Dynamic Product Ads: automatically showing the right products in 2026

Imagine a visitor browsing three electric folding bikes on E-4motion.com, closing the site, and later that evening scrolling through Instagram. At that moment, an ad automatically appears showing exactly the three models they just viewed, complete with current pricing and availability. This is Dynamic Product Ads in action. Dynamic Product Ads (DPAs) are personalised ads that automatically display the most relevant products to the right user, at the right moment, via a live connection to your product catalogue. In this article, we explain what Dynamic Product Ads are, when they work best, how to set them up effectively, and how AdBrains uses proprietary AI technology to take DPA campaign performance to a structurally higher level.

What are Dynamic Product Ads?

Static ads vs. Dynamic Product Ads

Comparison of performance indicators in e-commerce campaigns

Static ads

• Manually select products per ad
• Fixed images and copy
• No personalisation at user level
• High management burden for large catalogues
• Outdated offers when inventory changes

Dynamic Product AdsRecommended

• Automatically most relevant products per user
• Real-time product feed integration
• Personalisation based on browse and purchase behaviour
• Scalable to thousands of products simultaneously
• Always up to date via live catalogue connection

Dynamic Product Ads are a campaign format within Meta Ads (Facebook and Instagram) where ads are dynamically assembled based on a product catalogue and the behaviour of individual users. Unlike static ads, where you manually select one product or offer, a DPA campaign lets the system automatically decide which product is most relevant for each user.

The technical foundation is a product feed: a structured file (XML, CSV, or via a direct API connection) containing all products with attributes such as name, price, image, availability and URL. Meta links this feed to the ad auction and combines it with data from the Meta Pixel or the Conversions API to determine which products a specific user has viewed, added to their cart, or almost purchased. Based on this, the algorithm automatically selects the most promising products per user and assembles the ad in real time.

DPAs are typically deployed in three strategic contexts:

• Retargeting: users who previously viewed a product but did not purchase are re-engaged with those same or related products.
• Cross-selling and upselling: users who have already made a purchase are shown complementary products that match their earlier buy.
• Broad audience (prospecting): DPAs are deployed to cold audiences, where the algorithm selects the most appealing products per user based on interest and behavioural data.

It is precisely these three layers that make DPAs so powerful: with one campaign structure, you can serve the entire funnel, from initial product exploration to repeat purchase.

Why Dynamic Product Ads are so effective for e-commerce

Dynamic Product Ads: conversion funnel

Average throughput per stage (e-commerce Meta Ads benchmarks, 2026)

Impressions

Users see a DPA in their feed

Clicks

Average DPA CTR: approx. 5%

Product page visit

Visitors view the product

Conversions

Purchase or lead form completed

Higher relevance = higher throughput per stage

The effectiveness of Dynamic Product Ads comes down to personalisation at scale. Where a traditional campaign for a webshop with a thousand products would require hundreds of manually managed ad sets, a DPA campaign needs just one well-configured structure. The system does the rest. This delivers not only a significant time saving, but also structurally better results, because every user sees an ad tailored to their specific interests and purchase intent.

Consider ToetsJeKennis.nl, a platform with dozens of online exams and courses. Without DPAs, all visitors would see generic ads showing the entire course catalogue. With Dynamic Product Ads, a visitor who browsed the driving theory page automatically sees an ad for that course with the current price and a direct call to action. A visitor who viewed an IT certification course sees exactly that course. This level of relevance significantly increases click-through rates and the likelihood of conversion.

A similar dynamic applies at E-4motion.com. The range of electric folding bikes includes multiple models at various specifications and price points. A DPA campaign ensures that someone who viewed the lightweight city model is not retargeted with the heaviest all-terrain model, but with the model most aligned with their browsing behaviour. This improves not just CTR, but the overall ROAS of the campaign.

Technical requirements for a successful DPA campaign

Before you can run Dynamic Product Ads, a number of technical building blocks need to be in place. Missing any one of these directly impacts performance. Below is an overview of the required and strongly recommended components:

A high-quality product feed is the foundation of every DPA campaign. Incomplete or outdated feeds directly result in less relevant ads, higher CPA and lower ROAS. Feed management therefore deserves ongoing attention, not just at the initial setup stage.

How AdBrains AI optimises Dynamic Product Ads

At AdBrains, we have developed proprietary AI technology that manages DPA campaigns at a fundamentally different level than standard manual management or Meta's default settings. While most agencies rely on Meta's own automation after the initial setup, AdBrains goes further with multiple specialised AI modules working in continuous coordination.

Our multi-agent verification system reviews every optimisation decision before it is executed. Four independent AI agents assess each adjustment, from budget shifts to audience changes, and only approve execution when the decision is consistently rated as positive across all agents. This prevents the impulsive or erroneous changes that regularly occur with manual management, especially in the dynamic environment of DPA campaigns where stock levels and prices change continuously.

Our automatic tCPA/tROAS optimisation adjusts bidding strategies daily based on current conversion volume and the margin targets per client. In DPA campaigns this is particularly valuable because performance can vary significantly across product categories. Our system detects which product sets generate the best ROAS and automatically shifts budget toward the most profitable segments.

We also deploy server-side signal enrichment through our own sGTM infrastructure. The Conversions API is enriched with first-party data, giving Meta a substantially more complete picture of the conversion journey. This is critical for DPA campaigns, as the product selection algorithms depend heavily on the quality of the signals they receive. More and richer signals lead directly to better product selection and higher relevance per user. In practice, advertisers with full server-side tracking see an average of 23% more conversions tracked, which steers the algorithm structurally better over time.

