CONFIDENTIAL — INTERNAL USE ONLY

Taboola Native Demand Analysis

Q1 2026 data · 2025 weekly trends · TBLA public filings context
Yahoo FP&A
Overview
Demand Mix
Supply Arbitrage
Weekly Trends
Fill Rate
TBLA Financials
Advertiser Overlap
Logic Flow
Reference & Notes

Executive Overview

The core finding: Yahoo's preagg ad server records 81.4M net Q1 2026 in Taboola native revenue. Yahoo Omni (Yahoo-sold demand via demand_supply_daily) accounts for 33.9M of that. The remaining 47.5M net (est. 66.3M gross†) is TBLA marketplace demand running on Yahoo's inventory — with zero advertiser identity visible to Yahoo in any data source.
† Gross estimate assumes same 71.625% rev share as Omni demand. Actual rate for TBLA marketplace demand is not confirmed. See Assumptions tab.
Yahoo Preagg Net (Q1 2026)
$81.4M
v_preagg_oo_daily · billing_use_case IN (TABOOLA, TABOOLA_NATIVE_IO)
TBLA Marketplace Demand (Net)
$47.5M
est. $66.3M gross† · 58% of Yahoo total
Yahoo Omni Demand (Net)
$33.9M
271 advertisers · demand_supply_daily
Avg Fill Rate (full period)
65.2%
71.9% Jan 2025 → 62.0% Q1 2026
CPM Gap: Omni vs TBLA-Sourced
1.8×
$1.8 Omni gross vs $0.98 TBLA-sourced (Q1 2026)
Unfilled Ad Opps (weekly avg)
~3.9B
At $0.78 CPM net = est. $158M/yr foregone

Q1 2026 Revenue Bridge — Where Does Yahoo's Native Revenue Come From?

Q1 2026 Demand Mix by Source — Gross Impressions

What Yahoo can see: 271 named Yahoo Omni advertisers (Intuit, Rocket Mortgage, Hulu, Capital One, Apple Card, Hilton, etc.) running $47.3M gross / $33.9M net in Q1 2026. These are demand relationships Yahoo controls and can measure.
What Yahoo cannot see: The other est. $66.3M gross / $47.5M net has no advertiser names, domains, or identifiers in any Yahoo-accessible data source. Every Taboola marketplace impression returns NULL for advertiser domain in both fact_aggregate and preagg. The Salesforce order IDs resolve to a single master record (Order O-672198 — the Omnibus deal). Verified by querying all advertiser-related fields across four TBLA BQ tables.
Display-in-Native — possible arbitrage: TBLA's FY2024 10-K explicitly references "testing of ad formats with Yahoo in the last six months of 2024" as a gross margin headwind. TBLA's external network reporting files show display-format demand at significantly higher CPM than native. Whether and at what scale this occurred on Yahoo O&O specifically is not directly observable in Yahoo's data — Yahoo sees all TBLA demand as billing_use_case='TABOOLA' regardless of creative format. Named buyers in Yahoo Omni native include Fisher Investments, eBay, La-Z-Boy.

Demand Quality Comparison

TBLA-sourced demand generates more impressions and clicks than Yahoo Omni — but at lower CPM and CPC. Omni advertisers are direct-sold Yahoo brands paying premium rates. TBLA marketplace advertisers are performance buyers with higher CTR but lower bid prices. The CPM gap is structural (different demand pools), not evidence of active price suppression — TBLA-sourced CPM has a slight upward trend over the period.
Yahoo Omni — CTR (Q1 2026 avg)
0.291%
~26.6B impr Q1 2026
TBLA-Sourced — CTR (Q1 2026 avg)
0.442%
~62.1B impr Q1 2026
Yahoo Omni — CPC gross (Q1 2026)
$0.62
Weekly avg of weekly weighted CPCs
TBLA-Sourced — CPC gross (Q1 2026)
$0.22
Weekly avg · derived by subtraction†
Yahoo Omni — CPM gross (Q1 2026)
$1.8
From demand_supply_daily gross revenue
TBLA-Sourced — CPM gross (Q1 2026)
$0.98
1.8× lower than Omni · derived†

† TBLA-sourced metrics are derived by subtracting Yahoo Omni (demand_supply_daily) from total (fact_aggregate). See Assumptions tab — confidence: Medium.

