> ## Documentation Index > Fetch the complete documentation index at: https://docs.vortexiq.ai/llms.txt > Use this file to discover all available pages before exploring further. # ROAS, AdRoll > ROAS 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](/nerve-centre/overview#metrics-types-explained) • **Category:** [Ad Platform](/nerve-centre/connectors#connectors-by-type) ## At a glance > Return on ad spend, the headline efficiency number for AdRoll. **`conversion_value ÷ spend`** at the advertisable level. A 6× ROAS means £1 of AdRoll spend produced £6 of AdRoll-attributed revenue. **Retargeting-platform ROAS expectation is 6, 10×**, materially higher than top-of-funnel acquisition platforms (Meta, Google) because the audience is warm. Below 4× on a retargeting-heavy account is a warning sign; below 2× is unprofitable for most DTC margins after COGS, fulfilment, and overhead. Post-iOS-14.5 ATT, AdRoll's reported ROAS over-states real business ROAS by 30, 60% on iOS-heavy accounts without Conversions API. | | | | --------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **The formula** | `SUM(conversion_value) ÷ SUM(spend)` over the period at advertisable level. Result is a unitless multiple. Both inputs are in advertisable currency. | | **What "spend" means** | Gross media cost in advertisable currency, before AdRoll platform fees, agency markup, or partner rebates. The figure you're billed for. | | **What "conversion value" means** | AdRoll-attributed revenue (`conversion_value` summed across all line items). **NOT order revenue from the commerce platform**; the two often differ by 30, 50%. See *Reconciling* for why. | | **Conversion attribution model** | **Last-touch with view-through credit.** AdRoll's default is 30-day post-click + 7-day post-view, configurable per advertisable. Tightening to 7-day click typically drops ROAS 15, 30% on retargeting-heavy accounts. | | **Attribution window** | **30-day click + 7-day view** (default). Wider than Meta's post-iOS 7d/1d, narrower than Google's 30d/1d on conversions\_value. | | **Retargeting ROAS bands** | **Warm-audience retargeting**: 6, 10× typical. **Cart-abandon orchestration**: 8, 15× (heavily cannibalised, see FAQ). **Lookalike prospecting**: 1.5, 3× typical. **Brand-safe contextual prospecting**: 1, 2.5×. The headline ROAS is a spend-weighted blend. | | **iOS 14.5+ ATT impact** | **Severe and structural.** AdRoll's cross-site cookie network was hit harder than walled-garden platforms (Meta, Google). Without server-side tagging, expect AdRoll-reported ROAS to over-state real business ROAS by 30, 60% on iOS-heavy audiences. With AdRoll Conversions API + Identity Graph live, gap narrows to 10, 25%. The card cannot correct for this; it reports what AdRoll reports. | | **View-through inflation** | **Material on retargeting line items.** AdRoll's 7-day view window means a user who scrolled past your ad on Tuesday and bought via Google search on Friday is claimed by AdRoll. The view-through share of `conversion_value` is typically 25, 45% on retargeting-heavy accounts. To see click-only ROAS, switch attribution to "Click Only" in advertisable settings (cuts reported ROAS by 25, 40%). | | **Bot / invalid traffic** | AdRoll filters detected IVT impressions before billing; conversion-side IVT is removed before reporting. Real-world IVT-conversion drag is \<0.5% on properly tagged accounts. | | **Currency** | Advertisable currency. Single-currency per advertisable; multi-currency merchants run separate advertisables. | | **Time window** | `30D vsP` (default 30D vs the prior 30D). Real-time updates with up to 4-hour ingest lag on "today"; view-through conversions accumulate over 7 days post-impression. | | **Alert trigger** | `<2 (warn)`, `<1 (critical)`. Below 2× is unprofitable for most DTC margins after COGS; below 1× is bleeding cash. Drives `sentiment_key: roas`. | | **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 A US beauty DTC brand running AdRoll for site retargeting + lookalike prospecting + cart-abandon orchestration. The 30-day window is 02 Apr 26 to 01 May 26. Account currency USD. AdRoll Conversions API live since 18 Mar 26. | Line item | Spend (\$) | Conversion value (\$) | Per-line-item ROAS | iOS share | | ------------------------------------- | ------------ | --------------------- | ------------------ | --------- | | Site retargeting (web visitors, 30d) | 9,800 | 78,400 | **8.00×** | 58% | | Lookalike prospecting (purchaser LAL) | 6,400 | 14,800 | **2.31×** | 61% | | Cart-abandoner email + display | 2,200 | 31,200 | **14.18×** | 56% | | Brand-safe contextual prospecting | 1,200 | 2,100 | **1.75×** | 49% | | **Account total (this card)** | **\$19,600** | **\$126,500** | **6.45×** | **57%** | Shopify-side total revenue for the same period: $284,000. Of that, UTM-tagged AdRoll-source revenue: $58,200. GA4 *Display* channel attributed revenue: \$74,400. What's interesting: 1. **The headline 6.45× is in-band for retargeting-heavy accounts.** Retargeting platforms should land in the 6, 10× range; if you're seeing 4× or below, audience match rate is degraded (most likely iOS / Safari ITP shrink) or the optimiser is buying low-quality inventory to hit pacing. **Below 4× on a retargeting-heavy account is a yellow flag, not a green light.** 2. **The 6.45× hides a 14× cart-abandon line item.** AdRoll's email + display orchestrated cart-recovery flow takes credit for purchases that your existing Klaviyo flow would also have caught. Real incremental ROAS on that line item is closer to 6, 8× (the 14× is the upper bound that includes cannibalisation). **Run a 10% holdout for 30 days to find the real number.** 3. **Site retargeting at 8.00× is the honest read on AdRoll's core product**, paying to re-target users you've already paid Google or Meta to acquire. The math: at a 60% gross margin, 8× ROAS contributes \~3.8× to gross profit, but you're double-paying on these conversions (top-of-funnel + retargeting). Subtract cannibalisation and the net incremental ROAS is closer to 4, 5×, still profitable, but not the 8× headline. 4. **Prospecting at 2.31× is acceptable cold-audience performance** for AdRoll. Pure cold prospecting on AdRoll typically lands 1.5, 3×, lower than Meta or TikTok cold audiences. If the growth lever is prospecting, Meta or TikTok will outperform AdRoll on a CPA basis. 5. **Pre-CAPI on this account, ROAS read 4.8× on similar spend; post-CAPI it reads 6.45×.** The \~34% lift is real measurement (catching iOS conversions Pixel-only had missed), not inflation. The underlying business ROAS was the same; the new number is more accurate, not better. If you're judging the rollout, expect a step-change in the 7, 14 days following CAPI go-live and *do not* compare year-on-year ROAS across the rollout boundary. 6. **The 30-day prior window had ROAS 6.81×, slight decline.** Spend up 18%, ROAS down 5%. That's mild scaling pressure; healthy provided ROAS holds above 5×. If next period shows ROAS down a further 10%+ alongside spend up, hit the brakes; the auto-optimiser is buying lower-quality inventory to hit pacing. 7. **Real business ROAS is closer to 3, 4×, not 6.45×.** UTM-tagged AdRoll Shopify revenue is $58k; AdRoll's claim is $126k. The gap is structural (view-through credit + iOS modeling). Use a weighted blend (40% AdRoll claim + 60% Shopify-UTM) for the CFO read: \~\$85k, ROAS \~4.3×. **Don't quote the headline 6.45× to finance.** Quick sanity tests: * ROAS up + spend up = healthy scaling, expand budget if pacing allows. * ROAS flat + spend down = budget cut without channel deterioration; channel is still healthy. * ROAS up + spend down = pulled back wisely from low-quality inventory; can scale back when audience refreshes. * ROAS down + spend up = scaling beyond efficient frontier. Cap budget, refresh creative, audit per-line-item ROAS. * ROAS down + spend flat = something changed: pixel breakage (most common), attribution shift to Meta or Google, or ATT audience collapse. **Investigate before cutting.