Skip to main content

Recommendations & Ad Networks

Written by Kevin Jabbour

Summary

  • You can now connect your ad networks to Ktrl so it understands your live bids and reacts to them.

  • Previously, Ktrl predicted long term ROAS and gave bid target suggestions as a reference point for your UA team to act on manually.

  • With your live ad-network bids connected, Ktrl goes further: it returns refined, "paced" bid targets that respect each network's auction dynamics, so every recommendation is ready to apply.

Why connect your ad networks

  • Bid targets are immediately actionable. No more second-guessing whether a change is too big or too fast. Pacing manages the timing and size of every bid change for you.

  • Smarter, safer changes. Bid targets are moderated, or “paced”, based on Ktrl’s analysis of over $6bn of UA spend. This means your bids benefit from more data and context than any single account holds.

  • Quick to set up. Connecting all 6 major networks usually takes under 15 minutes.

How to integrate ad networks

Recap: How Ktrl Optimises ROAS campaigns

  • ROAS campaigns ask an ad network to find users who reach a target ROAS by a certain day since install (DSI), typically D7 or D28. That required ROAS acts as your bid target.

  • Ktrl predicts the ROAS curve of your most recent cohorts at your Target DSI (e.g. currently predicted to hit 80% at D180 → solid line).

  • From that, Ktrl derives the optimal ROAS curve needed to hit your target (e.g. a target of 100% at D180 means shifting the predicted line up 20 points → dashed line).

  • Ktrl translates that optimal curve into a bid target (green dot) that signals the user quality you want, and the network optimises toward it.

Introducing Pacing

  • Pacing closes the loop. Ktrl ingests your live bid target, then calculates the optimal paced bid target to reach your ROAS goal. It does three things:

    • Changes at the right speed and cadence: Bids move no more often than every 7-14 days depending on the network.

    • Changes of a realistic size: Each step is capped at a ±20% move from your live bid, so no network-breaking swings in either direction.

    • Accounts for each network's behaviour: A network's bid target isn't always the same as realised ROAS, so Pacing applies the direction and magnitude of change Ktrl predicts from realised performance to your live bid instead.

  • For each campaign card, you can track your paced target against the live bid, and the raw recommendation vs predicted ROAS at the Optimisation DSI it's derived from over time on the Recommendation History chart.

  • Note: Pacing currently only applies to ROAS campaigns on connected networks. Other campaign types (e.g. CPI, CPE) and networks continue to show standard recommendations based on Ktrl target settings.

Reading Campaign Cards

On the card

What it means

What to do

Raw only

  • Unconstrained raw recommendation

  • Ad network is not supported for integration or pacing

  • Nothing for now

  • Let us know if you'd like us to prioritise a particular ad network integration next!

Live → Raw

  • The ad network is supported

  • But no pacing until:

    • Ad network is connected

    • Live bid data exists for the specific campaign or campaign-geo

  • Connect the network to unlock pacing

  • Check back after the next recommendation run to see if new ad network data is ingested successfully

  • Check 'Sort by Geo' - sometimes there are multiple geo-level target overrides, so we will not aggregate the live bid target at campaign level

Live → Paced

  • Your paced target recommendation, which differs from your live bid

  • Move your live bid target toward the paced, recommended value

Live = Paced

  • Your live bid already matches the paced target recommendation

  • No actions needed; your bid is already optimised

Live = Paced

  • A change is due, but the live bid target changed recently

  • The frequency limit is holding it until the next allowed update (e.g. every 7 days)

  • Wait; the new paced recommendations will show once the frequency limit passes

  • In the meantime, the raw recommendations are always available in the Recommendation History tab

Did this answer your question?