CTR: Yahoo Omni vs TBLA-Sourced (weekly, 2025–Q1 2026)

CPC (Gross $): Yahoo Omni vs TBLA-Sourced

CPM (Gross $): Yahoo Omni vs TBLA-Sourced

Weekly Gross Revenue ($M)

CPM decomposition: CPM = CPC × CTR. TBLA-sourced CPM ($0.98) is lower than Omni ($1.8) primarily because CPC is lower ($0.22 vs $0.62) — TBLA marketplace advertisers bid less per click. CTR is actually higher for TBLA-sourced demand (0.442% vs 0.291%). Both CPC and CPM on TBLA-sourced demand show a slight upward trend over the period, which is inconsistent with the throttling hypothesis (which would predict declining CPC/CPM).

Supply-Side Arbitrage

Possible arbitrage mechanisms (both require caveat — see below): (1) Display-in-Native — display-format creatives at higher CPM running through native ad slots; (2) Native-over-Display (NoD) — native demand routed into display/header-bidding positions at lower CPM. CPM figures below are from TBLA's external network reporting files (non-Yahoo publishers) and from the TBLA Advertiser Daily file. Yahoo's preagg classifies all TBLA-served impressions as billing_use_case='TABOOLA' regardless of creative format, so these rates cannot be directly verified on Yahoo O&O from available data.

Q1 2026 — CPM by Format × Supply Position

Q1 2026 Revenue by Format × Supply Position ($M gross)

Arbitrage Detail

Supply PositionAd FormatRevenue (Gross) CPM (Gross)Notes
Yahoo O&O — NativeNative - Static $48.97M$1.79From TBLA Advertiser Daily file (TBLA-provided, Yahoo O&O supply group). Baseline — native demand in native positions.
Yahoo - Channel/HBNative (NoD) $4.00M $1.11 From TBLA Advertiser Daily file. Native demand in display/HB channel slots — 38% lower CPM than native positions. Source is TBLA-provided; not independently verified in Yahoo BQ.
Yahoo - Channel/HBDisplay - Vertical $0.98M $7.41 From TBLA Advertiser Daily file. Display creatives in Yahoo slots — 4.1× premium vs native baseline. TBLA 10-K confirms "testing of ad formats with Yahoo in H2 2024." Top buyers: Fisher Investments, eBay, La-Z-Boy. Source is TBLA-provided reporting; Yahoo sees all TBLA demand as billing_use_case='TABOOLA'.
Possible display-in-native mechanism: A display advertiser bids and wins a native ad slot via TBLA's platform. Yahoo would receive native-channel rates while TBLA charges the advertiser display rates — capturing the spread. TBLA's FY2024 10-K referenced "testing of ad formats with Yahoo in the last six months of 2024" as a gross margin headwind, which confirms the practice occurred but does not specify the scale on Yahoo O&O or the revenue split. The specific CPM figures in this tab are from TBLA's own advertiser reporting files and cannot be independently cross-checked against Yahoo's data because Yahoo's ad server does not record TBLA creative format separately.

Fill Rate Analysis

TBLA fills an average of 65.2% of Yahoo's native ad opportunities. Fill rate declined from 71.9% (Jan 2025) to 62.0% (Q1 2026). Whether this reflects deliberate demand routing, insufficient TBLA demand at scale, or floor-price management cannot be determined from fill rate data alone — all three explanations are consistent with the observed pattern.
Avg Fill Rate (full period)
65.2%
fact_aggregate: impressions ÷ ad_opportunities
Jan 2025 Fill Rate
71.9%
Highest observed in dataset
Q1 2026 Fill Rate
62.0%
Lowest observed — down ~10pp from Jan 2025
Weekly Ad Opportunities (avg)
~11.2B
ad_opportunities from fact_aggregate
Weekly Unfilled (avg)
~3.9B
ad_opps − impressions
Foregone Revenue (est.†)
~$158M/yr
~3.9B/wk × $0.78 CPM net × 52

Fill Rate (%) and Ad Opportunities — Weekly 2025 → Q1 2026

CPM Net ($) vs Revenue Per Opportunity (RPM Net $)

Filled vs Unfilled Impressions (billions, weekly)

Minimum guarantee context: TBLA's 10-K discloses that guarantee costs (TAC paid above revenue-share minimums) represent ≤19% of total TAC. Yahoo's contract "in general" contains minimum guarantees but the specific floor is redacted. If TBLA is paying a floor regardless of fill rate, they have less incentive to push fill rate up beyond the contractual minimum.

