WhoKnew Views™ | Seasonality Detection (Hourly, Daily, Weekly)

Skip to main content
Table of Contents
Print

How Prophet AI discovers your store’s repeating traffic rhythms across hours, days, and weeks

Every store has a heartbeat. Traffic doesn’t arrive randomly—it follows patterns. Your 7pm rush. The Tuesday spike. The weekend surge. WhoKnew the Future™ uses multi-frequency seasonality analysis to detect and learn these rhythms automatically.


🧬 The Seasonality Component: s(t)

Remember Prophet’s formula? Seasonality is the s(t) component:

📐 PROPHET PREDICTIVE MODEL

ŷ(t) = g(t) + s(t) + m(t)

s(t) = Multi-Frequency Seasonality

Detects repeating patterns at multiple time scales—hourly rhythms within days, daily patterns within weeks, and weekly cycles within months. All learned automatically from your traffic.

Unlike simple averages, Prophet uses Fourier series to model smooth, wave-like seasonal patterns. This means it captures gradual transitions (like the slow 6pm→8pm traffic ramp) rather than treating each hour as isolated.


⏰ The Three Layers of Seasonality

Prophet analyzes your traffic at three frequency levels simultaneously:

🕐

Layer 1

Hourly Patterns

Which hours are busiest? Prophet learns that 7pm consistently beats 3am, that lunch hours spike, and that late-night traffic is minimal.

EXAMPLE HOURLY PATTERN

12am 6am 12pm 6pm 12am

📅

Layer 2

Daily Patterns (Day-of-Week)

Which days perform best? Prophet discovers that Saturdays crush Mondays, that Wednesdays are surprisingly strong, and that Sundays are browsing-heavy but conversion-light.

EXAMPLE WEEKLY PATTERN

Sun

Mon

Tue

Wed

Thu

Fri

Sat

🗓️

Layer 3

Extended Seasonality (Monthly/Yearly)

Which seasons dominate? With 90+ days of data, Prophet detects monthly trends. With 730+ days, it unlocks true annual seasonality—Black Friday patterns, summer slumps, and holiday surges.

EXAMPLE ANNUAL PATTERN

JanAprJulOctDec


📈 When Seasonality Features Unlock

Prophet can’t detect patterns it hasn’t seen. Here’s when each seasonality capability activates:

Data AgeSeasonality UnlockedStage
0-7 days❌ None — Learning basic trends1
7-14 days✅ Day-of-week patterns — First weekly cycle observed2
14-28 days✅ Multi-week validation — Weekly patterns confirmed3
28-90 days✅ Monthly trend modeling — Quarterly patterns emerge4
90-180 days✅ Extended seasonality + Holiday/event modeling5
180-365 days✅ Year-over-year comparison — Full annual cycle6
730+ days 🏆✅ Prophet-Grade Seasonality — Validated annual patterns (Meta standard)6+

💡 Why 730 Days?
Meta’s original Prophet research recommends at least 2 years of data to confidently detect true annual seasonality (vs. one-off events). At 730 days, WhoKnew the Future™ upgrades to “Prophet-Grade Intelligence” with validated seasonal patterns—the same standard used by Fortune 500 companies.


🔍 How Prophet Uses Seasonality

Once detected, seasonal patterns directly influence forecasts:

🧠 REAL-TIME FORECAST ADJUSTMENT

⏰

Hourly Adjustment

“It’s 7pm on this product. Historically, 7pm gets +45% more traffic than average. Boost forecast.”

📅

Day-of-Week Adjustment

“Tomorrow is Saturday. Saturdays average +32% higher traffic than weekdays. Increase prediction.”

🎄

Annual Adjustment

“It’s November. Last November saw +85% traffic spike for Black Friday. Factor into 30-day forecast.”


⚙️ Automatic vs Manual Pattern Detection

Unlike traditional analytics that require you to set up rules, Prophet’s seasonality detection is 100% automatic:

❌

Traditional Analytics

  • Manually define “peak hours”
  • Guess which days are busy
  • Set up seasonal rules
  • Constantly adjust settings
  • Miss patterns you didn’t configure

✅

WhoKnew the Future™

  • Learns patterns from data
  • Discovers hidden rhythms
  • Adapts as patterns change
  • Zero configuration required
  • Finds patterns you’d miss

🎯 Practical Examples

📦 What Prophet Might Discover About Your Store:

  • Hourly: “Traffic peaks at 8-9pm, dips 2-5am, and has a small lunch spike at 12-1pm”
  • Daily: “Saturday is your best day (+35%), Monday is worst (-22%), mid-week is average”
  • Weekly: “First week of month is slower, last week picks up (payday pattern?)”
  • Annual: “November-December are 2× normal, July-August are 40% below average”

All of these patterns are learned automatically from your traffic. You don’t need to tell Prophet anything—it figures it out from the data.


🏆 The Takeaway: Prophet’s seasonality detection means your forecasts get smarter every day. The longer you use WhoKnew Views™ Pro, the more accurately it predicts your traffic rhythms. At 730 days, you’ll have enterprise-grade seasonal intelligence matching Meta’s original Prophet standards.

Share