** * ROAS spike (>2× normal) on a single day = view-through window catch-up or a fraudulent conversion event. Don't celebrate; verify in commerce-platform truth first. ## Sibling cards merchants should reference together | Card | Why pair it with ROAS | What the combination tells you | | ----------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- | | [Total Spend](/nerve-centre/kpi-cards/adroll/total-spend) | The denominator. ROAS up + spend up is healthy scaling; ROAS up + spend down is just a retreat. | The shape of growth. | | [Total Revenue](/nerve-centre/kpi-cards/adroll/total-revenue) | The numerator. Tells you whether ROAS moved on cost-side or revenue-side. | If ROAS dropped because revenue cratered (not spend rising), the issue is on the conversion path or pixel. | | [ROAS Trend](/nerve-centre/kpi-cards/adroll/roas-trend) | Daily series. Retargeting ROAS is volatile day-to-day; the 7-day rolling is the actionable read. | Detect the shape of decline (gradual creative fatigue vs sudden pixel break vs ATT audience collapse). | | [ROAS by Campaign](/nerve-centre/kpi-cards/adroll/roas-by-campaign) | Headline ROAS hides per-line-item variance. Cart-abandon at 14× drags the average up; prospecting at 2× drags it down. | **Open this card before making decisions.** Headline alone is misleading. | | [Conversion Lag](/nerve-centre/kpi-cards/adroll/conversion-lag) | View-through latency. Today's spend may credit revenue 3, 7 days from now. | If lag is rising, ATT is degrading attribution; if lag is falling, inventory mix shifted toward higher-intent placements. | | [Clicks vs Conversions](/nerve-centre/kpi-cards/adroll/clicks-vs-conversions) | The broken-tracking canary. Clicks rising while conversions stagnate = pixel issue. | ROAS will look like it dropped, but the cause is measurement. Don't cut spend on a measurement bug. | | [CTR Trend](/nerve-centre/kpi-cards/adroll/ctr-trend) | Click-through rate. CTR drops typically precede ROAS drops by 1, 2 weeks (creative fatigue). | Early-warning shape for ROAS deterioration. | | [Wasted Spend](/nerve-centre/kpi-cards/adroll/wasted-spend) | Zero-conversion line items eating budget. | Pause-or-fix list; eliminating wasted spend lifts headline ROAS without changing performance on the productive line items. | | [Meta Ads ROAS](/nerve-centre/kpi-cards/facebook-ads/roas) | Walled-garden retargeting peer. Meta's retargeting ROAS sits at 5, 9× typically. | If AdRoll ROAS is materially below Meta retargeting ROAS, AdRoll's audience match rate is degraded post-iOS; consider consolidating spend on Meta. | | [Google Ads ROAS](/nerve-centre/kpi-cards/google-ads/roas) | Top-of-funnel acquisition peer (different funnel position). | Retargeting ROAS should be 2, 3× higher than top-of-funnel ROAS. If they're equal, retargeting is not adding incremental value. | | [Shopify Total Revenue](/nerve-centre/kpi-cards/shopify/total-revenue) | The truth side. Real business ROAS = (Shopify revenue × AdRoll channel share) ÷ AdRoll spend. | If your business ROAS is 3.5× and AdRoll-claimed ROAS is 6.5×, that's a 1.85× over-claim, normal post-iOS without CAPI. | ## Reconciling against the vendor's own dashboard **Where to look in AdRoll's own dashboard:** > [AdRoll Dashboard → Reporting → Performance Report → "ROAS"](https://app.adroll.com/reporting) (filter to the same advertisable and date range used in this card). The "ROAS" column in AdRoll's Performance Report is the same ratio this card surfaces. Headline number on the *Home* tile shows it for the chosen window. Match the date range and the figure should reconcile to within sub-percent rounding. Other AdRoll views that *look* similar but aren't: * **"View-Through ROAS"**: a separate view that isolates view-only-credited conversions. This card uses the blended click + view ROAS. * **"Click-Only ROAS"**: a toggle in advertisable settings; if you've switched to click-only, this card matches that setting. * **"Influenced Revenue"**: AdRoll's broader claim that includes any user touched by AdRoll inventory regardless of last-touch. Not a real ROAS; ignore for decisions. **Why our number may legitimately differ from AdRoll itself (rare):** | Reason | Direction | Why | | --------------------------------- | --------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Real-time event-ingestion lag** | Marginal on most recent 24h | AdRoll Reporting API has a 2, 4 hour lag; view-through events accumulate over 7 days post-impression. | | **Attribution model toggle** | Direction depends | If the merchant changed click-only / click+view mid-window, AdRoll retroactively reflows the data; this card matches whatever the current model says. | | **Time zone** | Boundary days off | AdRoll uses advertisable time zone; Vortex IQ uses UTC. For "today" / "yesterday" the boundary shifts can move ROAS 5, 12% on US Pacific advertisables. | | **Modeled-conversion fill** | None expected | Both