TBLA Public Financials Context

Taboola (NASDAQ: TBLA) is a profitable and growing company. Yahoo is a declining share of their business — from 13% to 10.5% of revenue in one year. Non-Yahoo revenue grows faster than Yahoo. TBLA's incentives increasingly diverge from Yahoo's interests.
TBLA FY2025 Revenue
$1,912M
+8.3% YoY · FY2026 guidance: $1,993–$2,054M
FY2025 Adj. EBITDA
$215M
+7% YoY · First consistently profitable year
Yahoo Revenue (FY2025)
$201.6M
10.5% of total · down from 13% (FY2024)
Yahoo TAC Paid (FY2025)
$348.9M
Includes non-Yahoo publisher costs on Omni campaigns
Publisher Partners
~14K
Yahoo + MSFT = ~32% of gross revenue
Yahoo Equity Stake
~25%
Board seat · issued at deal close Jan 2023

TBLA Revenue and Adj. EBITDA ($M) — FY2023–FY2025

Yahoo as % of TBLA Total Revenue

Key Disclosures

TopicTBLA 10-K Language Implication
Yahoo revenue definition "Revenues from the related party are derived from Yahoo's advertiser spend on the Company's network, for which Yahoo is the billing entity." Only Yahoo Omni (Yahoo-sold) demand counts as "Yahoo revenue." TBLA marketplace demand on Yahoo's pages is booked as non-Yahoo revenue.
Unaffiliated TAC "Certain traffic acquisition costs noted herein are unaffiliated with the Yahoo revenues recorded for this period." The $349M TAC includes costs paid to non-Yahoo publishers for Yahoo Omni campaigns running across TBLA's network. Cannot derive Yahoo gross margin from these numbers.
Ad format testing "…partially offset by a margin decrease on certain digital property partners, including the impact of the testing of ad formats with Yahoo in the last six months of [2024]." Direct confirmation of the display-in-native arbitrage identified in BQ data. TBLA booked this as a margin headwind — meaning it cost them more than expected, but they still did it.
Minimum guarantees "Our guarantee costs…as a percentage of our total payments to digital properties, or TAC, was approximately 19% or less." TBLA has guaranteed minimums. Yahoo's specific floor is redacted. Low fill rates may reflect floor-price management rather than demand shortage.
Deal risk language "If the Yahoo partnership is not as long-term financially accretive as we expected…the expected long-term financial benefits of the partnership may not be fully realized." TBLA is hedging its own expectations on the Yahoo deal. They are investing in non-Yahoo growth to reduce concentration risk.
The structural tension: Yahoo holds ~25% of TBLA's equity. TBLA pays Yahoo ~$349M/yr in TAC. TBLA's "Yahoo revenue" ($202M) is declining while TAC grows — this is the 30-year deal working as structured: TBLA gets Yahoo's audience scale, Yahoo gets traffic monetization. But TBLA's growing non-Yahoo business means Yahoo's negotiating leverage weakens over time as exclusivity becomes less valuable to TBLA.

Advertiser Cross-Format Overlap

Three-way comparison of who is buying ads on Yahoo and which channel Taboola controls. (A) Yahoo display inventory won by non-TBLA DSPs via header bidding; (B) Yahoo display inventory won by TBLA as bidder (bidder_name='taboola'); (C) Yahoo Omni native advertisers sold by Yahoo's sales team, managed through TBLA. Sets A & B: Jan 5–11 2026, GAM_PREBID, top 500 domains each. Set C: Q1 2026 full quarter, 28 named advertisers mapped to domains covering 91% of Omni gross revenue.

Key finding: 17 Yahoo Omni native advertisers also appear as winners in TBLA's display header bidding — meaning TBLA controls both their native campaigns AND their Yahoo display spend. For 8 of these brands, competing DSPs also win display impressions, but at higher CPM than TBLA. TBLA is effectively the single point of contact for these advertisers across all Yahoo ad formats.
Set A — Yahoo Display (non-TBLA bidder)
496 domains
GAM_PREBID display, bidder != taboola · Jan 5–11 2026
Set B — TBLA as Display Bidder
500 domains
bidder_name='taboola' in GAM_PREBID · ~265M impr/day
Set C — Yahoo Omni Native Advertisers
28 mapped domains
271 total advertisers · 91% gross rev mapped · Q1 2026
A & B cross-bidder overlap
62 domains
TBLA wins cheaper than open market in 60 of 62
B & C (TBLA: display + native)
17 domains
TBLA holds both channels · $29.2M Omni native gross Q1
All three (A + B + C)
8 domains
TBLA wins display vs open market AND runs Omni native
Open display + Omni native, no TBLA display
1 domain
ebay.com only

