general code ltd / internal memo draft v2 · apr 2026
gatekeeper scenario · revised financial model

The three-SSP
open-web lift,
re-examined.

cohort
1.0M US desktop users
scope
open-web standard display
horizon
monthly, 2026 baseline
change from v1
cpm, fill, demand-destruction
01 / 11 → tl;dr
tl;dr03 revisions · 01 sign-flip
executive summary

The v1 model was directionally right, numerically generous.


finding 01 · pricing
$4.50 → $2.75
$4.50 is a PMP-inclusive blend; pure open-auction US desktop standard display clears closer to $2.50–3.00 in 2026.
finding 02 · fill
90% → 75%
Removing AdX destroys demand, not just redirects it. Google Ads SMB & DV360-locked spend has no OpenRTB path to the Big 3.
finding 03 · scenario d
+$1.77M → −$0.36M
Static-only creative under a Big-3-only regime destroys Big 3 revenue vs. the status quo. The purity play is self-defeating.
the new narrative Moderate restriction (Scenario B — partial AdX retention) beats aggressive restriction (Scenarios C / D). The gatekeeper thesis is strongest when it disciplines the auction, not when it tries to replace Google demand.
02 / 11→ assumption delta
assumptions · v1 → v2what changed & why
input deltas

The three knobs that move the answer.


parameter v1 v2 rationale
Pageviews / user / month · Display slots / page 1,200 · 3.0 1,200 · 3.0 Retained. Not the sensitive knobs.
Monetizable share 70% 60% Tighter definition: viewable × bid-matched × rendered. 70% assumed near-perfect bid coverage.
Baseline CPM $4.50 $2.75 2026 open-auction desktop display clears in the $2–3 band; $4.50 implied PMP/PG in the blend.
Fill retention — Scenario B 97% 93% Half of AdX is still competing, so bid density holds up.
Fill retention — Scenario C 90% 75% Demand destruction — Google Ads SMB + DV360-locked spend can't follow to the Big 3.
Fill retention — Scenario D 85% 55% Static-only disqualifies most modern HTML5 campaigns. Demand pool collapses.
CPM damage — Scenario B 0% −5% Halving AdX presence removes marginal bid pressure.
CPM damage — Scenario C −15% −25% Second-price theory: losing a ~40%-share bidder shaves clearing prices 20–35%.
CPM damage — Scenario D (additional) −30% −30% Retained. Static creatives price as remnant; directionally right.
03 / 11→ baseline economics
baseline · before any restrictionmonetized US desktop open-web display
baseline economics

The $5.94M/mo pool at stake.


users
1.0M
US desktop
monetized imp / mo
2.16B
1,200 pv × 3 slots × 60%
blended cpm
$2.75
2026 open-auction
baseline pool / mo
$5.94M
gross display spend
pool distribution — before
participantsharerevenue / mo
Google AdX40%$2.376M
Big 3 SSPs (combined)35%$2.079M
↳ SSP 1 (42% of Big 3)14.7%$0.873M
↳ SSP 2 (33% of Big 3)11.6%$0.686M
↳ SSP 3 (25% of Big 3)8.8%$0.520M
Long-tail SSPs25%$1.485M
TOTAL100%$5.940M
what moves under restriction
AdX · 40% Big 3 · 35% Long tail · 25%
key modelling assumption
The Big 3 absorb redirected volume subject to two frictions: (1) fill retention — not all displaced demand has a path via OpenRTB; and (2) CPM damage — fewer bidders → lower clearing prices. Both are under-modelled in v1.
04 / 11→ scenario A
scenario AAdX unchanged · long tail removed
A
AdX unchanged, long tail removed.
Most benign case — the Big 3 simply inherit the long-tail slice. No fill or CPM damage.
big3 share60% fill100% cpm$2.75
incremental monthly lift · per SSP
SSP 1
+$0.481M
SSP 2
+$0.526M
SSP 3
+$0.478M
TOTAL
+$1.485M
revenue distribution — after
participantrev / moΔ
AdX$2.376M
Big 3 combined$3.564M+$1.485M
Long tail$0−$1.485M
TOTAL POOL$5.940Mflat
interpretation Zero-sum transfer from long tail to Big 3. This is the cleanest scenario — and the most politically achievable, since Google is untouched.
05 / 11→ scenario B
scenario BAdX 50% retained · long tail removed
B
AdX 50% retained.
The optimal case in v2. Big 3 absorb long tail + half of AdX volume; fill and CPM hold up because AdX still competes.
big3 share80% fill93% cpm$2.61 (−5%)
incremental monthly lift · per SSP
SSP 1
+$0.722M
SSP 2
+$0.741M
SSP 3
+$0.656M
TOTAL
+$2.119M
revenue distribution — after
participantrev / moΔ
AdX (half)$1.188M−$1.188M
Big 3 combined$4.198M+$2.119M
Long tail$0−$1.485M
TOTAL POOL$5.386M−$0.554M
the winner under v2 The Big 3 gain more here than in C, because keeping AdX in the auction preserves bid density. The total pool shrinks, but the Big 3's slice grows faster.
06 / 11→ scenario C
scenario CBig 3 only · AdX & long tail removed
C
Big 3 only.
AdX fully removed. Demand destruction bites: Google-Ads & DV360-locked spend has no OpenRTB path forward.
big3 share100% fill75% cpm$2.06 (−25%)
incremental monthly lift · per SSP
SSP 1
+$0.397M
SSP 2
+$0.450M
SSP 3
+$0.416M
TOTAL
+$1.262M
revenue distribution — after
participantrev / moΔ
AdX$0−$2.376M
Big 3 combined$3.341M+$1.262M
Long tail$0−$1.485M
TOTAL POOL$3.341M−$2.599M
counter-intuitive finding Big 3 lift is lower in C than in B. Capturing 100% of nothing < capturing 80% of something. The pool contraction swamps the share gain.
07 / 11→ scenario D
scenario DBig 3 only + static-only creative
D
Big 3 only · static-only.
The purity play. Static creatives are remnant-tier; HTML5-native campaigns are disqualified. Demand collapses.
big3 share100% fill55% cpm$1.44 (−47.5%)
incremental monthly lift · per SSP
SSP 1
−$0.221M
SSP 2
−$0.103M
SSP 3
−$0.040M
TOTAL
−$0.364M
revenue distribution — after
participantrev / moΔ
AdX$0−$2.376M
Big 3 combined$1.715M−$0.364M
Long tail$0−$1.485M
TOTAL POOL$1.715M−$4.225M
the aha The purity version of the gatekeeper scenario is self-defeating. Even the intended beneficiaries (Big 3) lose revenue vs. the status quo. Static-only is a narrative, not a business case.
08 / 11→ v1 vs v2
sensitivityv1 model vs v2 model · total Big-3 lift
original vs revised

