RTB House does product-based dynamic retargeting like Criteo, but it builds its bidding engine around a different engineering decision. That difference isn’t a marketing footnote: RTB House trains a separate deep-learning model for every advertiser, rather than running one shared model for all advertisers.
How Does RTB House Work?
Most programmatic ad platforms pool every advertiser’s data and process it in a single general model; that lets a new advertiser’s campaign “warm up” quickly, but it limits how much the model can learn about that specific brand’s behavior patterns. RTB House does the opposite: for each customer, it builds a separate deep-learning model trained only on user behavior from that brand’s own site.
What this model produces isn’t a single bid figure. It predicts which product, shown at which moment, in which ad format, gives a given user the highest likelihood of converting; the bid, the creative choice, and the timing of the impression are all decided together from that prediction. Training a model per advertiser needs more data and more time for the campaign to mature; in exchange, it carries the potential to produce more accurate bids on sites with high traffic and a clear conversion pattern.
RTB House’s Pros and Cons
| Pro | Con |
|---|---|
| A brand-specific model can produce more accurate predictions on high-traffic sites | The model needs more data and time to mature |
| Transparency reporting (viewability, brand-safety metrics) is a standout feature in the sector | Low-traffic sites may not generate enough data for the model |
| Ads update automatically when the product feed updates | Setup and model training can take longer than Google/Meta remarketing |
| Campaign performance keeps improving with its own data past a certain point | Pricing is negotiated, no fast self-service start is offered |
How Does the Pricing Work?
Like Criteo, RTB House doesn’t publish a public price list and runs on a model set up through a sales conversation, typically performance-based (CPC or close to a target CPA). The core variables that determine price are the site’s traffic volume (which affects how much data the model needs to learn), the size of the product catalog, and the number of markets targeted. Because the model is trained specifically for each advertiser, on a low-traffic site the model may take longer to mature at the same budget compared to Criteo — which can affect early performance.
Who Does RTB House Fit?
RTB House makes sense for mid-to-large e-commerce sites with high traffic and conversion volume, where a brand-specific model can make a difference over time. It particularly suits advertisers who value brand-safety and viewability reporting and can treat the campaign as a long-term investment. For a low-traffic new site or a short-term campaign expecting fast results, Criteo, which runs on a shared model, or Google Ads’ own dynamic remarketing feature usually warms up faster.