Retargeting ROAS Dropped Below Threshold, AdRoll

Retargeting ROAS Dropped Below Threshold for AdRoll stores. Tracked live in Vortex IQ Nerve Centre. How to read it, why it matters, and how to act on it.

Metrics type: Key Metrics • Category: Ad Platform

At a glance

An alert that fires when AdRoll’s retargeting ROAS falls more than a tunable margin below its own recent baseline. AdRoll is retargeting-first: the bulk of spend chases site visitors and basket abandoners that the AdRoll Pixel has already seen, so those audiences are the warmest, highest-ROAS part of the account. A sharp drop here is usually the first visible symptom of something structural: a retargeting pool exhausting as frequency caps bite, the AdRoll Pixel losing events on a tracking break, a catalogue-feed gap starving Dynamic Ads of product data, or a view-attribution loss as browser privacy enforcement tightens. This is an early-warning card, not a reporting card. It exists to catch the slide before a full reporting window bakes the loss in.

What it counts	The trailing-window return on ad spend for AdRoll’s retargeting audiences only (site visitors, product-page browsers, basket abandoners), compared against the same day-of-week reading from the prior 7 days. The alert is the boolean breach state; the displayed value is the current retargeting ROAS and the percentage gap to baseline.
Cost basis	CPM and CPC blended. AdRoll buys display, native, social and connected-TV inventory programmatically, so most retargeting spend is CPM-priced impressions with effective CPC derived from click-through. The ROAS numerator is AdRoll-attributed conversion value; the denominator is retargeting spend.
Currency	Advertiser-account currency. ROAS itself is a ratio, so it is currency-neutral; the underlying spend and revenue are in account currency.
Conversion attribution	AdRoll defaults to a 30-day post-click and 1-day post-view window, configurable per account. Retargeting carries meaningful view-through credit, so this card is sensitive to view-attribution loss on Safari and iOS.
Attribution window	30-day click plus 1-day view default. A genuine drop and an attribution-measurement drop can look identical here; pair with the AdRoll Pixel Tracking Broken card to tell them apart.
Bot / invalid traffic	Filtered pre-billing across NextRoll’s supply partners. Invalid-traffic slippage is generally low on retargeting because the audience is a known, pixel-matched pool rather than open prospecting.
iOS / Safari privacy impact on the card	High. Retargeting is the line most exposed to identity loss. As Safari ITP and iOS App Tracking Transparency enforcement tighten, the matchable pool shrinks and post-view credit evaporates, so this is often the first card to breach after a privacy change.
Catalogue-feed dependency	Indirect but strong. A feed gap forces Dynamic Ads to fall back to static or generic creative, conversion rate collapses, and retargeting ROAS drops even though spend holds. Always check feed health when this alert fires.
Time window	RT (real-time evaluation) against a prior 7-day same-day-of-week baseline. Same-DOW comparison removes the weekly seasonality that would otherwise trip false alerts on, for example, weekend apparel spikes.
Alert trigger	retargeting ROAS drop >25% vs prior 7D same-DOW. An illustrative threshold; the breach fires when current retargeting ROAS is more than a quarter below the same weekday a week earlier. The margin is tunable per account.
Roles	owner, marketing, finance

Calculation

Calculated automatically from your AdRoll data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading.

Worked example

Northfell Outfitters, a UK outdoor-apparel DTC retailer on BigCommerce, runs AdRoll retargeting (web Dynamic Ads plus a native layer) with a small prospecting budget on top. The alert evaluates retargeting ROAS each day against the same weekday a week prior. Account currency GBP.

Day	Retargeting spend (£)	AdRoll-attributed revenue (£)	Retargeting ROAS	Same-DOW baseline	Gap	State
Mon 08 Jun 26	940	6,580	7.0x	7.2x	-3%	OK
Tue 09 Jun 26	970	6,890	7.1x	7.0x	+1%	OK
Wed 10 Jun 26	990	4,650	4.7x	7.1x	-34%	BREACH
Thu 11 Jun 26	1,010	3,840	3.8x	7.0x	-46%	BREACH

What the pattern tells you:

