> ## 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, MediaMath > ROAS for MediaMath 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 on MediaMath. **`conversionValue ÷ spend`** with view-through credit included. **Status note (May 26)**: post-restructure pixel instability can suppress reported revenue intermittently, leading to ROAS reading lower than truth on this connector. **Always reconcile to T1 UI before treating a ROAS drop as a campaign issue**; pixel/platform issues are more common here than on peer DSPs. | | | | -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ | | **The formula** | `conversionValue ÷ spend` from T1 reporting at organisation level. | | **What "spend" means** | Gross media cost (incl. platform fee) in account currency, after IVT credits. | | **What "revenue" means** | T1-pixel-attributed conversion value within configured attribution window. View-through included where enabled. | | **Cost basis** | Mostly CPM (display, video, CTV, audio); CPC for some display. ROAS is unit-agnostic. | | **Currency** | Account currency. | | **Conversion attribution** | Click-through + view-through. Generous attribution; CTV view-through can lift ROAS 30, 60% over click-only. | | **Attribution window** | 30/1 default. CTV-heavy accounts should use 30/7. | | **Bot / invalid traffic** | IVT-filtered. | | **Time window** | `30D vsP`. 1, 4 hour ingest lag (longer than peers). | | **Alert trigger** | `<2 (warn), <1 (critical)`. **Caveat**: false-positive alerts more common on MediaMath because of pixel instability. Reconcile to T1 UI before acting. | | **Sentiment key** | `{'type': 'gauge', 'thresholds': {'good': 4, 'warn': 2}}` | | **Roles** | owner, marketing, finance | ## Calculation Calculated automatically from your MediaMath data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading. ## Worked example The same enterprise homeware brand. Window 02 Apr 26 to 01 May 26. | Campaign | Channel | Spend (\$) | Total revenue (\$) | Per-campaign ROAS | | ----------------------------- | ------- | ------------ | ------------------ | ----------------- | | Display Prospecting (Brain) | Display | 24,800 | 72,600 | **2.93×** | | CTV Premium PMP, NBC | CTV | 28,400 | 38,200 | **1.35×** | | Video Pre-roll | Video | 12,200 | 23,200 | **1.90×** | | Retargeting | Display | 6,800 | 30,600 | **4.50×** | | **Account total (this card)** | | **\$72,200** | **\$164,600** | **2.28×** | Shopify-truth context: UTM-tagged MediaMath revenue \$58,400; **business ROAS = 0.81×**. Combined with TTD the cross-DSP business ROAS is closer to 1.6×. What's interesting: 1. **Account ROAS 2.28× sits in the warn band** (good ≥4, warn ≥2). For DSP this is on the lower side; the CTV-heavy mix (39% of spend at 1.35× last-click) drags the headline down. 2. **Retargeting at 4.50× ROAS** is the standout campaign, same dynamic as on every DSP. 3. **CTV at 1.35× is misleading.** CTV's job is intent-seeding; conversions appear weeks later via Branded Search or Direct. Real CTV ROAS via incrementality testing on this account historically runs 2.5, 3.5×. 4. **Prior 30D ROAS was 2.44×.** Down to 2.28×, a 7% decline. **Important MediaMath check**: does this drop correlate with a known pixel instability event? Check T1 status page or account management before treating as campaign-level issue. 5. **Compare to TTD ROAS (running in parallel during migration evaluation).** If TTD ROAS on the same audience and creative is materially better, that's the migration signal. Quick sanity tests: * ROAS down + revenue down + spend held + clicks held = MediaMath pixel issue first; campaign issue second. * ROAS down + revenue held + spend up = scaling beyond efficient frontier (standard DSP issue). * ROAS up after platform issue resolved = the prior drop was pixel-related, not campaign-related. * ROAS materially worse than TTD on like-for-like inventory = migration signal. * ROAS very low + view-through revenue zero = CTV view-through window probably misconfigured. ## Sibling cards merchants should reference together | Card | Why pair it with ROAS | | ------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------- | | [MediaMath Total Spend](/nerve-centre/kpi-cards/mediamath/total-spend) | Denominator. | | [MediaMath Total Revenue](/nerve-centre/kpi-cards/mediamath/total-revenue) | Numerator; pixel-health check important on this connector. | | [MediaMath ROAS by Campaign](/nerve-centre/kpi-cards/mediamath/roas-by-campaign) | Per-campaign breakdown; CTV at 1.3× and retargeting at 4.5× tell different stories. | | [MediaMath ROAS Trend](/nerve-centre/kpi-cards/mediamath/roas-trend) | Daily series; 7-day rolling is the actionable read. | | [MediaMath Clicks vs Conversions](/nerve-centre/kpi-cards/mediamath/clicks-vs-conversions) | The pixel-health canary, **read this whenever ROAS alerts on MediaMath**. | | [The Trade Desk ROAS](/nerve-centre/kpi-cards/the-trade-desk/roas) | Migration target peer; like-for-like ROAS comparison drives the migration decision. | | [GA4 Revenue by Channel](/nerve-centre/kpi-cards/google-analytics/revenue-by-channel) | Independent attribution check. | | [Shopify Total Revenue](/nerve-centre/kpi-cards/shopify/total-revenue) | Truth side. | ## Reconciling against the vendor's own dashboard **Where to look in MediaMath T1:** [T1 → Reports → Campaigns → ROAS column](https://t1.mediamath.com/) at organisation level. Should match this card to within sub-percent rounding once ingest catches up. **MediaMath-specific caveat**: Reconcile this card to T1 UI more often than you