views read the same blended (real + modeled) `conversion_value` field. | **Why the BUSINESS ROAS often differs from this card (the important one):** The `conversion_value` AdRoll reports is whatever the pixel + Conversions API saw, attributed back through AdRoll's last-touch model with view-through credit. That value flows from your tag setup: * If the pixel sends *order revenue with VAT*, ROAS is on a tax-inclusive basis. * If the pixel sends *only first-purchase value*, ROAS understates returning-customer revenue. * If the pixel sends *modelled value*, ROAS includes some imputed revenue. For the *true business ROAS* (the one your CFO cares about), divide [`shopify.total_revenue`](/nerve-centre/kpi-cards/shopify/total-revenue) attributed to AdRoll UTMs by AdRoll spend. The two figures should land within 30, 50% post-CAPI; bigger gaps usually mean (a) commerce-platform attribution disagrees with AdRoll's view-through window, (b) the conversion pixel is misconfigured, or (c) iOS / Safari ITP impact is degrading attribution despite CAPI. **Cross-connector reconciliation:** This card is *AdRoll's view of AdRoll-driven performance*. The same purchase will be claimed differently by every platform with a tag in the path: | Card | Expected relationship | What causes legitimate divergence | | ------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | [`google_analytics.ga_revenue_by_channel`](/nerve-centre/kpi-cards/google-analytics/revenue-by-channel) | GA4 *Display* revenue ÷ AdRoll spend ≈ this card × 0.5, 0.75 | GA4 uses last-non-direct click; AdRoll uses last-touch with 7-day view. AdRoll typically wins more clicks because of the wider window and view-through credit. **AdRoll inflates by 30, 60% versus GA4** structurally. | | [`shopify.total_revenue`](/nerve-centre/kpi-cards/shopify/total-revenue) | (AdRoll `conversion_value`) is a subset claim of Shopify total revenue **but** AdRoll-attributed Shopify revenue (UTM-tagged) is much smaller than AdRoll claims. **AdRoll typically claims 1.5, 2.5× the UTM-truth Shopify number due to iOS gap and view-through credit.** | The "true" business ROAS sits between AdRoll's claim and Shopify's UTM count. Use a weighted blend (40/60 AdRoll/Shopify) for the CFO read. | | [`facebook_ads.fac_roas`](/nerve-centre/kpi-cards/facebook-ads/roas) | Independent paid channel; comparable retargeting peer. Meta retargeting typically sits at 5, 9×. | Different audiences, different inventory. AdRoll's match rate is structurally weaker post-ATT because Meta has a closed ID graph. | | [`google_ads.gads_roas`](/nerve-centre/kpi-cards/google-ads/roas) | Different funnel position. Google is intent / search; AdRoll is awareness / retargeting. | Comparable benchmark only; retargeting ROAS should be 2, 3× higher than Google's blended ROAS. If equal or lower, retargeting is not adding incremental value. | ***
Documentation cross-reference (retargeting ROAS bands and CAPI rollout) Retargeting platform ROAS expectations differ structurally from top-of-funnel acquisition. Reference bands for healthy DTC accounts: | Audience type | Expected ROAS band | Notes | | --------------------------------------- | ------------------ | ------------------------------------------------------------------------------------- | | Cart-abandon orchestration | 8, 15× | Heavily cannibalised; run a holdout for true incremental ROAS. | | Site retargeting (warm visitors) | 6, 10× | Core retargeting; ROAS below 5× signals audience-match degradation. | | Custom audience retargeting (CRM lists) | 5, 9× | Depends on list freshness and recency. | | Lookalike prospecting (purchaser seed) | 1.5, 3× | AdRoll's prospecting inventory is mostly GDN remnant; Meta or TikTok will outperform. | | Brand-safe contextual prospecting | 1, 2.5× | Cold audience, AdRoll has limited differentiation here. | A clean Conversions API rollout typically lifts AdRoll-reported ROAS by **20, 40%** in the 7, 14 days following implementation, as AdRoll backfills previously-missed iOS conversions. The lift is real measurement, not inflation. Annotate the Change History below when CAPI goes live. After CAPI is stable, modeled-conversion fill in `conversion_value` should drop from \~20% to \~8, 12%. If modeled fill stays above 20% post-CAPI, your CAPI implementation has gaps.
