Inside the Blurb · The MarTech Matrix

Why Your Retargeting Partner Is Only Reaching 15% of Your Shoppers

The big picture: Most ecommerce brands believe their retargeting is handled. They’re running Criteo or Performance Max, hitting their ROAS targets, and moving on. What the numbers actually show is that their platform is reaching 15–20% of the visitors who bounced from their site — and leaving the other 80–85% behind.

That gap is the starting point for this conversation with Jason Blom, VP of Strategic Opportunities at RTB House and a former Criteo sales leader. In this episode of Inside the Blurb, Jason breaks down what deep learning retargeting actually does differently, why 60% of RTB House-driven purchases come from products shoppers never viewed, and how to set up a clean incrementality test before you add a second retargeting partner.

Why it matters: The difference between 15% and 35% reach on your bounced visitors isn’t marginal. It’s a meaningful slice of revenue that your current setup structurally cannot recover — and no optimization of your existing partner will fix it.

---

The Problem With “We Already Have Retargeting Covered”

The most common objection Jason hears from brands is also the most expensive one. “We already have retargeting covered. We run Criteo — or Performance Max — and we don’t need to do more here.” The problem is that this assumes one partner can saturate the eligible audience. None can.

Budget constraints and ROAS performance guardrails mean that any single retargeting bidder — regardless of how good it is — tops out at roughly 15–20% reach on visitors who bounced from a site. A million shoppers hit your site; your retargeting platform serves impressions to 150,000–200,000 of them. The other 800,000 never see your brand again.

“Whether it’s Google or Criteo or Meta or RTB House, you’ll never achieve 100% reach rate on your eligible audience,” Blom said. “The performance constraints just won’t allow you to invest as much into getting those impressions out there.”

This isn’t a failure of execution. It’s how the math works at ROAS-managed budgets. And understanding it changes the conversation from “should I replace my retargeter” to “how do I reach the audience my retargeter structurally can’t get to.”

---

What Is Deep Learning Retargeting — and How Is It Different From Machine Learning?

Machine learning has powered retargeting for roughly 20 years. Google, Criteo, Meta — all built their performance infrastructure on machine learning algorithms that automate audience identification, bidding, and creative delivery. It works. It’s also reached its ceiling.

Deep learning retargeting runs on a more powerful subset of that same foundational logic — with significantly more processing power. What that unlocks in practice is the ability to ingest more first-party signals, test more product and creative combinations in real time, and make better bidding decisions further out from the original bounce event.

“When you talk about agentic AI and all the buzzwords around ChatGPT — deep learning is really just more processing power,” Blom said. “And the reason we want it is to create more efficiency and higher incrementality with the media spend.”

The output shows up in three concrete ways: impression timing, product recommendation, and creative quality. Standard retargeting platforms front-load impressions immediately after a bounce and taper off. RTB House elongates the delivery window across days — which matters because many purchase decisions happen on a different timeline than the original bounce.

Why Standard DSPs Can’t Do What Deep Learning Does

Standard self-serve DSPs — The Trade Desk, DV360, even Yahoo — are rules-based. You manually curate audiences, create deal IDs, select PMPs, set CPMs, and define frequency caps. That’s an inherently inefficient way to run performance campaigns, and it rarely produces strong ROAS in lower-funnel retargeting.

Criteo and Google are better because their systems automate most of that decisioning. But their impression dispersion still clusters right after a bounce — they’re good at pulling someone back immediately, then the targeting stops. RTB House’s deep learning engine is designed to yield-manage media delivery across a longer window, continuously testing offer and product combinations to find what actually converts.

---

Why 60% of Purchases Come From Products Shoppers Never Viewed

This is the counterintuitive stat at the center of the RTB House value proposition: more than 60% of purchases driven by the platform involve products the shopper never previously viewed or added to cart. For a channel traditionally associated with “you looked at these brown shoes, here are those same brown shoes,” that number reframes what retargeting can actually be.

The explanation starts with how people shop online. When someone visits an ecommerce site, they search for what they already know they want. They’re not browsing a wall of products the way they would in a store — they’re navigating toward a known target. They may never discover that there’s a green-and-white shoe that fits them better than the one they searched for, because they never went looking for it.

“The consumer isn’t just going to a website and then bouncing off — it takes multiple visits,” Blom said. “You have to constantly be in front of them with some type of advertising and product recommendation. Our technology is designed to test those offer and product recommendations in real time, leveraging AI, to get the consumer to engage.”

What drives that 60% number is continuous real-time testing against first-party engagement data. The deep learning engine cycles through product assortments — new items, different colors, different price points, different categories — and finds the combination of timing and offer that brings a shopper back. Retargeting the exact item they carted and abandoned is often the wrong answer, because there’s usually a reason they left without buying it.

---

What RTB House Collects That Standard Retargeters Don’t

Standard retargeting implementations track five events: homepage, search results page, product detail page, cart, and conversion. It’s a clean setup. It’s also incomplete.

RTB House’s integration collects additional on-site engagement signals — email landing pages, wish lists, favorites, blog content, and other events that indicate intent but don’t fit neatly into the standard five-pixel framework. App signals are also pulled in broadly, which most DSPs don’t do. Criteo handles app in a limited way; Meta is primarily app. RTB House tries to work across all of them.

