Amazon Performance Plus: Setup, Strategy, and ROAS Gains

The Skinny

Amazon Performance+ is reshaping how brands use Amazon DSP, but you don’t need to unpack every technical detail to decide if it belongs in your plan. The points below summarize what it is, how to think about its role in your funnel, and the conditions you need in place before you turn it on.

Performance+ is an AI-powered campaign type in Amazon DSP that automates audience building and bidding across display, online video, and streaming TV while keeping critical levers like supply, brand safety, and geo in your hands.
Brands see the best ROAS when they treat Performance+ as a demand generator and a bridge between upper-funnel media and lower-funnel conversion, not just another last-click retargeting tool.
A structured readiness check—conversion volume, creative, budget, tagging, and measurement governance—determines whether Performance+ becomes a scalable always-on program or an expensive one-off test.

What Amazon Performance Plus Actually Is (Beyond the “AI-Powered” Marketing Copy)

Amazon DSP has evolved from a niche add-on for Amazon sellers into a full-funnel, omni-channel engine that can move the needle for both endemic and non-endemic brands. At the center of that evolution is Performance+, an AI-driven campaign type that uses Amazon’s shopping and streaming signals plus your own conversion data to find the audiences most likely to buy.

Amazon’s marketing will tell you Performance+ is “simpler and smarter.” For most brands, what matters more is how it changes your day-to-day planning, what you can stop doing manually, and where you still need to stay hands-on if you care about outcomes.

That’s the theory. In practice, Performance+ only pays off when you launch it with the right goals, inputs, and guardrails, then measure its impact beyond last-click ROAS. In this guide, we’ll walk through what Performance+ actually is, how it works under the hood, how to run a fit check before you turn it on, and the launch, optimization, and measurement plays our team uses to turn AI automation into durable business outcomes.

Sign Up for Tinuiti's Commerce Newsletter

Where Performance Plus Sits Inside the Amazon Ads Stack

Performance+ is a predictive campaign type within Amazon DSP, not a new ad unit. It uses Amazon’s first-party signals and machine learning to dynamically build audiences and optimize bids toward a performance KPI, typically ROAS for retail-focused campaigns or CPA for lead and subscription goals.

That makes it fundamentally different from Sponsored Products, Sponsored Brands, and Sponsored Display, which are keyword- and product-level CPC formats built to capture explicit intent on Amazon search and product detail pages. Performance+ runs wherever your audience spends time, including display placements, online video, and streaming TV inventory across Amazon-owned properties and third-party publishers connected through Amazon’s supply integrations. Instead of asking you to hand-pick dozens of static audiences and line items, it uses predictive modeling to decide which impressions are most likely to convert within the guardrails you set.

For brands that have already exhausted the obvious Sponsored Ads tactics and non-branded keyword expansions, Performance+ becomes the bridge between that high-intent aisle and a broader pool of prospects and customers across the open internet, connected TV (CTV), and premium streaming content.

The Glass Box Approach to Automation and Advertiser Control

Most marketers have been burned by “black box” automation that hides what the algorithm is doing or strips away too much control. Performance+ was built around the opposite concept: a “glass box” approach where Amazon’s AI handles the heavy lifting, but you still see and steer the strategy.

On the automation side, Performance+ takes audience modeling, line-item scaffolding, and bid optimization off your plate. Instead of spinning up separate line items for every combination of audience, inventory, and device, you declare your KPI, budget, and guardrails, and the model continuously scores users and impressions on their likelihood to drive that outcome. Over time, predictive bidding replaces the weeks of manual “test every targeting mix” work that Skai and Amazon’s teams identified as a major pain point in our joint webinar below.

Crucially, that doesn’t mean ceding control. Advertisers and agencies still decide:

Which supply to include or exclude across Amazon-owned inventory and third-party deals.
Brand safety thresholds and blocklists.
Geo, device, and dayparting strategy.
Pacing and frequency caps, including how aggressive to be with premium CTV and OLV.

Amazon provides granular reporting at the domain, app, placement type, and creative size levels, so buyers can see exactly where dollars went and adjust those levers as new insights emerge. That makes Performance+ a fit for marketers who want AI to work harder for them without giving up the supply transparency and governance their finance and brand teams expect.