## Known limitations / merchant FAQs **My AdRoll ROAS is 8× and AdRoll says I'm profitable, but my P\&L disagrees.** AdRoll's `conversion_value` is gross revenue claimed via last-touch + view-through. Your P\&L profitability also factors in: COGS (typically 30, 50% of revenue), fulfilment cost (5, 15%), payment processing (\~3%), returns (5, 15%), overhead, *and* cannibalisation (the share of AdRoll-claimed conversions that would have happened anyway via email, organic, or direct). Rule of thumb: divide ROAS by 2 for a "true contribution multiple" on typical DTC margins, then halve again for incremental contribution if the line item is cart-abandon or warm retargeting. **An 8× ROAS on cart-abandon is roughly a 2× true incremental contribution multiple.** **Why is my AdRoll Branded retargeting ROAS so high (12×+)?** Same dynamic as Branded Search on Google Ads or returning-audience Advantage+ on Meta. Users who already know your brand were going to come back anyway; AdRoll captures \~50, 80% of that intent for warm-audience CPMs. **Branded / repeat-visitor retargeting ROAS is a poor measure of incremental health**, you're paying AdRoll to defend revenue you'd capture for free. The "true acquisition" ROAS is on cold prospecting or new-visitor retargeting only. **My ROAS dropped 30% overnight, what should I check first?** In order of likelihood: (1) **Pixel breakage**. Check AdRoll → Pixel → Diagnostic; broken events show as warnings. (2) **Conversion-tracking regression**. Check [Clicks vs Conversions](/nerve-centre/kpi-cards/adroll/clicks-vs-conversions), if clicks held but conversions cratered, the tag broke. (3) **Attribution-model change**. Check advertisable settings; someone may have switched click-only / click+view. (4) **iOS audience pool collapse**. ATT opt-out continues to grind audience match rates down; if you haven't refreshed creative or expanded audiences in 30+ days, expect a drift. (5) **Promo period rollover**. ROAS during a 30%-off campaign rolls into "normal" the day after; sales attribution catches up unevenly. (6) **Cross-platform competition**. Meta or Google launched a new campaign that's now winning last-click on conversions AdRoll used to claim. The conversions still exist; AdRoll just lost claim. **Should I optimise for ROAS or for Spend × ROAS (revenue)?** Depends on your phase. If you're profitability-constrained (need every retargeting dollar to break even on its own), optimise for ROAS, retargeting ROAS should sit at 6, 10×. If you're growth-focused with profitable unit economics on top-of-funnel, optimise for absolute incremental conversion volume; ROAS will compress as you push into prospecting line items. The right answer is rarely "highest ROAS"; usually it's "max profit subject to spending the budget". **iOS 14.5 ATT impact, the structural answer:** ATT (App Tracking Transparency) gave iOS users a system-level prompt to opt out of cross-app tracking. Around 60, 75% of UK / US iOS users opt out. AdRoll, as a cross-site retargeting network, was hit *harder* than walled-garden platforms (Meta, Google) because: * AdRoll's pixel can no longer reliably link iOS users across sites (the basis of retargeting). * Safari ITP independently kills third-party cookies, compounding the damage on Safari users (\~40% of UK / US web traffic). * For opted-out users, AdRoll backfills with statistical modeling and Identity Graph (its post-cookie ID solution). **Net effect on this card**: without server-side tracking, expect AdRoll-reported ROAS to over-state real business ROAS by 30, 60% on iOS-heavy audiences. With AdRoll Conversions API + Identity Graph live, narrows to 10, 25%. Action: roll out CAPI immediately if you haven't, and treat ROAS readings as directional, not exact. **What's the practical AdRoll Conversions API rollout playbook?** 1. **Shopify**: Install AdRoll's official Sales Channel app → enable Conversions API → choose "Maximum" data sharing. 30-minute setup. Auto-deduplicates Pixel + CAPI via `event_id`. **Easiest.** 2. **BigCommerce**: AdRoll BC app + manual CAPI via Tag Manager + custom event\_id. 2, 4 hours. 3. **Adobe Commerce / Magento**: AdRoll extension + custom server endpoint. 1, 3 days. 4. **Custom / headless**: Build a server-side proxy that mirrors Pixel events. 1, 2 weeks. 