piece is the post — which is also why /draft's queue mode writes one portable core piece rather than a thin fan-out. Read the rest of this page as the strategy the slice is proving toward, not the one it is running. Two of its priorities invert because the country is Sudan rather than Saudi — see the notes on format and season below.

Organic — the default is $0

• No ad spend by default. The billing doctrine (subscription-only, lean burn) applies to marketing too. Any paid experiment — boosts, ads, aggregator SaaS — is a billing change: /decide + captain.
• Organic reach is not equal across platforms. 2026 directional benchmarks (vendor studies, not platform figures — treat as ranges):

• Implication — formats: the strategy is short vertical video + carousels, not text posts into the void. Only TikTok (and Reels, to a degree) give a small account genuine follower-independent discovery in 2026 — they get the creative energy.
  • Inverted for the Sudan slice (2026-08-06). The reach numbers above are Saudi/Gulf. For a Khartoum audience the Sudan reality governs — intermittent connectivity favors text + image over video — so the template lane's static Arabic cards lead, and they are also the only lane that is free and unlimited. Video keeps its place in this table for the channels the slice is proving toward.
• Implication — LinkedIn: Ali's personal profile is the LinkedIn strategy; the company page is the storefront his posts link back to. Employee advocacy beats page posting.
• Owned beats rented. Telegram channels and WhatsApp broadcast lists are algorithm-proof: 100% of subscribers see the message, and Telegram is free to automate. Every rented-reach post should CTA toward an owned channel. Owned lists are the asset; follower counts are weather.
• Borrowed is a third thing, and it is spendable. The Telegram and WhatsApp groups other people run — school admins, landlords, MENA edtech — are where prospects already are, but we post there as guests: an admin can eject us, and the room closes for good. Rented reach recovers from a weak post; borrowed reach does not. Post there rarely, in the room's own register, and measure per room. Model and guardrails: Groups & Destinations — planned, and ranked last on purpose.

Viral — engineered luck

Virality is a lottery. The strategy is to buy many cheap tickets (the multiplier gives us volume at near-zero marginal cost), never to bet the plan on winning one.

What raises the odds per ticket:

Viral shape per audience:

• Dev-rel (databayt): build-in-public — show the machine working. "Watch an agent ship a feature", real terminal, real diffs. The open-source story is inherently shareable.
• Consumer (Mkan): before/after and beautiful property tours — visual envy travels.
• Schools (Hogwarts): parent-shareable moments and admin time-savings — with consent, always (see moral).
• Sudan SMEs (Sijillee): practical bookkeeping tips a shopkeeper forwards to another shopkeeper — WhatsApp-forwardable is the viral unit there, not the algorithm.

What we refuse for reach: rage-bait, fake controversy, trend-jacking tragedy. → moral.

When — rhythm and seasons

Weekly rhythm (the honest week — capacity-true, per brand):

Total: ~4 core pieces/week — at the edge of capacity, which is correct. If a week is tight, brands drop in reverse-payoff order (Sijillee → Mkan → databayt; Hogwarts posts last to die, because it feeds live sales).

Daily timing (MENA heuristics — regional studies, not guarantees):

• Evenings dominate — the band is 19:00–24:00 local, strongest 21:00–23:00 for KSA audiences.
• The week peaks Thursday–Saturday; the weekend is Friday–Saturday in both Saudi Arabia and Sudan, and Thursday evening behaves like the Western Friday evening.
• Friday midday is a dead zone (Jumu'ah) — schedule around prayer times for anything live.
• Sudan has no reliable timing benchmarks right now (connectivity disruption) — apply the same evening pattern as an assumption and favor WhatsApp/Facebook delivery.

Seasons (plan calendars around them):

• Ramadan (≈ Feb 7 – Mar 8, 2027) — media use rises ~30% and goes nocturnal: engagement jumps ~40% after iftar, peaking post-iftar (20:00–24:00), post-Taraweeh, and pre-suhoor (02:00–06:00). Flip the calendar nocturnal for the month; soften sales tone, raise community/giving tone.
• Eid (both) — greeting + campaign windows.
• Saudi school year 2026–27 — starts Aug 23, 2026, mid-year break from Jan 8, 2027 (dates vary by school). Hogwarts pushes: back-to-school Jul–mid-Aug, re-enrollment late Dec–early Jan — the windows when administrators feel the pain we solve.
  • Not the clock during the slice (2026-08-06). The slice is Sudan-first, so the Aug 23 Saudi start does not set its pace — and that is deliberate: a proving run should not be raced against a marketing deadline. The window returns as a driver when the slice widens to Saudi.
• Exam seasons — Moallimee's future prime time.
• Sudan reality — connectivity is intermittent; Sijillee cadence favors lightweight formats (text + image over video) and WhatsApp/Telegram delivery.

