Real-Time Attribution for Ecommerce: What Updates Instantly and What Takes 72 Hours
Real time attribution is near real time in AdBeacon: it syncs from all of your connected sources roughly every 10 to 15 minutes, and you can refresh manually at any time. What it cannot do is make the ad platforms finish counting faster. Platform-reported conversions, refunds, and returns arrive later, and some get restated after you have already acted on them. The practical answer is to make fast decisions on fast data, hold slower decisions until the lagging inputs settle, and know which clock each number runs on. This post maps those clocks and tells you what is safe to do now versus later.
Log scale. Platform timings from Google's data freshness documentation; refund timing varies by store.
How fast does AdBeacon update, and how does that compare with GA4?
AdBeacon is near real time: it syncs from all sources roughly every 10 to 15 minutes, so you see new data in minutes, not hours, and you can refresh manually whenever you need the latest. That includes your store orders and your connected ad accounts, so an intraday check does not wait on an overnight job.
GA4 works on a different clock. Google's data freshness documentation says GA4 data processing can take 24 to 48 hours, and that data in your reports may change during that time. GA4 does have a Realtime report that is typically a few minutes behind, but it covers a limited set of dimensions and metrics. For full reporting, expect a delay.
What it means: if you want an early read on whether today's traffic is converting, AdBeacon's near real time view of store orders is the most dependable signal you have. What to do: build your intraday view from it, check the Orders report against your store, and label the number as provisional until the day closes.
Which attribution data is delayed or restated?
Platform-reported conversions inside the ad platforms lag and get revised, and refunds and returns arrive later still. Each has a different delay, and mixing them in one live number is the main source of misleading real-time ROAS. Syncing every 10 to 15 minutes gets you what a platform has published so far. It does not change when a platform decides it is finished counting.
- Ad platform spend and conversions: Google Ads says clicks, impressions, and cost are generally refreshed hourly, while conversion data lags about 3 hours under last-click and 15 hours under other models, per its data freshness documentation. The same page notes metrics can change a day or more after an event because of invalid traffic removal and late conversions.
- Late conversions: Google also explains that conversions can be reported up to 90 days after the click. Because spend is reported in full while conversions are still arriving, recent periods look worse than they will end up.
- Definition changes: Platforms change what they count. Meta narrowed click-through attribution to link clicks and expanded engage-through in 2026, and stopped returning 7-day and 28-day view-through window data in the Insights API on January 12, 2026. A number you saved last year may not reproduce today.
- Refunds and returns: A sale can be reversed days or weeks later. Shopify's Order object tracks refunds separately from the original order, so revenue that looked final can be reduced afterward.
The headline "72 hours" is a working rule of thumb, not a platform guarantee. Some effects settle within a day, some take weeks. A number from this morning's platform dashboard is a draft, not a final.
What decisions are safe to make in real time, and which should wait?
Decisions that are cheap to reverse and based on store-verified data can happen in real time, and decisions that move meaningful budget or depend on platform-reported conversions should wait until those inputs settle. A simple rule: the harder a decision is to undo, the older the data should be.
- Monitor order volume and conversion rate against a normal day for traffic anomalies.
- Catch a broken checkout, a dead landing page, or a tracking outage.
- Pause an obviously broken ad, such as a wrong price or an expired offer.
- Watch a launch or sale day for pacing against store orders, with the date range compared to the previous period. Our post on what a real-time attribution dashboard should show on Cyber Monday walks through the metrics worth watching hourly.
- Cutting or scaling a campaign by more than a modest amount.
- Judging a creative test on platform-reported conversions from the last day or two.
- Reallocating budget between channels based on blended ROAS.
- Reporting performance to a client or leadership as final.
- Any decision that depends on profit, since returns and refunds are not in yet. Our guide to calculating true ROAS covers what to subtract.
What to do: write the two lists on one page, share it with your team, and agree on a minimum age for data before a budget change. Many teams settle on a closed prior week for scaling calls and a closed prior day for pauses.
How do you build a reporting setup that respects data latency?
Show provisional and settled numbers separately, and never put a lagging number inside a live headline. Three habits get you there.
- Tag every number with its source and age. A column that says "store orders, live" and another that says "platform conversions, settling" removes most arguments before they start.
- Use AdBeacon for the live read. Its 10 to 15 minute sync and manual refresh make it the right place for intraday checks. Platform views can still sit alongside it for in-platform optimization, as long as you remember they are still settling.
- Close the books on a schedule. Re-pull platform data for a prior period after it has settled, then compare. Our post on why holiday attribution windows break explains why platform dashboards keep restating conversions for days afterward.
Synced every 10 to 15 minutes. Use for intraday health checks, pacing, and pauses.
Still being counted and restated. Use for in-platform optimization, and re-pull before budget calls.
Part of any gap between a platform and AdBeacon is what each system counts, and part is when each finishes counting. Both are worth knowing before you decide.
Real time attribution is valuable when you use it for what it can actually see. Fast data tells you whether the store is healthy and whether orders are coming in. Slow data tells you whether a campaign was profitable. Keep them apart, give each a clock, and decide accordingly. If you want to see how AdBeacon separates fast store data from settling platform data, book a demo with AdBeacon.
FAQ
Is real-time attribution accurate?
It is accurate for data that arrives quickly, such as store orders you can check against the Orders report. It is unreliable for anything that depends on platform conversions or refunds that have not arrived yet.
How often does AdBeacon sync?
AdBeacon syncs from all sources roughly every 10 to 15 minutes, and you can refresh manually at any time.
How long should I wait before trusting platform-reported ROAS?
Wait at least several days for a rough read and longer for a final one. Google Ads notes conversions can post up to 90 days after the click, so recent periods almost always get revised.
Why does today's ROAS look worse than last week's?
Spend is reported in full right away while conversions are still arriving. The recent period catches up as late conversions are added.
Sources
- About data freshness - Google Ads Help
- Google Ads performance fluctuations: Conversion tracking setup & conversion delay - Google Ads Help
- Data freshness - Analytics Help
- Order - GraphQL Admin (Shopify.dev)
- PPC Land: Meta rewrites click attribution rules, finally aligning with Google Analytics
- PPC Land: Meta restricts attribution windows and data retention in Ads Insights API