CPM by Channel — 8 brands where TBLA wins display vs open market AND manages Yahoo Omni native (Q1 2026)

Display Impression Share by Channel — Jan 6 2026 (single day)

B & C: 17 Brands Where TBLA Controls Both Display and Yahoo Omni Native

Yahoo Omni advertisers (sold by Yahoo's sales team) who also appear as winning display bids under Taboola's DSP. Native CPM from demand_supply_daily Q1 2026 (gross). Display CPM from preagg GAM_PREBID Jan 5–11 2026 (net). “Exclusive” = TBLA display not in open-market top 500 for that week (no competing DSP recorded a win).

Brand Omni Native CPM (gross) Native Rev Q1 (gross) TBLA Display CPM (net) Open Mkt Display CPM (net) TBLA vs Open Mkt Observation
intuit.com $1.97$8.54M$0.54$0.97 −44% Yahoo's largest Omni native advertiser. TBLA also wins display at steep discount.
rocketmortgage.com $2.28$9.33M$0.87Exclusive — TBLA exclusive across both channels. No competing DSP in display top 500.
capitalone.com $1.86$4.97M$0.92$1.02 −10% TBLA controls both channels; competing DSPs win at slightly higher CPM.
apple.com $1.10$3.55M$0.21$1.06 −80% TBLA wins Apple display at 80% discount to open market. Largest discount in sample.
hilton.com $1.09$1.81M$0.50Exclusive — TBLA exclusive across both Yahoo display and native. ~100M display impr/wk.
carecredit.com $3.57$0.64M$1.10$2.21 −50% Highest native CPM in dataset. TBLA wins display at 50% discount.
xfinity.com $1.48$0.47M$0.75$1.31 −43% Comcast residential — TBLA controls both display and native for Xfinity.
comcast.com $1.59$0.71M$0.61$0.92 −34% Comcast Business (separate from Xfinity); both managed through TBLA on Yahoo.
sofi.com $3.14$0.57M$1.06Exclusive — Premium fintech — second highest native CPM. TBLA exclusive.
smartasset.com $2.33$0.47M$0.96Exclusive — Financial content platform; TBLA exclusive in both channels.
veteransunited.com $2.91$0.20M$1.00Exclusive — High-intent mortgage advertiser. TBLA exclusive both channels.
fisherinvestments.com $1.73$0.50M$3.10Exclusive display premium Only brand where TBLA display CPM exceeds native. TBLA exclusive.
frontier.com $1.65$0.57M$1.04$0.80 +30% One of two cases where TBLA display CPM exceeds open market. Telecom vertical.
synchrony.com $1.35$0.36M$0.52Exclusive — Consumer bank arm of Synchrony (CareCredit parent). TBLA exclusive both channels.
schwab.com $1.01$0.34M$0.51$0.48 +6% TBLA display slightly above open market; unusual. Both channels managed by TBLA.
homedepot.com $1.06$0.28M$0.55Exclusive — TBLA exclusive display + Omni native for Home Depot.
lendingtree.com $1.27$0.12M$0.57Exclusive — Financial marketplace. TBLA exclusive across both formats.
Data notes: Sets A & B: preagg GAM_PREBID display rows, Jan 5–11 2026, top 500 domains by impressions per bidder group (net CPM). Set C: demand_supply_daily Q1 2026; 28 domains mapped from 271 advertiser names by brand-name heuristic (91% gross rev coverage, gross CPM). Direct CPM comparison across A/B vs C requires applying 71.625% rev share to Set C figures. TBLA native advertiser domains structurally unavailable in Yahoo BQ (NULL in preagg and fact_aggregate for billing_use_case IN ('TABOOLA','TABOOLA_NATIVE_IO')). Sample caveats: Sets A and B capped at 500 domains — advertisers outside top 500 are not captured in either set.

March 2026 — Yahoo Ads Demand vs. Yahoo O&O Supply: TBLA Markup Analysis

Reconciliation of Yahoo Ads campaign spend (demand-side gross, what advertisers paid TBLA) against Yahoo O&O display revenue (supply-side net, what Yahoo received via prebid) for March 2026. Impression matching caveat: supply reports paid impressions; demand reports viewable impressions — these differ structurally so impression alignment is imprecise. Revenue comparison is directional. 35 advertisers matched; 8 campaigns confirmed within 5% impression variance.