When does it pay off — timeline honesty:

Grow — loops, not spikes

Stages:

1. 0 → 100 (seed): manual and personal — Ali's LinkedIn network, Sedon's Saudi connections, Abdout's dev circles, team members sharing. No automation needed; credibility is hand-made.
2. 100 → 1,000 (traction): consistency + series formats + collaborations (guest posts, MENA tech communities, education groups).
3. 1,000+ (scale): community — groups where we answer questions beat audiences who scroll past.

Loops (each output feeds the next input):

• Dev-rel flywheel: blog/docs (SEO) → social clips → GitHub stars → contributors → their work becomes content → blog. The open-source story compounds; ads never do.
• Umbrella amplification: the databayt account boosts every product launch; product wins feed the databayt story back.
• Rented → owned: every platform post CTAs to Telegram/WhatsApp lists; the list is the moat.
• Social ↔ sales: Ali's outreach cites the content; conversations from outreach become case-study content.

First milestone stays growth.md's: 500 followers across platforms — then per-brand targets get set from real data, not guesses.

Pay off — the accounting

Social serves the drive: cash flow first. It supports sales; it rarely closes B2B alone. We run it like an investment with a thesis, checkpoints, and a stop-loss:

• Payoff map (fastest revenue-linkage first):
  1. Hogwarts — social is a trust asset for Ali's sales calls. A school checks the page before signing; an active, calm, Arabic-first presence closes doubt. Direct linkage to the North Star (active paying schools).
  2. Mkan — direct consumer funnel; fastest feedback on whether content converts.
  3. databayt — dev-rel pipeline: contributors, talent, partners. Long-tail, high compounding.
  4. Sijillee — market education in Sudan; slow, cheap, trust-first.
• Measurement: UTM on every link → PostHog → CRM pipeline → North Star. If a post can't carry a UTM (IG bio limits, WhatsApp), count DMs and mentions manually in the weekly review.
• Kill criteria (the stop-loss): a channel earns its slot. Zero signal after 3 months of consistent posting → drop to monthly maintenance or park it. Quarterly review in the captain's cadence. Effort is a cost; zombie channels pay negative interest in focus.
• What we don't measure by: follower count alone. Lists, pipeline mentions, and inbound DMs are the real scoreboard.

Moral — the code

We sell trust — to schools handling children's data, to businesses handling their books, to developers reading our source. The brand is the moral posture. This section is not PR; it is operating law, rooted in the same share-economy ethos as the company.

1. Truthful. Real metrics, real screenshots, real testimonials or none. No overclaiming: we "automate the boring", we don't "replace your staff". If a number is projected, say so.
2. AI transparency. Label AI media where required — and default to labeling even where not. The 2026 duties: TikTok mandates labeling realistic AIGC and exposes an is_aigc flag in its posting API (unlabeled realistic AI content risks removal); Meta requires self-disclosure for photorealistic video / realistic audio and auto-labels "AI Info" from C2PA/IPTC metadata; X, LinkedIn, and Snapchat impose no duty on ordinary marketing images (Snapchat: monetized content only). Beyond compliance: never pass AI imagery as reality — no "customer photo" that never existed, no AI "property shot" of a real listing. Mkan listings use real photos; AI is for brand/mood visuals, and our text-free charcoal/cyan aesthetic is deliberately non-photoreal.
3. Children and privacy. Hogwarts hard rule: no student faces without written consent; blur by default. Saudi PDPL and regulator expectations (ADEK) treat student data as radioactive — so do we, in marketing as in product.
4. Cultural respect. Modest imagery. Correct Arabic — the no-broken-Arabic rule applies to copy as much as renders. Ramadan sensitivity. And for Sudan: humility in tone — never exploit crisis for engagement. A bookkeeping app for shopkeepers rebuilding their business markets itself with respect or not at all.
5. No dark growth. No bought followers, no engagement pods, no spam DMs, no misleading hashtags, no unofficial APIs that violate platform ToS. Growth that needs deception is debt.
6. Accountability. A named human owns every account. Full automation (L4) ships only with the moral gate: an automated judge that checks truthfulness, Arabic correctness, cultural fit, AI-labeling, and consent before anything publishes — plus rate limits and a kill switch. The autonomy ladder is a moral structure, not just a technical one.

See also

• Social Automation — the pipeline and architecture.
• Brand & Voice — palette, tone, the no-AI-text rule.
• Brand pages: Databayt · Hogwarts · Mkan · Moallimee · Sijillee