Want to hear how this “glass box” feels in real life? In our Amazon & Retail Media Summit session below, leaders from Tinuiti, Amazon Ads, and Skai unpack how Performance+ automation changes their day-to-day work—what they’ve handed to the model and where they still insist on staying hands-on.

What Performance Plus Optimizes For

Under the hood, every Performance+ campaign revolves around a clearly defined performance goal. For endemic brands, that often means a ROAS target tied to on-Amazon sales; for non-endemic brands, it might be a CPA target for off-site leads, signups, or app events. In both cases, Amazon’s models ingest your conversion events and learn which combinations of audience signals, supply, and creatives drive that outcome most efficiently.

Where Performance+ sits in your broader plan depends on how you set those goals. With a ROAS or lower-funnel CPA target, it behaves like a mid-to-lower-funnel engine, turning upper-funnel awareness from Brand+, live sports, and other premium inventory into measurable conversions. With more flexible efficiency targets and a bigger budget, it can also play higher in the funnel, modeling new-to-brand audiences that your Sponsored Ads and retail media programs can later harvest.

The throughline: Performance+ is at its best when it’s not siloed as a “DSP test,” but woven into a full-funnel architecture that spans Brand+, Sponsored Ads, and Amazon Marketing Cloud (AMC).

How Amazon Performance Plus Works Under the Hood

First-Party Amazon Signals Plus Your Conversion Events

Amazon brings a uniquely rich set of first-party signals to Performance+: shopping behavior, browsing and purchase history, streaming and Prime Video consumption, and other entertainment context data across its ecosystem. Performance+ taps these signals to understand who is in-market, who’s an existing customer, and who looks like your best buyers, even if they haven’t seen your brand yet.

On your side, the inputs are conversion events: purchases, leads, signups, app events, and other outcomes you accurately define and instrument. When those events are clean and consistent, the model can connect the dots between upstream impressions and downstream results, constantly re-weighting which combinations of signals and supply are most likely to produce the next conversion. That dynamic approach is why Performance+ tends to outperform static audience targeting that relies on one-time segment definitions and quarterly refreshes.

Predictive Bidding and What the Learning Phase Actually Feels Like

Instead of targeting “all users in audience X,” Performance+ scores each impression on its likelihood to drive your chosen conversion goal at an efficient cost. The model updates frequently as new performance data comes in, shifting spend toward combinations of audience, inventory, and creative that demonstrate they can meet your ROAS or CPA targets.

From a practitioner’s perspective, the learning phase feels like a structured exploration period. In the first days and weeks, you’ll see the model test different pockets of supply and audience clusters as it builds confidence about where to lean. Results typically stabilize once the campaign has sufficient conversions and budget throughput to identify patterns. That’s why we recommend making goal and structure changes in 15–20% increments on a weekly cadence, rather than swinging targets dramatically and resetting the model’s progress.

The net effect on day-to-day management is a shift in focus. Instead of micromanaging line-item bids and audiences, your time goes into setting smart guardrails, reading signal patterns, and using insights to refine creative, landing experiences, and budget allocation.

The Three Performance Plus Tactics: Acquisition, Remarketing, Retention

Performance+ supports three core tactics that roughly map to major stages of the customer lifecycle.

Acquisition: Uses modeled expansion off seed behaviors and conversions to find net-new customers who look like your best buyers, reaching them across display, OLV, and streaming TV.
Remarketing: Re-engages site visitors, content viewers, and product browsers who haven’t yet converted, adapting bids and creative based on how close they are to purchase.
Retention: Focuses on repeat purchase, cross-sell, and lifetime value by targeting past customers with tailored messaging and suppressing recent buyers to avoid waste.

A mature Performance+ program typically runs all three tactics in concert, with budgets and guardrails tuned to your category dynamics and business goals. Acquisition drives new-to-brand growth, remarketing protects your investment in upper-funnel media, and retention protects and expands customer value over time.

The Amazon Performance Plus Fit Check

The readiness checklist before you launch

There are four non-negotiables we look at before recommending Performance+:

Conversion signal volume and consistency: Performance+ needs a steady drumbeat of conversions to learn from; if you only see a handful of conversions a week or your tagging is inconsistent, the model can’t separate signal from noise.
Creative readiness: At minimum, plan for multiple display sizes and at least one approved video asset so the model has creative coverage across key placements, especially in streaming and OLV.
Budget and learning period: You’ll need enough budget to hit meaningful conversion volume and a flight length of at least 30 days to give Performance+ a fair learning window.
Measurement infrastructure: Clean tags, deduplicated conversions, and access to AMC (or equivalent) are required if you want to prove incremental ROAS rather than just stare at platform last-click.