What this shows: For advertisers running Yahoo Ads campaigns through TBLA's platform, TBLA charges them (gross demand) meaningfully more than Yahoo receives (net supply). The spread is TBLA's margin on the intermediary role. On confirmed matched campaigns, TBLA's gross is 40–80% above Yahoo's net for most advertisers — consistent with the ~28–29% rev share structure (Yahoo nets ~71.6% of gross), but several advertisers show significantly higher gaps suggesting additional TBLA platform/DSP fees layered on top.
Advertisers Matched
35
March 2026 — Yahoo Ads on native-to-display line items
Impression-Confirmed Campaigns
8
Impression diff <5% — same campaign confirmed
Avg Markup (confirmed only)
+72%
TBLA gross vs Yahoo net on matched campaigns
In existing B∩C set
14 of 35
TBLA also controls their display bidding + Omni native

Markup % — 8 Impression-Confirmed Campaigns (Yahoo Net vs TBLA Gross)

Net (Yahoo) vs Gross (TBLA) Revenue — Top 10 Advertisers by TBLA Gross (March 2026)

Impression-Confirmed Matches (diff <5%)

Advertiser (Legal) Brand / Domain Site Supply Impr Demand Impr Yahoo Net TBLA Gross Markup In B∩C?
Park Hotels & Resorts hilton.com Yahoo Mail 402.1M398.6M $405K$579K +43% Yes — TBLA controls display + native
Rocket Mortgage, LLC rocketmortgage.com Yahoo Mail 99.6M99.7M $131K$202K +55% Yes — TBLA controls display + native
Duke Energy Corporation duke-energy.com Yahoo Mail 7.5M7.4M $13K$19K +46% No
Healthgrades healthgrades.com Yahoo Mail 6.7M6.5M $5.9K$9.2K +54% No
The Home Depot homedepot.com Yahoo News 513K535K $330$1,386 +319% Yes — TBLA controls display + native
Point Digital Finance point.com Yahoo Mail 641K634K $1.2K$1.4K +20% No
LendingTree, LLC lendingtree.com AOL Mail 134K130K $148$267 +80% Yes — TBLA controls display + native
Hims Inc. hims.com AOL 48.4K49.4K $116$182 +57% No

All 35 Advertisers — Directional Revenue Comparison (March 2026)

Impression mismatch is expected (paid vs viewable) — treat revenue figures as directional, not exact. Advertisers in the existing B∩C set (TBLA controls display bidding + Yahoo Omni native) are flagged. Markup = TBLA gross ÷ Yahoo net − 1. Extreme values (>200% or <−50%) reflect impression measurement mismatch, not real economics.

Advertiser Domain(s) Sites Yahoo Net (Mar) TBLA Gross (Mar) Markup (dir.) Conf. Matches In B∩C?
Park Hotels & Resorts hilton.com8 $463,738$659,652 +42%1 Yes
Rocket Mortgage rocketmortgage.com3 $136,896$210,673 +54%1 Yes
Fisher Asset Mgmt fisherinvestments.com10 $142,581$172,304 +21%0 Yes
eBay Inc. ebay.com11 $60,134$74,226 +23%0 Yes (A∩C)
Comcast / Xfinity xfinity.com, comcast.com9 $61,796$67,776 +10%0 Yes
Apple Inc. apple.com2 $45,734$51,204 +12%0 Yes
American Honda honda.com4 $26,440$47,462 +79%0 No
Frontier Communications frontier.com8 $43,588$45,666 +5%0 Yes
Social Finance (SoFi) sofi.com8 $25,746$19,507 −24%0 Yes
FedEx Corporation fedex.com7 $15,487$19,342 +25%0 No
Skechers skechers.com9 $18,888$17,584 −7%0 No
Duke Energy duke-energy.com5 $13,723$19,838 +45%1 No
United Healthcare uhc.com4 $3,663$8,174 +123%†0 Yes
Synchrony Financial synchrony.com4 $3,794$13,690 +261%†0 Yes
Healthgrades healthgrades.com4 $7,133$9,718 +36%1 No
LendingTree lendingtree.com4 $1,025$1,624 +58%1 Yes
Hims Inc. hims.com7 $2,970$4,183 +41%1 No
The Home Depot homedepot.com3 $4,168$5,492 +32%1 Yes
Google LLC google.com8 $24,929$2,752 −89%†0 No
HSBC Bank USA hsbc.com4 $5,721$3,708 −35%†0 Yes
Schwab schwab.com6 $4,294$2,455 −43%†0 Yes

