Real-time vs nightly ad sync: how fresh does ad data need to be?

How fresh ad platform data really is

Ad platform data is never fully live, and each platform publishes its own refresh schedule. Meta's Insights API limits and best practices page says insights refresh every 15 minutes and do not change after 28 days of being reported. The same page says metrics may keep updating for a couple of days after an ad has ended, and that Meta reports insights in the ad account's time zone.

Google splits its metrics by speed. Google's data freshness page gives clicks, impressions and cost a 1-hour freshness Service Level Objective, so they refresh hourly. Conversions credited with the last click model are on the same hourly schedule. Conversions credited with any other model arrive the next day: Google's own example has Sunday's figures ready at 3:00 PM on Monday for an account in San Francisco. Impression share and auction insights update once a day, and search click share runs 3 days behind.

Snapchat tells you in each answer how much of it is final. Snap's Marketing API measurement docs describe two fields on every stats response. finalized_data_end_time marks the moment up to which non-conversion metrics are final. conversion_data_processed_end_time does the same for conversion metrics. Anything after those times may still change.

The comparison on Google's data freshness page also shows how much imported conversions add. Goals and transactions imported from Google Analytics take 12 hours under last click and 24 hours under other models. Google Ads' own conversion tracking takes 3 and 15 hours.

Why yesterday's numbers keep changing

Yesterday's ad numbers keep changing because conversions arrive after the ad interaction, and platforms keep processing them for days. Which day a late sale lands on depends on the platform. Google's conversion delay page says Google Ads reports conversions by impression date, so a purchase on Friday from an ad shown on Monday raises Monday's numbers on Friday. Google adds that this lag can make CPA look higher and ROAS look lower than they really are.

Meta and Snapchat file web conversions differently. Meta's Insights best practices page says that since June 10, 2025 its API reports off-Meta actions, such as web purchases, by when they happened, and on-Meta actions, such as link clicks, by impression time. Snap's measurement API reports conversions by the time of the conversion unless you ask for impression time instead.

The credit window sets how far back a sale can reach. Google's conversion windows page gives the 30-day default and the 1-to-90-day range. On Google, a recent day can keep gaining conversions for as long as that window runs. On Meta and Snapchat, a recent day already holds sales from older ads, so it moves less for that reason, but Meta still says its figures can change for up to 28 days.

Late conversions are not the only revision. Google's freshness page names three reasons a past day can move: invalid traffic, such as bot clicks, is removed after the fact; late conversions arrive; and billable events are adjusted during the billing cycle for things like overdelivery. On every platform, the newest days in a report are the least settled.

Real-time vs nightly sync: the four designs compared

A real-time sync reads the platforms every few minutes, and a nightly sync reads them once a day; what decides accuracy is whether each read goes back over recent days. The table sets out four designs a reporting tool or an in-house pipeline can use. The trade-offs are this article's analysis, built on the platform limits above.

Rate limits are the real cost of polling. Meta's Insights best practices page says it counts insights calls against limits per app and per ad account, and that once an app reaches its limit, only a share of requests go through. It recommends spreading queries through the day, planned around the ad account's time zone, and backing off as usage nears 100%. A tool that polls hundreds of accounts every few minutes has to spend that budget carefully, and a throttled read shows stale numbers without saying so.

When real-time ad data matters

Real-time ad data matters when the decision is about delivery or spend, not about results. Delivery metrics are close to live on Meta and Google, while conversions are not, so the cases where minutes count are the ones you can judge from spend, impressions and clicks:

• Launch day. Whether a new campaign is spending at all shows in spend and impressions, which Meta refreshes every 15 minutes and Google hourly.
• Spend pacing. A budget near its cap, or a sale that ends at midnight, needs same-day spend.
• Broken tracking. Clicks with no conversions at all for several hours, on an account that usually converts within hours, often point to a tag or pixel problem rather than a bad ad.

In each case the native Ads Manager is the fastest view, because it reads the platform directly with no sync in between.

When a daily read is enough

A daily read is enough for decisions based on results: moving budget between campaigns, judging a creative, or tracking ROAS over weeks. Those decisions rest on conversions, and conversions for the last few days are incomplete at any refresh rate. Two habits matter more than the cadence:

1. Judge days of the same age. Compare last Monday with the Monday before it as each looked 7 days later, not with yesterday. Otherwise the newer day loses for being new.
2. Wait out your own lag. Google's conversion windows page points to the time lag report, which shows how long your customers take to convert. Treat a day as provisional until most of that lag has passed.

How to check what your reporting tool actually does

Checking a reporting tool's freshness takes five questions, and the answers matter more than the schedule on its marketing page. A schedule says when a sync is meant to start. Jobs can be skipped, throttled, deduplicated or fail, so the only proof of freshness is a timestamp of the last successful read.

1. When did each account last sync? Look for a per-account timestamp, not one for the whole workspace.
2. How far back does each sync re-read? It should cover your longest conversion window, or at least Meta's 28 days.
3. Does a re-read overwrite or append? Overwriting lets late conversions land; appending freezes them out.
4. Which time zone cuts the day? It should be the ad account's own, or totals by day will not match the platform.
5. What does a failed sync look like? A good tool says the figures are stale; a weak one shows the last numbers it had as if they were current.

To test the answers, pick a day more than 28 days old and compare it with the platform's own Ads Manager: it should match. Then read yesterday twice, a day apart: conversions should rise.

How overads refreshes ad data

overads reads connected Meta, Google, X and Snapchat ad accounts on a schedule, at least once a day; it does not stream data in real time. The overads home page describes the result as read-only ad reports. Meta, Google and X sync every campaign, ad set and ad; Snapchat shows spend totals only.

Each pass re-reads recent days rather than appending yesterday. In the sync code on October 7, 2026, a pass re-reads up to 90 days by default and overwrites the stored daily rows, so any revision Meta, Google or Snapchat makes to an earlier day replaces the figure overads held before. That window is longer than Meta's 28 days and Google's default 30-day conversion window.

overads shows when each figure was read. The Integrations page labels every connected account with when it last synced. On paid plans, the overads API's analytics endpoints and its MCP server return a collectedAt time beside the figures, plus a plain sentence that the figures describe that moment. When collection has been failing, the sentence says the figures are real and not current.

Where overads fits

overads fits a team that reviews ad results once a day across Meta, Google, X and Snapchat in one place. overads is an AI marketing agent built on that rhythm: it reads recent results, writes the weekly ad read and suggests what to change.

What overads does not do:

• It is not a live spend monitor. It does not read ad accounts minute by minute, so launch-day checks and end-of-sale pacing belong in each platform's own Ads Manager.
• It cannot change ads. Access is read-only: overads proposes a change and a person applies it in the platform, for the reasons in our guide to why many ad tools can read campaigns but not change them.
• It does not read LinkedIn, Pinterest or TikTok ad accounts.

A team that needs intra-day reporting across many platforms in a spreadsheet or BI tool may be better served by a data connector built for that, and the platform's own Ads Manager remains the reference for any single number. For brands comparing ad revenue with store revenue, our Meta vs Shopify ROAS guide shows how to line the two up, and overads for DTC operators describes the morning read overads is built around.