Operating brief: reallocating spend after a network exits
With ironSource's ad network days from shutting down, a framework for reallocating spend that avoids recreating the same concentration risk elsewhere.
The ironSource Ads network shuts down at the end of this month, and most of the buyers affected by it have by now built a replacement media plan. The harder question isn't where the budget lands but whether the new plan repeats the mistake that made this exit painful in the first place: too much dependency concentrated in one place.
Treat this as an operating question rather than a one-off logistics problem, because it will happen again: networks get sold or shut down or deprioritised on timelines a buyer does not control, and the teams that come through the following disruption calmly are the ones that used this one to build a more resilient mix rather than simply patching the immediate gap.
Reallocating after a network exit: diagnose the concentration, not the gap
Before assigning the freed-up budget, look at what ironSource actually represented in your overall mix. If it was 10% of spend spread evenly across several networks, this is a straightforward top-up exercise. If it was 40% of spend because it consistently delivered the best CPI in one core geography or genre, the real lesson is not "find a replacement for ironSource" but "this account was too concentrated in a single network's performance, and that concentration happened to be tolerable right up until it wasn't." Reallocating without addressing that underlying pattern just moves the same risk onto whichever network absorbs the largest share.
Split the reallocation by function, not by convenience
It is tempting to route the entire freed budget to whichever network your team already knows best, because that is the path of least resistance during a busy migration. Resist that by default. Separate the reallocation into at least two buckets. Proven performance goes to networks with a track record on your specific audience; deliberate testing goes to one or two networks you have not scaled before. The testing bucket does not need to be large, even 10 to 15% of the freed budget, but skipping it entirely means the following disruption finds you in exactly the same position, with no tested alternative ready to absorb volume quickly.
Rebuild attribution before you rebuild volume
A network exit is also an attribution event. Confirm with your MMP how it will report historical ironSource-attributed cohorts once postbacks stop, and mark the transition point clearly on your dashboards so that a dip or spike in blended performance around the cutover does not get misread as a campaign problem when it is actually a reporting artefact. Teams that skip this step tend to spend the following month debugging a metrics anomaly that was really just a discontinuity in the data, not a change in campaign quality.
Expect a slower ramp than the spreadsheet suggests
Whatever model you used to redistribute the ironSource budget across replacement networks, build in a slower ramp than the numbers imply is necessary. New budget entering an auction takes time to find its footing, and a network absorbing a sudden volume increase from a competitor's exit is learning your audience at the same time you are learning its delivery patterns. Judging the new mix against the old baseline after one week will produce a misleading verdict in either direction. Give it a full attribution cycle before deciding whether the reallocation actually worked, and resist the temptation to make a second round of adjustments before the first one has had time to show its real shape.