A simple readiness scorecard consisting of Criterion, What “Good” Looks Like, Risk if Missing, and How to Fix Fast helps teams align on whether to launch now, fix prerequisites, or stay with more traditional DSP structures for the moment.

Which Brands Get the Most Out of Performance Plus

We see Performance+ over-deliver for brands that check a few boxes:

Stable conversion goals and products: Categories with reasonably consistent demand and clear conversion events give the model the clean feedback loop it needs.
Full-funnel ambitions: Brands that want to connect premium awareness (live sports, CTV, OLV) to lower-funnel sales, and are willing to use AMC and other tools to see beyond last-touch metrics, unlock more value.
Operational appetite for automation: Teams that are comfortable letting the model handle line-item scaffolding and audience building so they can reallocate time to strategy, creative, and measurement see the biggest efficiency gains.

If your organization is already running Sponsored Ads at scale, has a baseline DSP program, and is investing in AMC, Performance+ is a natural next step rather than a leap into the unknown.

When to Wait or Skip Performance Plus Entirely

There are also scenarios where we recommend pressing pause:

Low conversion volume or erratic seasonality without enough runway for the model to learn, such as ultra-short flights or one-week tests.
Measurement constraints, including unclear attribution ownership, broken tags, or an inability to see off-Amazon conversions that matter to the business.
Over-restrictive brand safety and inventory rules that starve the model of scale and force it to hunt for performance in a tiny corner of the available supply.

In those cases, it’s usually better to shore up signals, creative, and governance first before layering in Performance+.

Launching Amazon Performance Plus Ads Without Sabotaging the Learning Phase

Once you’ve passed the fit check, launch discipline is what separates stable performance from “we tried it once and it didn’t work.”

The Pre-Launch Checklist

Before you activate a single tactic, work through a tight pre-launch QA:

Tag audit: Confirm that all relevant pixels and server-side events are firing, that matchback windows are appropriate for your buying cycle, and that conversions are deduplicated across on-Amazon and off-Amazon sources.
Creative upload: Load and QA all required display sizes and video assets, including brand-safety checks and alignment with Amazon’s creative policies for CTV and OLV.
Goal setting: Set ROAS or CPA targets with a 20–30% margin above your minimum acceptable floor so you don’t choke delivery before the model learns how to find efficient conversions.
Audience exclusions: Suppress recent converters and, where appropriate, existing customers to keep acquisition honest and avoid paying again for people you already won through other channels.

Codifying these steps in a runbook pays for itself in fewer false starts and cleaner tests.

Structure Decisions: Single-Tactic vs. Multi-Tactic and Budget Allocation

The next decision is structural: do you bundle acquisition, remarketing, and retention into a single campaign or separate them?

Single-campaign setups can work for smaller budgets where you want the model to allocate flexibly across tactics while you watch how spend and performance shake out.
Separate campaigns by tactic give you more control over budgets, pacing, and guardrails at each stage of the funnel, which is often preferable as investment grows.

Either way, we recommend:

Ensuring prospecting gets a meaningful share of budget so Performance+ can build new-to-brand audiences instead of just circling existing customers.
Committing to a minimum 30-day flight to allow for exploration and stabilization.
Reviewing what the automated setup creates before launch so you can adjust geo, frequency caps, and inventory options where needed.

That balance preserves the benefits of automation while keeping your media strategy intentional.

Optimizing Amazon Ad Campaign Performance When Performance Plus Stalls

Even well-set-up Performance+ campaigns can hit plateaus. The key is to read symptoms correctly, diagnose likely causes, and pull the right levers without resetting the model every week.

Reading the Signals: Symptom, Cause, Lever

Once Performance+ is live, the real work is interpreting what the platform is telling you. Instead of reacting to every fluctuation, we translate campaign behavior into a simple diagnostic framework: Symptom → Likely Cause → Lever.

The table below outlines some of the most common patterns we see in Performance+ campaigns and the first levers we recommend testing when they show up in your reports.