† Values with extreme markup (positive or negative) reflect impression metric mismatch (paid vs. viewable), not real economics. Only confirmed-match rows (impression diff <5%) should be used for revenue conclusions. Rows greyed out have negative or extreme markup driven by impression measurement issues. Source: DB - TBLA Dashboard.html (March 2026). "In B∩C?" = from Jan 5–11 2026 preagg + Q1 2026 demand_supply_daily analysis.

Analytical Logic Flow

How every key finding in this dashboard is derived — from raw data sources through to conclusions. Confidence ratings use the same scale as the Assumptions tab.

Layer 1 — Raw Data Sources

BQ Yahoo Preagg (Ground Truth)

data-revads-p.revads_vw.v_preagg_oo_daily
Yahoo's internal O&O ad server — the authoritative revenue ledger. Filtered to billing_use_case IN ('TABOOLA','TABOOLA_NATIVE_IO'), valid=TRUE, billable_status=3.
Output: $81.4M net revenue, Q1 2026. Used as the total denominator for all demand-split calculations.

BQ Yahoo Omni Demand (Visible Piece)

data-revads-p.revads_taboola_vw.vw_taboola_demand_supply_daily
Yahoo-sold demand only (yahooomnichannel-*, yahoodanads-* networks). Has advertiser names. Revenue share confirmed at 71.625%: spent_usd × 0.71625 = net_usd.
Output: 271 named advertisers · $47.3M gross · $33.9M net, Q1 2026.

BQ Total TBLA Demand Feed

data-revads-p.revads_taboola_vw.vw_taboola_fact_aggregate_daily
Total TBLA demand (Yahoo Omni + TBLA marketplace combined). All advertiser identity fields (demand_local_advertiser_*) are NULL — verified by exhaustive schema scan. Used for fill rate, impressions, and demand quality metrics.
Output: Q1 2026 total TBLA impressions, ad_opportunities, and revenue (cross-check vs preagg).

BQ Display Advertiser Domains (Indirect)

v_preagg_oo_daily, GAM_PREBID channel — pruned_advertiser_domain is populated only for display demand. TBLA native rows always return NULL for this field (confirmed, not missing data). Sample: Jan 5–11 2026, top 500 domains per bidder. TBLA appears as bidder_name='taboola'.
Output: 496 display domains (non-TBLA bidder) · 500 display domains (TBLA bidder) · 62 overlap.

TBLA-Provided Network & Advertiser Files

TBLA reporting files: Weekly core network file (non-Yahoo publishers, 2025 + Q1 2026) and Advertiser Daily file (Yahoo O&O supply group, Q1 2026). These are TBLA's own records and cannot be independently verified in Yahoo BQ. Used for supply arbitrage tab.
Output: Format-level CPM/revenue breakdown for display-in-native analysis. Non-Yahoo network CPM for ecosystem comparison.

SEC TBLA 10-K Filings (FY2023–FY2025)

Public SEC filings providing: Yahoo revenue definition, TAC structure, minimum guarantee disclosures, ad format testing language, deal risk factors.
Key verbatim: "testing of ad formats with Yahoo in the last six months of [2024]" — direct confirmation of display-in-native practice.

Layer 2 — Key Derivations & Computations

D1 TBLA-Sourced Demand Derivation MEDIUM confidence

Method: Gap = Preagg total − Yahoo Omni (demand_supply_daily)
    $81.4M (preagg net) − $33.9M (Omni net) = $47.5M TBLA-sourced net
Gross estimate: $47.5M ÷ 0.71625 = $66.3M (assumes same rev share rate as Omni)
Risk: If demand_supply_daily under-counts Yahoo Omni (e.g., missing network IDs), TBLA-sourced figure is overstated.
Result: $47.5M net / est. $66.3M gross · 58% of total. Zero advertiser identity in any Yahoo-accessible data source.
↓