5. **In all cases**: deduplicate via `event_id` (unique per purchase) and matching event timestamps. Test in AdRoll → Pixel → Server-Side Events. 6. **Allow 7, 14 days for retraining** before judging the ROAS lift; AdRoll's models recalibrate over that window. **How do I interpret modeled conversions on AdRoll?** AdRoll auto-fills attribution gaps with statistically-modeled values. They're invisible (rolled into `conversion_value`) but typically run 8, 18% of total revenue for Pixel-only DTC accounts. **Patterns**: * **Pre-CAPI**: modeled fill \~20, 30%. Inflated. * **Post-CAPI clean**: modeled fill \~5, 12%. Healthy. * **Post-CAPI but still high (>20%)**: implementation gaps, audit `event_id` deduplication and event coverage. You cannot opt out. Trust the rolling 30-day, not the daily. **Why is AdRoll's claimed ROAS often higher than Meta's claimed ROAS on the same merchant?** Two structural reasons: (1) **View-through window**. AdRoll defaults to 7-day view; Meta defaults to 1-day view post-iOS 14.5. AdRoll claims more view-through conversions. (2) **Audience overlap**. AdRoll often re-targets users Meta already targeted; both claim the same conversion. Meta's pixel typically fires before AdRoll's on cart-and-checkout flows, but AdRoll's last-touch + view-through wins on users who saw an AdRoll display ad mid-funnel. **Sum of "Meta-claimed + AdRoll-claimed" almost always exceeds total commerce-platform revenue because of double-claim.** Use commerce truth (Shopify, BigCommerce, Adobe) as the denominator. **AdRoll vs Criteo, who wins on ROAS?** Different audiences, different inventory. **AdRoll skews toward Google Display Network + Facebook Audience Network mix; Criteo skews toward open-exchange catalogue retargeting.** On the same DTC merchant, AdRoll typically reports 10, 20% higher ROAS, but this is largely a measurement artefact (AdRoll's wider view-through window). Real incremental ROAS, measured via holdout, is usually similar between the two. **Don't pick the platform on claimed ROAS alone**; pick on inventory fit, audience reach, and incrementality testing. **Can I trust the "today" ROAS?** Less than the rolling 7-day. Today's ROAS is built from incomplete data: spend ingests with 2, 4 hour lag; view-through conversions accumulate over 7 days; modeled-conversion fill takes up to 72h to converge. **The 7-day rolling average is the most actionable.** Don't restructure line items based on a single day's ROAS. **My multi-currency setup, how does ROAS work?** AdRoll advertisables are single-currency by design. Multi-currency advertisers run separate advertisables per currency. Each advertisable reports ROAS in its own currency. This card is per-advertisable. To roll up across currencies for a unified business view, weight each advertisable's ROAS by its spend in a common currency (FX-converted at month-end rates). **A simple average of ROAS across advertisables is wrong** because it gives equal weight to a $50k advertisable and a $500 advertisable. **Pixel + CAPI dual-implementation, do I get double-counted on ROAS?** **Only if dedup isn't configured.** With proper `event_id` matching, AdRoll server-side dedupes; you get the union (whichever event arrived first wins on attribution credit). Without dedup, you double-count and your ROAS reads \~80% inflated. **Test it**: AdRoll → Pixel → Server-Side Events → check dedup status. The official Shopify Sales Channel app handles this automatically; custom implementations need explicit dedup logic. **Is my AdRoll ROAS incremental or cannibalised?** The single most important question on retargeting. Most warm-audience retargeting is partially cannibalised, those buyers would have converted anyway via email, organic, or direct. **Run a holdout test**: exclude 10% of retargetable users from AdRoll for 30 days, measure their conversion rate vs the targeted 90%. The lift is the *incremental* AdRoll ROAS. Typical finding: 30, 60% of AdRoll-claimed conversions are incremental; 40, 70% are cannibalisation. Use the incremental ROAS for true performance evaluation; use the headline ROAS for AdRoll's own optimisation. *** ### Tracked live in Vortex IQ Nerve Centre *ROAS* 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](https://app.vortexiq.ai/login) or [book a demo](https://www.vortexiq.ai/contact-us) to see this metric running on your own data.