A restaurant marketing framework built from live client work — seven modules that compose into a 90-day operating cadence. Every step ties back to a seated, paying cover.
Strategic beliefs about how F&B growth actually works — different from the operational principles in Approach. Each one is backed by data across the TWS portfolio.
Follower count and reach are correlated at r = 0.405 across the TWS portfolio — barely more than half of one axis explains the other. Growth doesn't come from adding followers; it comes from action rate on borrowed audiences.
r = 0.405The Stack is five layers deep. Working from Layer 1 to 5 wastes cycles if Layer 3 is the real block. Diagnose first, then sequence.
5-layer stack86 creator posts across the TWS portfolio produced 39.1% of all Instagram engagement. A 10k F&B-native creator outdraws a 300k lifestyle creator on covers-per-post. Silence between posts costs momentum faster than a bad post does.
39.1% of engagementSubstance quality beats frequency, and it isn't close. The pace that scales is 3× a week, cast tight, briefed for format.
3× per weekEvery TWS engagement is a 90-day arc. Named owners. Weekly cadence. The compound interest of running the same loop.
90-day arcEvery engagement composes these seven modules into a single 90-day operating rhythm. Modules can start at any layer — we diagnose the lowest broken one first.
Before anything ships, we reset how you measure growth.
Highest effort-to-return ratio in the course; most operators see movement within three weeks.
The highest-leverage reach in F&B. Most brands cast for followers, not fit — and post once when they should be posting six.
Own content closes borrowed traffic — it doesn't generate reach. Three a week beats seven mediocre.
Most operators arrive wanting to spend more and leave spending less, on fewer things, tagged properly.
Where the method earns credibility — by teaching the limits of its own metric rather than selling a dashboard.
Ninety days, in order, with a named owner for every recurring task.
Creator marketing is the single highest-leverage lever in F&B — and the one most operators run backwards. They pick by follower count, post once, and wonder why nothing moved. The two things that actually decide outcomes are relevance and cadence. Both are engineerable. Here's how we run it on every engagement.
A 10k F&B-native creator whose feed is food, hawkers, and neighbourhood cafés outdraws a 300k lifestyle creator whose audience is fashion, travel, and gym. Category fit is the first filter. Follower count is a tiebreaker, not a criterion.
Fit > reachOne post lands and evaporates. Six posts in a rolling six-week rhythm compound. Gaps longer than three weeks cost roughly 40% of the momentum you built. We plan the cadence before we plan the casting.
6 in 6 weeksCategory fit → format literacy → audience overlap with your neighbourhood → delivery reliability → follower count. Reverse this order and you buy reach that doesn't convert to covers.
5 filtersNano (5–25k, F&B-native) · Micro (25–100k, primary vertical F&B) · Anchor (100k+, city-scale). Portfolio target ~70/25/5 — the inverse of what most operators default to.
70 / 25 / 5Brief → shoot → whitelisting → paid amplification (M04). Named owner per stage. Skip whitelisting and you forfeit the compounding half of the spend. Miss the brief and delivery drifts off format.
4 stages86 creator posts across the TWS portfolio = 39.1% of all IG engagement. Wrong-category anchors underperform right-category nanos on covers-per-post. Cadence gaps ≥3 weeks halve the next post's reach.
TWS portfolio · 2024–26Client-safe framing of the patterns we most often walk into. All observed in live TWS engagements.
Anonymised patterns from live TWS engagements. Every shape composes the same seven modules — the difference is the starting layer.
One arc, one calendar — modules layer in and stay active rather than running one after another.
Every engagement runs this rhythm. Weekly loops. Named owners. Ledger every Sunday.
A two-week audit — honest findings, specific recommendations, PSG-eligible.