D2 Demand Quality Metrics (TBLA-Sourced) MEDIUM confidence

Method: Subtract Yahoo Omni totals (demand_supply_daily) from fact_aggregate totals to isolate TBLA-sourced impressions, clicks, and spend. Then compute: CPM = revenue / impressions × 1000; CTR = clicks / impressions; CPC = revenue / clicks.
Risk: Any weeks where demand_supply_daily is missing Yahoo Omni data inflate TBLA-sourced metrics.
Q1 2026 results: Omni CPM $1.8 · TBLA-sourced CPM $0.98 · Ratio: 1.8×  |  Omni CTR 0.291% · TBLA CTR 0.442%  |  Omni CPC $0.62 · TBLA CPC $0.22
↓

D3 CPM Trend (Linear Regression) HIGH confidence

Method: Linear regression (numpy polyfit degree=1) on 69 weeks of weekly TBLA-sourced CPM gross from throttle CSV. Independent variable: week index (0–68). Dependent variable: tbla_cpm_gross.
No assumptions: Direct computation on observed weekly averages.
Result: Slope = +$0.00248/week (RISING). Direction: rising. Range: ~$0.90 (Q1 2025) → ~$0.99 (Q1 2026). This directly contradicts the throttling hypothesis.
↓

D4 Fill Rate & Foregone Revenue MEDIUM confidence

Method: Fill rate = impressions / ad_opportunities from fact_aggregate.
Foregone revenue estimate: (avg weekly unfilled impressions) × (avg net CPM) × 52 weeks.
~3.9B/wk × $0.78 CPM net × 52 = ~$158M/yr
Risk: Unclear whether ad_opportunities is pre-filtered by TBLA before logging. The CPM applied is average filled CPM — unfilled slots may have lower inherent value (floor-price rejects).
Result: Avg fill rate 65.2% · Jan 2025: 71.9% → Q1 2026: 62.0% (−9.9pp) · Est. $158M/yr foregone (directional only, see risk above).

Layer 3 — Observations

O1 Advertiser Opacity HIGH confidence

Verified by: Exhaustive schema scan of fact_aggregate, demand_campaign_daily, dsp_adv_mdm_mapping_dim, and all related tables. All demand_local_advertiser_* fields return NULL for billing_use_case IN ('TABOOLA','TABOOLA_NATIVE_IO'). SFDC order lookup: all TBLA marketplace impressions resolve to Order O-672198 (single Omnibus deal record).
Conclusion: Yahoo has structurally zero advertiser visibility into $47.5M (58%) of Q1 2026 TBLA native revenue.

O2 TBLA Bids on Yahoo Display (Separate from Native) HIGH confidence

Verified by: Direct query of preagg GAM_PREBID display rows. bidder_name='taboola' returns ~265M impression-wins on a single day (Jan 6, 2026) — ~12% of total GAM_PREBID display volume. Completely separate from the native exclusivity arrangement. For every domain where TBLA and other DSPs both bid and win, TBLA's CPM is lower than the open market in all 62 cases.
Implication: TBLA has a secondary demand footprint on Yahoo O&O display that is entirely outside the native exclusivity framework and currently not tracked in this analysis.

O3 Display-in-Native Format Testing MEDIUM confidence

Source A (TBLA-provided): Advertiser Daily file shows $0.98M gross at $7.41 CPM labeled "Display - Vertical" in Yahoo O&O supply. Source B (SEC filing): TBLA FY2024 10-K cites "testing of ad formats with Yahoo in the last six months of [2024]" as a gross margin headwind — TBLA's words, not an inference. Limitation: Yahoo BQ sees all TBLA impressions as billing_use_case='TABOOLA' regardless of creative format. Scale on Yahoo O&O cannot be independently verified.
Conclusion: Display-in-native occurred and TBLA self-reported it as a cost. Scale in Q1 2026: $0.98M gross per TBLA files. Directionally confirmed by 10-K disclosure.

O4 TBLA Revenue vs Yahoo Preagg Mismatch HIGH confidence

TBLA's 10-K "Yahoo revenue" for FY2025 = $201.6M. Yahoo's preagg records far more TBLA-related native revenue. Reconciled by 10-K language: "Revenues from the related party are derived from Yahoo's advertiser spend…for which Yahoo is the billing entity." TBLA marketplace demand (Yahoo not the billing entity) is counted as non-Yahoo revenue in TBLA's books.
Implication: TBLA's $201.6M understates total demand flowing through Yahoo O&O. The gap represents ~$47.5M net in Q1 2026 alone (~$190M annualized at Q1 rate).