Finally, our audience management automation keeps PROD, Incubator and RLSA audiences updated on a weekly basis. For DPA campaigns, this means retargeting audiences are always current and fresh product interactions are immediately translated into updated audience segments. No manual intervention is needed to place the most recent visitors in the correct DPA audience. Our approach to campaign management is explained in detail on our methodology page.

DPA strategy: retargeting, prospecting and everything in between

Performance improvements after implementing Dynamic Product Ads

Average improvement vs. static campaigns (e-commerce benchmarks 2026)

CTR increase

42%

ROAS improvement

35%

CPA reduction

28%

Conversion value increase

31%

A common mistake when setting up Dynamic Product Ads is limiting the strategy to retargeting alone. Retargeting is indeed the most direct application, but DPAs are equally powerful as a prospecting tool for new customers, provided they are configured correctly.

The recommended strategy structure for most e-commerce advertisers in 2026 consists of three layers:

1. Retargeting (high intent): users who viewed a product or added it to their cart within the past 7 to 14 days. This is the most direct conversion phase, with the highest ROAS. Use a separate DPA campaign or ad set, with a tailored message that adds urgency or social proof.
2. Warm prospecting: lookalike audiences based on existing buyers, combined with DPA product selection. Meta uses the purchase behaviour of your best customers to determine which products are most relevant for similar users.
3. Broad prospecting: DPAs based on interest and behavioural targeting without specific pixel data. The algorithm selects products that match the broad interests of the target audience. This works best for advertisers with a wide assortment and sufficient historical data in the catalogue.

For Elletens.nl, a fashion webshop with an extensive product range, this three-layer approach means retargeting DPAs recover abandoned carts, while prospecting DPAs bring in new visitors through relevant product recommendations based on their style preferences. Combining both layers within one overarching campaign structure consistently delivers a lower CPA across the full funnel.

Feed optimisation: the difference between average and excellent

The quality of your product feed largely determines how effectively the DPA algorithm can do its job. Meta uses product attributes not only for ad display, but also as signals for relevance ranking. The richer and more accurate the feed, the better the product selection per user.

The most impactful improvements to a product feed are:

• Optimise product titles: lead with the most search-relevant attribute (e.g. brand, model, colour), not the internal product code.
• Expand descriptions: add specifications, benefits and use cases. This strengthens contextual matching.
• Google Product Category and custom labels: use these fields for granular segmentation and budget differentiation per product group.
• Image quality: use consistent, high-resolution product images. Lifestyle imagery outperforms plain white backgrounds in many categories.
• Price accuracy: ensure the price in the feed always matches the landing page. Discrepancies lead to ad disapprovals and lower relevance scores.
• Keep stock status current: set a feed refresh of at least once per day. Out-of-stock products that continue to be advertised waste budget without delivering conversions.

For HACCP-cursus.com, an online food safety training platform, feed optimisation translates to clear course titles with audience-specific descriptors (e.g. "HACCP Basics for hospitality staff" rather than a generic course code), current pricing and clear availability per start date. This drives higher relevance scores and a better match between ad content and user intent.

Measuring and optimising DPA performance

Dynamic Product Ads generate a rich dataset that goes beyond standard campaign metrics. In addition to the usual KPIs such as CTR, CPA and ROAS, there are DPA-specific dimensions that provide valuable optimisation insights:

• Performance per product set: which categories or brand groups deliver the best ROAS? Split budgets based on these insights.
• Frequency per audience segment: high frequency in the retargeting layer indicates ad fatigue. Refresh the creative setup or narrow the audience.
• Attribution window analysis: DPA conversions are partly determined by the chosen attribution window. Compare results across 1-day click, 7-day click and 28-day view to get a realistic picture of contribution to total revenue.
• Audience overlap check: ensure retargeting and prospecting audiences do not overlap, to prevent bidding against yourself in the auction.

A solid conversion tracking setup is essential here. Without accurate purchase and add-to-cart events, the DPA algorithm loses its steering capability. Server-side tracking via the Conversions API is the standard for every serious DPA advertiser in 2026, partly due to increasing limitations on browser-based pixels caused by cookieless browsing and privacy legislation.

Frequently asked questions about Dynamic Product Ads

What is the minimum product count for an effective DPA campaign?

There is no official minimum, but in practice DPA campaigns work best with an assortment of at least 20 to 30 products. With a smaller range, dynamic product selection is limited and it is often more efficient to use manually assembled carousel or collection ads. The larger the assortment, the greater the advantage of DPAs, because the system has a broader pool to match against individual user intent.

Do Dynamic Product Ads work for services and online courses?

Yes, DPAs are not exclusive to physical products. Platforms like ToetsJeKennis.nl and HACCP-cursus.com use catalogue-based ads for their online course offerings. The product feed in that case contains course data rather than physical product information. The principle of dynamically matching user behaviour with the most relevant offering works equally well in online education and digital services.

What is the difference between Dynamic Product Ads and Advantage+ Shopping Campaigns?

Advantage+ Shopping Campaigns (ASC) are a full campaign type where Meta automates virtually all targeting, budget and creative decisions. DPAs are a specific ad format that can be used within various campaign types, including ASC. They are complementary concepts: you can use DPAs as an ad format within an ASC structure, where Meta determines who sees the ad, while the DPA format determines which product is shown. Many advanced e-commerce advertisers combine both for maximum funnel coverage.

How long does it take for a DPA campaign to perform optimally?

Like all Smart Bidding-based campaigns, a DPA campaign requires a learning phase of typically 7 to 14 days. During this period, the algorithm collects sufficient data to develop a stable product selection and bidding strategy. It is strongly inadvisable to make major changes to campaign structure, audiences or budgets during this learning phase, as this restarts the learning period and delays optimisation. After the learning phase, performance is typically significantly more stable and predictable.