Someone once said, “what you don’t measure, you can’t manage.” Ad ops has a worse version of that problem: what you don’t measure, you don’t even know you’re losing.
Because nothing breaks. Dashboards don’t crash. Revenue doesn’t fall off a cliff. It just underperforms quietly, consistently, and at scale.
Most studios aren’t watching revenue drop. They’re watching dashboards that look completely normal.
This article puts a number to that loss, by DAU tier, using real variables. Find where your studio sits. Then decide if the structure you are running is still worth it.
The two types of loss
Manual ad ops creates two kinds of cost. Most studios only see one.
- Direct cost: the rev-share you pay a managed service to run your ops. Visible, negotiated, sitting in a contract at 15 to 30% of revenue
- Invisible cost: the revenue that never gets captured. Stale floors. Slow partner decisions. Fill lost in lower markets. Misconfigured units poisoning your auction data
The direct cost is predictable. The invisible cost compounds daily and shows up nowhere on any report. If you want to understand what ad mediation is actually supposed to do versus what manual management delivers, that gap is where the answer lives.
How the calculation works
Every scenario below uses five variables:
- Daily ad requests: 30 to 100 per DAU per day for a mid-sized mobile game
- Blended eCPM: $6 to $9 across mixed global traffic
- Fill rate loss from ops lag: 4 to 8% below an optimised setup, from stale floors and unmonitored partners
- Floor miscalibration impact: 5 to 10% off optimal kills fill in lower markets entirely
- Rev-share: 20% of gross revenue, the industry midpoint for managed services
What an optimised setup actually achieves
Before the tier calculations, here is what studios running automated execution layers achieve versus manual ops. These are the numbers your current setup is being measured against every day:
- Fill rate: automated setups run 6 to 12% higher fill rates, driven by real-time partner prioritisation rather than weekly reviews
- eCPM uplift: market-specific floor logic delivers 15 to 25% higher eCPM in mid-tier markets like Brazil, Southeast Asia and Eastern Europe, where a single global floor kills fill entirely
- Response time: a floor miscalibration is caught and corrected in minutes on an automated platform. On manual ops, the average lag between signal and action is 48 to 72 hours
- Partner management: automated platforms run 10 to 15 active demand partners simultaneously with no additional ops overhead. Each new partner on a manual setup adds hours of weekly reconciliation
The numbers by DAU tier
50K DAU
At 50K DAU, the game generates 1.5 to 2 million ad requests per day. At a blended eCPM of $7 and 70% fill rate, that is approximately $220,000 per month in gross revenue.
What manual ops costs at this scale:
- Fill rate loss at 6%: $13,200 per month
- Floor miscalibration at 7%: $8,400 per month
- Managed service rev-share at 20%: $44,000 per month
Total: $65,600 per month. $787,000 per year.
Most studios at 50K DAU assume they are not big enough for this to matter. The numbers say otherwise.
100K DAU
At 100K DAU, daily requests sit between 3 and 5 million. At a blended eCPM of $7.50 and 68% fill rate, monthly gross revenue is approximately $459,000.
What manual ops costs at this scale:
- Fill rate loss at 6%: $27,540 per month
- Floor miscalibration at 7% across five to ten markets: $18,900 per month
- Managed service rev-share at 20%: $91,800 per month
Total: $138,240 per month. $1.66 million per year.
The invisible cost alone, before the rev-share, runs over $550,000 per year. That is a UA budget, a new hire or a new title. For context on how much of this comes down to how your mediation stack is structured, the partner mix matters as much as the floor settings.
500K DAU
At 500K DAU, daily requests reach 15 to 25 million. At a blended eCPM of $8 and 65% fill rate, monthly gross revenue is approximately $2.4 million.
What manual ops costs at this scale:
- Fill rate loss at 6%: $144,000 per month
- Floor miscalibration at 7% across a fragmented global audience: $98,000 per month
- Managed service rev-share at 20%: $480,000 per month
Total: $722,000 per month. $8.66 million per year.
At this scale, the rev-share alone is funding an ops function a platform could run continuously, without a Monday morning lag and without a human in the loop.
Where the money actually goes
Annual totals feel distant. Here is what the loss looks like inside a single week:
- A partner drops fill on Tuesday. The weekly report runs Thursday. Two days of millions of requests go unoptimised. That revenue does not come back
- A US-calibrated floor runs flat across India, Brazil and Indonesia all weekend. Fill in those markets hits near zero. Three days, three markets, gone
- A misconfigured ad unit inflates request counts for two weeks. Networks quietly pull back spend. eCPM dips. The report calls it soft demand. The real cause is never found
No alarm fires. Nothing looks broken. The money just does not show up. These are exactly the kind of tasks AI can handle without a human in the loop and the reason more studios are making the shift.
Five signals that manual ops is costing you money right now
You do not need to run the full calculation to know if manual ops is bleeding revenue. These signals show up in your data before the loss becomes visible in your monthly numbers:
- Your floors have not changed in more than five days
- You have a single global floor across all markets. If India, Brazil and the US share the same floor, you are losing fill in at least two of those three every day
- You cannot identify which demand partner dropped fill last week without pulling a manual report
- Your placement IDs have not been audited in 30 days. Misconfigured units suppress win rates and networks pull back spend before you notice the eCPM shift
- Your ops review cadence is weekly. At 5 million daily requests, every problem that surfaces between Monday and Friday runs unchecked. That is not a gap. It is a structural leak
If three or more of these apply, the annual loss figures in this article are already in your numbers. They are just not labelled.
What the same studio looks like without the manual layer
There are 6 reasons publishers are moving from managed ad ops to autonomous platforms, but the operational difference comes down to this:
- Floors update per market in real time, not globally once a week
- Partner prioritisation shifts based on live fill behaviour, not last week’s report
- Misconfigured units are caught before they affect auction data
- The team focuses on format strategy, market expansion and monetisation structure
The revenue gap this article has been calculating is the difference between those two setups.
The cost of switching versus the cost of staying
The obvious objection: switching to an automated platform has a cost too. Here is how the comparison actually looks:
- Automated platforms typically charge 5 to 10% of revenue versus the 15 to 30% a managed service takes
- At 100K DAU on $459,000 per month, the difference between a 20% managed service fee and a 7% platform fee is $59,670 per month before any fill rate or eCPM improvement is counted
- Onboarding and SDK integration typically takes two to four weeks of engineering time. At 100K DAU, that cost is recovered in the first month of the fee difference alone
- Modern autonomous platforms operate with full transparency. Every floor, every partner decision and every market configuration is visible and adjustable. The platform acts on them in real time instead of waiting for a human to catch up
- For a broader view of what AI is actually doing in ad operations, the underlying logic is less opaque than most studios assume
The cost of switching is a one-time adjustment. The cost of staying compounds every day the structure does not change.
Conclusion
No single day of manual ops feels costly. A missed partner review, a stale floor, a quiet weekend. Each one is a small loss. Together, across a year, at any of the three tiers above, they add up to a number that changes how you think about your current setup.
The studios pulling ahead are not doing manual ops better. They have taken it off the table entirely at the execution layer. For a broader view of what a modern monetisation stack looks like, that is the next read.
If you have not calculated what yours is costing you, now you have the framework to do it.