Symptom	Likely cause	Lever to pull first
Spend will not scale	ROAS/CPA target is too aggressive, frequency caps are too tight, or inventory settings are overly narrow.	Loosen goals by 15–20%, widen eligible supply, and ease frequency caps to give the model room to explore.
CPA spikes or ROAS flatlines	Creative fatigue, landing-page friction, or noisy/duplicate conversion events.	Refresh creatives, QA site speed and UX, and audit conversion tagging for duplicates or misfires.
Remarketing saturates	Over-indexed on lower-funnel tactics with too few net-new users entering the funnel.	Shift budget toward acquisition tactics and upper-funnel supply to feed more new-to-brand prospects.
Streaming TV spends without visible site conversions	Misaligned expectations; CTV impact is showing up as incremental lift and assisted conversions, not last-click sales.	Use AMC path-to-purchase, assisted conversion, and LTV analyses; add holdout or geo-lift tests where possible.

The Goal-Target Rule: Change by 15 to 20 Percent, Not by Panic

When performance drifts, the impulse is to slash or double targets overnight. That instinct is exactly what keeps Performance+ stuck in a permanent learning loop.

Instead, we recommend:

Adjusting ROAS or CPA targets in 15–20% increments so the model can adapt without treating every week as a brand-new campaign.
Making weekly, not daily, goal changes, giving enough time between adjustments to see whether the previous tweak improved stability and efficiency.
Documenting each change and its rationale so you can correlate shifts in performance with specific decisions rather than guessing after the fact.

That discipline usually produces better long-term ROAS than reactive “panic editing.”

The Four Pitfalls That Quietly Break Performance Plus Campaigns

Four issues come up again and again in underperforming Performance+ programs:

Bad conversion signals: Mis-tagged events, double-counted conversions, or goals that don’t match business value teach the model to optimize for the wrong outcomes.
Over-constrained inventory and brand safety: Starting with highly restrictive supply and strict brand-safety settings can prevent the model from discovering efficient performance pockets.
Set-and-forget mindset: Performance+ reduces manual work, but it still needs a human to develop a strategy around creative, messaging, budgets, and measurement.
Last-click measurement: Holding Performance+ to a pure last-touch standard ignores its role in demand generation and upper-funnel influence, especially in CTV and OLV.

Fixing these issues often yields more improvement than any tweak to bids or budgets.

Measurement That Proves Amazon ROAS Improvement

The Reporting Views That Matter Weekly vs. Monthly

Not every metric belongs in every meeting. We split Performance+ reporting into:

Weekly views: Domain and app delivery, creative performance by format, geo performance, and tactic-level outcomes (acquisition vs remarketing vs retention) to catch issues early.
Monthly and quarterly views: Comparisons against legacy DSP orders and prior periods, plus deeper insight from AMC and holdout tests to understand incrementality and long-term contribution.

That cadence keeps the team focused on structural improvements rather than chasing every short-term fluctuation.

Amazon Marketing Cloud as the Measurement Backbone

In the webinar, both Amazon and Tinuiti describe AMC as the backbone of decisioning for DSP strategies, including Performance+. AMC unlocks:

Path-to-purchase analysis, showing how Performance+ impressions interact with Sponsored Ads, retail media, and other channels on the way to a sale.
Lifetime value views, helping you understand whether Performance+ is bringing in higher-value customers who buy more often or across more categories.
Cross-channel insights, including off-Amazon conversions (like DTC sales) for brands that have instrumented those events properly.

At Tinuiti, we use AMC and clean-room integrations as the core of a measurement governance cadence: weekly insight reviews feeding monthly strategy changes, with shared views across ecommerce, media, and finance stakeholders. That structure gives leaders confidence to keep investing when the data shows Performance+ is driving incremental growth.

Platform ROAS vs. Blended ROAS vs. MER

Performance+ will show one set of numbers in the Amazon DSP UI; your business will care about a broader set of economics.

Platform ROAS reflects conversions that Amazon attributes to Performance+ based on its default lookback windows and attribution model.
Blended ROAS looks at revenue or margin across a cluster of channels—for example, DSP plus Sponsored Ads—to see whether the combined mix is getting more efficient over time.
MER (marketing efficiency ratio) zooms all the way out to revenue divided by total media investment, which is often how finance leaders judge success.