“We want to create a better picture for the AI of what that user’s experience is with that website,” Blom said. “If we can identify somebody that’s more engaged — a super customer — then those are good events to decision off of.”

The richer signal feeds the deep learning retargeting engine and produces better bidding decisions — especially for the elongated targeting windows that run days after a bounce, when the original last-click signal from standard platforms has already gone cold.

---

How to Set Up a Clean Incrementality Test With a Second Retargeting Partner

The right question when adding a second retargeting partner isn’t “will there be overlap?” — there will always be overlap. The right question is whether that second partner is driving incremental revenue above what your existing stack would have produced anyway. That’s what the test is designed to answer.

Blom’s recommended approach: start at 20–30% of your primary partner’s budget. If you’re spending $100K/month with Criteo, bring RTB House in at $20–30K and let them hit standard attributed metrics first — last click, post-click, view-based. That phase typically takes 30–60 days.

After 60 days, deploy the incrementality test. RTB House uses a ghost ads methodology — a standard test/control setup comparing exposed vs. unexposed populations — to measure conversion lift and revenue lift. The test typically resolves within 90 days from the start of the campaign.

The presence of Criteo, Meta, or any other active partner during the test doesn’t compromise the results. Your baselines are already established with existing partners. What RTB House needs to demonstrate is lift above those baselines — not performance in isolation.

“The most important metric is: can you prove incrementality?” Blom said. “That’s really what this comes down to with the vast majority of our customers.”

---

The Creative Philosophy Behind RTB House Ads

Retargeting has a reputation for ugly ads. Product shots on a white background, three-across carousels, the same item you looked at yesterday following you across the internet. It worked — barely — because it was automated and cheap to produce. But as performance marketers increasingly report to CMOs who value brand experience, “it converts” is no longer enough if it makes the brand look cheap.

RTB House’s approach: high-density product carousels still perform best from a pure conversion standpoint, but they’re designed to incorporate video, static brand imagery, and templated brand moments. The goal is an ad that feels like it belongs in the feed — not a DPA unit that screams “you left something in your cart.”

“Meta has done such a good job of blending video into a product carousel into other potential static content,” Blom said. “Our philosophy is how do you blend that? How do you actually work with your customers to create the right balance between showing product and showing brand elements? We definitely templatize most of the creative, but we offer more templates that have branding moments so that it creates better-looking ads.”

The practical implication: when getting started with RTB House, bringing video assets — even existing brand video — meaningfully improves creative performance. The platform will blend them into the product recommendation carousel. If you only have static images, they’ll work with that too, but video + product rec is the combination that tends to perform best.

---

How RTB House Becomes the Only Partner — and When That Makes Sense

Several RTB House customers that started deep learning retargeting as a complementary add-on to Criteo have since consolidated their lower-funnel budget entirely into RTB House. The path there is almost never a head-to-head competition. It’s an accumulation of trust built through performance, creative experience, and service quality.

Performance drives the decision, but it’s not the only thing. CMOs care about what their ads look like. They’ll sometimes sacrifice conversion rate for brand experience — which means creative quality is a competitive factor that shows up in budget decisions even when it doesn’t show up in ROAS reports.

Service levels matter too, in a way that doesn’t get discussed enough. Campaigns break. Pixels go down. Measurement stops working. A retargeting partner that identifies problems proactively and surfaces them before the client notices builds a different kind of loyalty than one that waits to be asked.

“If you can solve for performance, then solve for the gaps with things like brand experience and better service levels — that’s the winning formula for creating partnership,” Blom said.

---

The Question Most Brands Forget to Ask

When we asked Jason what question brands most commonly fail to ask when evaluating a retargeting partner, his answer cut to the core problem with how the industry evaluates anything.

“What’s in it for me two or three years down the road?” he said. “The problem with the way we operate as an ecosystem is that it’s so focused on near-term performance — how are you going to help me with my low-funnel ROAS — as opposed to how can you partner with me to drive better performance over time from a full-funnel standpoint?”

It’s worth naming honestly: this conversation focused on retargeting and lower-funnel performance because that’s where RTB House’s strongest differentiation lives today. The broader full-funnel story is there — do your homework — but it starts with performance. That’s the proof that earns the longer relationship.

No meaningful change to a performance stack can be fully evaluated in 90 days. The brands getting the most out of RTB House are the ones willing to measure a longer arc and hold both themselves and their partners accountable to it.

---

The Bottom Line

If your retargeting partner is hitting ROAS targets, that’s not the finish line — it’s the baseline. The real question is how much of your eligible audience you’re actually reaching, and whether you’re recovering the shoppers that your current setup structurally cannot get to.

Deep learning retargeting isn’t a replacement for what’s working. It’s an additive layer that extends reach, improves product recommendation through real-time testing, and proves its value through incrementality — not overlap arguments. The starting point is simple: pixel the site, bring in a conservative test budget, and let the 90-day incrementality test do the talking.

If you’re evaluating RTB House — or any lower-funnel partner — check out their profile on Blurbs for case studies, vendor Q&As, and the key questions you should be asking before you take a demo call.