Lookback window

The lookback window defines the time period that a connector will retrieve data for during each synchronization, relative to the reference date.

Why use a lookback window?

Data on source platforms (Google Ads, Meta, etc.) is not always immediately finalized. Several factors can cause retroactive changes:

• Data processing delays: metrics may be updated hours or days after the actual event

• Attribution adjustments: conversions can be attributed retroactively to earlier clicks

• Platform corrections: source platforms may correct or recalculate historical data

• Time zone differences: events near midnight may be reassigned to the correct date

1

How it works

Instead of only synchronizing data for the reference date, the connector retrieves data for a period including past days (the lookback window) to capture retroactive updates.

2

Example with a 7-day lookback window

Today's date: 2025-01-22 Reference date: 2025-01-15 Lookback window: 7 days

Data retrieved for dates:

• 2025-01-15 (reference date)

• 2025-01-14

• 2025-01-13

• 2025-01-12

• 2025-01-11

• 2025-01-10

• 2025-01-09 (reference date - 6 days)

All data for these 7 days is retrieved and synchronized, ensuring any retroactive updates are captured.

3

Practical example

Initial sync on 01/16:

| date       | campaign_id | clicks |
| 2025-01-15 | camp_123    | 100    |

Sync on 01/20 with 7-day lookback:

| date       | campaign_id | clicks |
| 2025-01-15 | camp_123    | 105    | ← updated (5 additional clicks attributed retroactively)

4

Impact by insertion method

The lookback window interacts differently with each insertion method:

• REPLACE Mode: Previous data within the lookback window is deleted and replaced with fresh data from the source.

• UPSERT Mode: Rows are updated with new values, preserving history if configured.

• INSERT Mode: May create duplicates if the same data is retrieved multiple times (not recommended with lookback).

5

Typical lookback window values

Platform Type	Common Lookback	Reason
Advertising platforms (Google Ads, Meta)	3-7 days	Attribution windows, conversion delays
Analytics platforms (GA4)	1-3 days	Processing delays, session completion
CRM/Sales (Salesforce, HubSpot)	1-2 days	Data entry delays, batch updates
E-commerce (Shopify, WooCommerce)	1-2 days	Order updates, refunds

6

Trade-offs

Critère	Longer Lookback Window	Shorter Lookback Window
Data completeness	✅ More complete and accurate data	⚠️ Risk of missing retroactive updates
Retroactive changes	✅ Captures all platform corrections	❌ May miss late attribution or corrections
Synchronization time	❌ Longer processing time	✅ Faster synchronizations
Data warehouse costs	❌ Higher (more partitions scanned/written)	✅ Lower (fewer partitions affected)
Data consistency	✅ More reliable for reporting	⚠️ Potential inconsistencies over time
Use case	Critical metrics, attribution analysis	High-volume data, stable metrics

7

Best practices

• Configure the lookback window based on the platform's attribution window.

• Monitor data stability: if values change significantly after several days, increase the lookback.

• For critical metrics, prefer a longer lookback window to ensure data accuracy.

• For high-volume tables with stable data, a shorter lookback may be sufficient.

8

Configuration in Quanti

The lookback window is configured at the connector level and can be adjusted based on your data quality requirements and platform-specific behavior.