would on TTD or DV360. Pixel-firing instability post-restructure means both sides can briefly disagree; the source of truth is T1 UI when they differ. **Why our number may legitimately differ from T1:** | Reason | Direction | Why | | ------------------------------- | ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Time zone** | Boundary days off | Org TZ vs UTC. | | **Real-time ingest lag** | Card slightly low for "today" | 1, 4 hour lag (longer than peers). | | **Pixel ingest instability** | Direction varies | Post-restructure platform has had episodic pixel outages; both T1 UI and this card miss the same conversions but timing of catchup may differ briefly. | | **View-through window changes** | Direction depends | Window changes reflow data retroactively. | **Cross-connector reconciliation:** | Card | Expected relationship | What causes legitimate divergence | | ------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------ | | [`google_analytics.ga_revenue_by_channel`](/nerve-centre/kpi-cards/google-analytics/revenue-by-channel) | (GA4 Paid Display ÷ MediaMath spend) ≈ this card × 0.5, 0.7 | View-through inflation. | | [`shopify.total_revenue`](/nerve-centre/kpi-cards/shopify/total-revenue) | Business ROAS = UTM-MediaMath ÷ spend ≈ 0.3, 0.5× this card | Multi-touch attribution. | | [`the_trade_desk.the_roas`](/nerve-centre/kpi-cards/the-trade-desk/roas) | Peer DSP; comparable structural attribution. **The migration-evaluation comparison.** | Different inventory pools, different bidder algorithms. Within 10, 15% is normal; >25% gap on like-for-like is the migration signal. | | [`stackadapt.sta_roas`](/nerve-centre/kpi-cards/stackadapt/roas) | Peer DSP; benchmark only. | Different segment (mid-market vs enterprise). | | [`google_ads.gads_roas`](/nerve-centre/kpi-cards/google-ads/roas) | Different ecosystem; not directly comparable. | Search vs programmatic. | ## Known limitations / merchant FAQs **My MediaMath ROAS dropped 20% overnight, what should I check first?** **On MediaMath specifically: pixel health, before anything else.** The post-restructure platform has had episodic conversion-tracking instability that suppresses revenue figures and depresses ROAS. Open T1 → Pixels and check fire-rate diagnostics for the last 24, 48 hours. Only after pixel health is confirmed should you proceed to standard checks (creative fatigue, audience exhaustion, attribution window). **My MediaMath ROAS is 2.3× and TTD ROAS on the same audience is 3.1×, should I migrate?** That's a 35% gap, materially different on like-for-like. **The migration math probably justifies the move**: 35% better ROAS on equivalent spend pays back operational migration cost (typically \$20, 50k for an enterprise migration) in 2, 4 months. Caveat: ensure you're comparing like-for-like inventory and audiences; a 35% gap can also reflect different audience targeting or creative. **Is MediaMath ROAS reliable enough for board reporting?** Use with caveats. The pixel instability means individual-month ROAS can be artificially depressed; quarterly averages are more reliable. **For board reporting, I'd flag MediaMath ROAS as "directional" in the methodology footnote** and emphasise the cross-platform ROAS view (TTD + DV360 + MediaMath aggregated). Single-platform ROAS on this connector is not the cleanest read. **Why is CTV ROAS so much lower than display ROAS?** Same reason as on every DSP: CTV cannot generate clicks; all CTV attribution is view-through. With 1-day view-through, CTV is artificially compressed. Set view-through to 7 days for CTV-heavy accounts. Real CTV incremental ROAS via holdout testing is typically 2.5, 4×, but the last-click reading will always be lower. **Should I optimise Brain for ROAS or CPA?** Brain (MediaMath's bidder) has both options. **For high AOV (>$200)**: optimise for ROAS. **For low AOV (<$80)**: optimise for CPA. Mid-range: try both for 14 days each, compare against Shopify-truth. Brain's ML is mature but not as recent as TTD's Koa or StackAdapt's BidCore; expect 5, 10% performance gap on identical campaigns vs peer DSPs. **Can I trust the "today" ROAS?** Less than rolling 7-day, more so on MediaMath because of slower ingest and pixel instability. Use 7-day rolling for actionable reads. **Why does MediaMath ROAS sometimes disagree with our internal multi-touch attribution model?** Multi-touch models attribute fractionally across all touchpoints; MediaMath's view is last-paid-click within its attribution window. Disagreement is structural; on this connector, also factor in pixel-instability noise. **Use MTA for budget allocation; use MediaMath's own ROAS only for within-MediaMath optimisation decisions.** **Should I extend view-through to 14 days for higher ROAS?** You can, but the additional view-through credit is increasingly noise-dominated past 7 days. The trade-off: higher reported ROAS but less actionable optimisation signal. 30/7 is the recommended ceiling; longer windows are vanity ROAS, not decision ROAS. **My retargeting ROAS dropped on MediaMath but is fine on TTD, what's happening?** Retargeting on MediaMath depends on the same pixel infrastructure as the rest of the account; pixel issues affect retargeting just as much as prospecting. Cross-check: if TTD retargeting on the same audience pool is healthy, the issue is MediaMath-platform-side. Run pixel diagnostics first. *** ### Tracked live in Vortex IQ Nerve Centre *ROAS* is one of hundreds of KPI pulses Vortex IQ tracks across MediaMath 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.