Same scenarios, different story.


0 +$1M +$2M +$3M +$4M +$5M −$1M A $2.84M $1.49M B $4.83M $2.12M C $4.71M $1.26M D $1.77M −$0.36M v1 (original) v2 (revised)
scenario A delta
−48%
scenario B delta
−56%
scenario C delta
−73%
scenario D delta
sign flip
09 / 11→ what breaks the thesis
stress testwhat breaks the gatekeeper thesis
stress test · v2

The thesis survives — if three things hold.


must hold
  • Partial-AdX preservation is viable. The gatekeeper can meter AdX access rather than block it. A hard-block regime (C/D) is worse than a throttle regime (B).
  • OpenRTB paths absorb long-tail spend. If long-tail demand is also locked to its SSPs (some DSP seat economics), the zero-sum transfer in A doesn't happen.
  • No retaliatory Chrome / AdSense / PAIR response. Any of those degrading Big-3 auction quality closes the gap v2 still shows.
sensitivity — Scenario B Big-3 lift ($M)
baseline CPM \ fill85%90%93%97%
$2.25+1.44+1.62+1.73+1.88
$2.50+1.60+1.80+1.93+2.09
$2.61+1.67+1.88+2.01+2.18
$2.75+1.76+1.98+2.12+2.30
$3.00+1.92+2.16+2.31+2.51
reading this
Scenario B lift stays in +$1.4M to +$2.5M/mo across the full plausible range. Even at the pessimistic corner ($2.25 CPM, 85% fill), the Big 3 still gain +$1.44M/mo. The B-scenario thesis is robust; it's C and D that are fragile.
thesis-breaker If the gatekeeper cannot preserve some AdX participation — i.e., if the only politically feasible version is "all or nothing" — then v2 implies the scenario collapses from a +$2M/mo transfer into either a +$1.3M/mo win (C) or a −$0.4M/mo loss (D).
10 / 11→ takeaways
takeawayswhat to do with this
conclusions

Four things to walk away with.


01 · the size of the prize
$1.5M – $2.1M / mo
Per 1M US desktop users, under realistic assumptions. Scales linearly with cohort size. Attractive, not transformative.
02 · optimal design
Throttle, don't block
Scenario B (50% AdX retained) outperforms C and D. Preserve bid density — the Big 3 earn more from a smaller, healthier auction than from a monopolised, broken one.
03 · creative restrictions are a tax
Don't touch HTML5
Any "static-only" rule removes the very campaigns that generate premium CPMs. Scenario D goes negative. The purity narrative is incompatible with the revenue narrative.
04 · what to model next
Competitive response
Google's reaction is not modelled. A Chrome-signal degradation or a PAIR/seller-side tightening could close >50% of the v2 lift. This is the next uncertainty to quantify.

— end — “Capturing 100% of nothing is less than capturing 80% of something.” gatekeeper v2
11 / 11↵ cover