Because DSP’s influence is often upstream and multi-touch, last-click platform ROAS alone tends to undervalue it. Layering in holdout groups, geo-based tests, and AMC analyses gives you a more accurate sense of how much Performance+ is contributing, so you can scale with conviction.

The Full-Funnel play: Where Performance Plus Fits Alongside Brand Plus and Sponsored Ads

Brand Plus for Awareness, Performance Plus for Conversion

Brand+ and Performance+ share the same AI foundation inside Amazon DSP but optimize for different KPIs. Brand+ is built for upper- and mid-funnel outcomes such as reach, awareness, and engagement; Performance+ is tuned for conversions and revenue.

Together, they allow you to:

Use Brand+ and premium supply (like Thursday Night Football or NBA inventory) to build large, high-quality audiences.
Hand those audiences to Performance+ to drive consideration, site visits, and purchases with more performance-oriented creatives and guardrails.

Both products use the same glass-box transparency and control model, so you maintain a consistent approach to supply, brand safety, and measurement across the funnel.

Pairing DSP Automation With Sponsored Products and Sponsored Display

DSP and Sponsored Display Ads are complementary, not competitive. A healthy architecture typically looks like this:

DSP and Performance+: Drive discovery, prospecting, and retargeting across the open internet, OLV, and CTV, feeding more people into your Amazon retail and DTC ecosystems.
Sponsored Products and Sponsored Display: Capture high-intent demand on Amazon search and PDPs, capitalizing on awareness and interest generated upstream.

Insights should flow both ways. Creative and messaging that prove out in Performance+ video can inform Sponsored Brands assets; high-performing search terms and audience signals from Sponsored campaigns can become inputs for contextual and audience strategies in DSP. Budget planning then becomes a question of protecting bottom-funnel efficiency while ensuring you fund enough upper- and mid-funnel activity to keep the demand pipeline healthy.

The Demand-Generator Argument (Don’t Turn DSP Into a Glorified Retargeting Tool)

In the webinar, our panelists didn’t mince words: treating DSP as a pure retargeting channel is overhyped and under-ambitious. Retargeting is a fine starting point, but when it becomes the center of your media strategy, you cap long-term growth.

Performance+ shines when you:

Use it to generate demand, not just capture it, by leaning into acquisition tactics and contextual strategies that find people before they hit your PDP.
Optimize for new-to-brand growth and incremental revenue, not just cheap last-click conversions from people who were already going to buy.
Pair remarketing with broader mid-funnel plays—like targeting related products, keywords, and content environments—so you build a richer pool of future searchers and shoppers.

That mindset shift is often the difference between DSP as a cost center and DSP as a growth engine.

The Underutilized Performance Plus Plays

A few Performance+ strategies come up repeatedly in our work and in the panel conversation as underutilized:

Premium content retargeting: Using live sports and tentpole content—Thursday Night Football, NBA, WNBA, the Olympics—as upper-funnel seed inventory, then retargeting exposed audiences with Performance+ to drive consideration and conversion.
Online video undercutting: For brands that can’t win on CPC against category giants, building audiences around competitor traffic and using OLV to generate cheaper glance views and downstream conversions.
Funnel-stage messaging maps: Aligning creative and offers to acquisition (education and problem-solution), remarketing (proof, reviews, urgency), and retention (loyalty, cross-sell), and letting Performance+ deliver the right story to the right audience at the right depth.

These plays require a bit more upfront strategy and creative planning, but the payoff in incremental ROAS and new-to-brand growth is often significant.

Scaling Amazon Performance Plus With a Partner Who’s Been There Since the Beta

Performance+ gives you a glass-box view into what Amazon’s AI is doing on your behalf, but that transparency only creates value if someone is actually reading the signals and pulling the levers.

Tinuiti was among the early beta-testing partners for Performance+, and we’ve since scaled it across categories with a full-funnel lens that spans DSP, Sponsored Ads, and AMC. That vantage point lets us pressure-test readiness before launch, establish governance and measurement frameworks that CFOs can trust, and architect media plans in which Brand+, Performance+, and Sponsored Ads work together rather than compete for budget.

If you’re evaluating Performance+ for your next growth phase or looking to move beyond a one-off test, the next step is a tailored fit check on your signals, creative, budget, and measurement so you know exactly what Performance+ can unlock for your business.