1. The breach on Wed 10 Jun 26 is the signal, not the Thursday number. By Thursday the loss is obvious to anyone reading the weekly report. The point of this card is that it fired on Wednesday, roughly 24 hours earlier, before a reporting window closed.
2. Spend held while ROAS fell. This is the classic pool-exhaustion or feed-gap fingerprint. AdRoll’s auto-optimisation is still pacing budget, still buying impressions, but each impression now lands on an over-served user (frequency caps hit) or renders fallback creative. Flat spend masking falling efficiency is the most dangerous failure mode on a retargeting-first account.
3. A 34% then 46% slide over two days points at a structural cause, not auction noise. Bid-density wobble produces single-day dips that recover. A two-day deepening slide usually means the retargeting pool has shrunk, a feed rejection has hit Dynamic Ads, or a top seller has gone out of stock.
4. First diagnostic stop is the retargeting pool size. In the AdRoll dashboard, open Audiences and check the matched pool count and the campaign frequency. If the pool has shrunk or frequency has spiked, you are over-serving a smaller audience: see Wasted-Spend Burst (retargeting pool exhaustion).
5. Second stop is the catalogue feed. If the pool is healthy, check the product catalogue under Dynamic Ads. A rejected or stale feed pushes Dynamic Ads to generic creative and conversion rate collapses. See Spend on Campaigns with Active Feed Rejections and Active Dynamic Ads on Out-of-Stock SKUs.
6. Third stop is the tracking layer. If pool and feed are both clean, check whether the AdRoll Pixel is still firing conversion events on a rising share of sessions. A measurement break shows up as falling attributed revenue with steady store orders. See AdRoll Pixel Tracking Broken.

Quick sanity tests:

• ROAS down, spend flat, pool shrunk and frequency up = pool exhaustion, refresh or expand the audience.
• ROAS down, spend flat, feed shows rejections or out-of-stock SKUs = feed gap, fix the catalogue first.
• ROAS down, spend flat, pixel events dropping = tracking break, fix the conversion signal, do not cut budget.
• ROAS down, CPM or CPC up = inventory competition or a smaller pool re-priced, check the auction conditions.
• ROAS down only on Safari or iOS split = attribution measurement loss, not necessarily real revenue loss.
• ROAS recovered same week after an audience refresh or feed re-sync = confirmed cause found, no further action.

Sibling cards merchants should reference together

Card	Why it matters next to Retargeting ROAS Dropped Below Threshold	What the combination tells you
ROAS	The headline account-level efficiency reading.	If account ROAS is steady but retargeting ROAS breached, the problem is isolated to retargeting audiences, usually pool, feed or tracking.
ROAS by Campaign	Per-campaign efficiency split.	Pinpoints whether the breach is one retargeting campaign (one shrinking pool) or all of them (an account-wide pixel or feed fault).
Conversions Trend	The conversion-volume side of the ratio.	A falling conversion count alongside the ROAS breach points at real demand loss or generic-fallback creative.
Conversion Lag	View-through and click latency, critical for retargeting.	Some apparent ROAS drop is just lagged credit that has not landed yet; rising lag softens the alarm.
Conversion Drop Alert	The conversions-volume counterpart alert.	Both firing together = real demand loss; only ROAS firing = an efficiency or attribution issue with volume intact.
Wasted Spend	Spend that returned little or nothing.	Rising wasted spend through a ROAS breach confirms the budget is pacing into an exhausted or poorly matched pool.
Total Spend	The denominator in absolute terms.	Spend flat through a ROAS breach is the dangerous case; spend rising is at least a controllable cause.

Reconciling against AdRoll

Where to look in AdRoll’s own dashboard:

In the AdRoll dashboard at app.adroll.com, open Reporting, choose the Campaigns or Performance view, filter to your retargeting campaigns (exclude prospecting), set the date grouping to daily, then compare the “Revenue” (or attributed conversion value) and “Spend” columns for the breach day against the same weekday a week earlier.

AdRoll does not surface a same-day-of-week retargeting-ROAS breach alert natively; the reporting view shows raw ROAS per period. To reconcile, divide the retargeting “Revenue” column by the retargeting “Spend” column for the breach day and for the same weekday a week before; the percentage gap should match this card within rounding. The account overview tile shows blended ROAS across retargeting and prospecting, so it will read higher and steadier than this retargeting-only card. Always isolate to retargeting campaigns in Reporting when reconciling, and confirm the attribution window in the dashboard matches the one this card is configured to use.

Why our number may legitimately differ from AdRoll itself:

A small gap is normal. Usual suspects:

Reason	Direction	Why
Time zone	Boundary days off	AdRoll reports in the account time zone; Vortex IQ uses UTC for period boundaries. Same-DOW comparison can shift by one calendar day at the boundary.
Ingest lag	Ours lower for “today”	The AdRoll reporting API has a few hours of lag on spend and conversions. A fresh breach can soften once catch-up revenue lands.
Attribution restatement	Ours marginally lower at first	Post-view and late-click conversions credit retroactively; a same-day ROAS reading rises as view-through credit accrues over the following day.
Campaign-type filter	Theirs higher if unfiltered	The account overview tile blends retargeting and prospecting; this card is retargeting-only, so a clean reconcile requires filtering to retargeting campaigns.
Currency	None expected	Both views use account currency; ROAS is a ratio.

Cross-connector reconciliation:

This card is AdRoll-only. The retargeting-ROAS breach concept has no direct counterpart on other platforms, but for merchants running parallel retargeting the reasonable comparison set is:

Card	Expected relationship	What causes legitimate divergence
facebook_ads.fac_total_revenue	Different retargeting inventory and identity graph; useful as a cross-vendor efficiency benchmark, not a reconciliation.	Meta’s closed identity graph degrades differently under App Tracking Transparency than AdRoll’s open programmatic supply and pixel-matched pools.
shopify.total_revenue / bigcommerce.total_revenue	Commerce-platform realised revenue is the source of truth for the ROAS numerator.	AdRoll attributes view-through generously; a meaningful over-state versus store UTM truth is normal, so an AdRoll ROAS breach should be sanity-checked against the store ledger before reallocating budget.

Known limitations / merchant FAQs

My retargeting ROAS breached but my account ROAS looks fine, which do I trust? Both, they answer different questions. Account ROAS blends retargeting and prospecting, so a retargeting slide can be masked by steady prospecting. This card isolates retargeting precisely because it is the warmest, most valuable, and most identity-exposed part of an AdRoll account. A retargeting-only breach is a real early warning even when the blended number looks healthy.

The alert fired then cleared by itself the next day, was it a false alarm? Usually not. Two common benign causes: post-view conversion credit landing late and pulling ROAS back up, or a single-day bid-density wobble that recovered. A breach that self-clears within 24 hours and does not recur is low priority. A breach that deepens over two or more days is structural, treat it as real.

How do I tell a real ROAS drop from an attribution-measurement drop? Check the AdRoll Pixel Tracking Broken card. If the AdRoll Pixel is firing fewer conversion events on a rising share of sessions, you are losing measurement, not necessarily revenue. Cross-check the store ledger: if realised orders held while AdRoll-attributed ROAS fell, it is a measurement problem and the fix is the conversion signal, not budget cuts.

Why same-day-of-week instead of a simple prior-7-day average? Retargeting has strong weekly seasonality; weekends and Sunday evenings convert differently from midweek on most DTC accounts. Comparing a Wednesday to the previous Wednesday removes that seasonality and cuts the false alerts a flat 7-day average would generate every Monday and Saturday.

Is a breach usually the retargeting pool exhausting? Often, yes, on a retargeting-first platform it is the single most common structural cause. When the matched pool shrinks (slower site traffic, a cookie or pixel match falling away, or a long lookback emptying out) and budget keeps pacing, frequency climbs and each extra impression buys less. Check the Audiences view first, then read Wasted-Spend Burst (retargeting pool exhaustion) for the matching spend-side signal.

Should I cut budget the moment this fires? No. The first action is diagnosis, not budget. Check the pool, then the catalogue feed, then the tracking layer. Cutting retargeting budget on a feed-driven or pool-driven dip throws away your most efficient line right when a refresh would have restored it. Cut budget only once you have confirmed the cause is genuine demand loss, not a fixable pool, feed or tracking fault.

Does this card account for AdRoll’s view-through attribution? The card reports AdRoll’s own attributed ROAS, which leans on the 1-day post-view window and over-states versus store UTM truth. That is fine for in-platform breach detection because the comparison is AdRoll-to-AdRoll (this week’s reading versus last week’s, same measurement basis). For absolute ROAS truth, reconcile against the commerce-platform revenue line.

---

Tracked live in Vortex IQ Nerve Centre

Retargeting ROAS Dropped Below Threshold is one of hundreds of KPI pulses Vortex IQ tracks across AdRoll and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English.

Start for free or book a demo to see this metric running on your own data.