Layer 4 — Conclusions & Confidence

✓ CONFIRMED: Advertiser opacity is structural, not incidental HIGH confidence

Yahoo has no path to see TBLA marketplace advertiser identity through existing data feeds. TBLA has deliberately structured data access to withhold this — a single master Salesforce order covers all marketplace demand. This is an architectural choice by TBLA, not a data gap.

✓ CONFIRMED: TBLA "Yahoo revenue" understates total O&O demand 2–3× HIGH confidence

Based on Q1 2026: $81.4M total vs $33.9M Yahoo Omni (what TBLA counts as "Yahoo revenue"), ratio = 2.4×. Confirmed by 10-K revenue definition. This is a known structural feature of the deal, not a misstatement.

✗ NOT CONFIRMED: Throttling (deliberate demand quality suppression) LOW confidence

The throttling hypothesis — TBLA routing lower-quality advertisers to Yahoo over time — is contradicted by the CPM trend data. TBLA-sourced CPM rose +$0.00248/week over 69 weeks. CPC also rose. A throttling pattern would predict falling CPM. The CPM gap between Omni and TBLA-sourced is structural (different advertiser pools) and has been stable.
Throttling cannot be definitively ruled out (we cannot see TBLA's auction logic), but there is no affirmative data supporting it.

~ DIRECTIONAL: Fill rate decline may represent foregone revenue MEDIUM confidence

Fill rate declined from 71.9% (Jan 2025) to 62.0% (Q1 2026). The foregone revenue estimate (~$158M/yr) is directionally meaningful but depends on assumptions about unfilled CPM value. Root cause (demand shortage vs floor enforcement vs routing) is not determinable from available data.

~ DIRECTIONAL: Display-in-native arbitrage occurred; scale uncertain MEDIUM confidence

TBLA confirmed it in their 10-K. Advertiser Daily files show $0.98M gross at $7.41 CPM (display in display/HB slots) for Q1 2026. Scale relative to total TBLA demand is small (<2% of gross). Cannot be independently verified in Yahoo BQ. The more significant arbitrage may be the $4.0M native-over-display (NoD) flow at $1.11 CPM.

Reference, Glossary & Assumptions

Glossary

Yield Metrics

CPM
Cost Per Mille — revenue per 1,000 impressions served. The primary yield metric. CPM = CPC × CTR × 1,000. A falling CPM can mean lower CPC (cheaper advertisers), lower CTR (less relevant creatives), or both.
CPC
Cost Per Click — revenue per click. Reflects advertiser quality. Higher-bidding, brand-safe advertisers command higher CPCs. A declining CPC trend signals TBLA routing lower-value demand to Yahoo.
CTR
Click-Through Rate — clicks ÷ impressions, expressed as a percentage. Reflects creative-to-audience relevance. Can be inflated by display-format creatives (banners) in native slots, which typically produce higher CTR than true native.
RPM
Revenue Per Mille — revenue per 1,000 ad opportunities (filled + unfilled). RPM = CPM × Fill Rate. A truer yield measure than CPM because it accounts for impressions TBLA chose not to fill.
Fill Rate
Impressions ÷ Ad Opportunities. The fraction of available ad slots that TBLA filled with a paying ad. Below-100% fill means TBLA passed on impressions — either due to insufficient demand, floor price management, or deliberate allocation to other publishers.
Ad Opportunities
Total ad slots made available to TBLA on Yahoo O&O pages in each period (field: ad_opportunities in fact_aggregate). Assumed to represent all eligible TBLA-eligible slots; see Assumptions #6.

Demand Types

Yahoo Omni Demand
Ads sold directly by Yahoo's sales team via the Omni network (yahooomnichannel-*, yahoodanads-* networks), served on Yahoo O&O pages through TBLA's platform. Yahoo is the billing entity. This is what TBLA reports as "Yahoo revenue" in their 10-K ($202M FY2025).
TBLA-Sourced Demand
TBLA marketplace advertisers buying inventory directly on Yahoo O&O pages. Yahoo is not the billing entity — this demand does not appear in TBLA's "Yahoo revenue" line. Identified in this analysis as the gap between total fact_aggregate and Yahoo Omni demand_supply_daily. Also referred to as "hidden demand."
TBLA Network (Non-Yahoo)
TBLA's publisher partners excluding Yahoo O&O. ~60× smaller than Yahoo by impression volume. CPM data from this network is used for external comparison. Sourced from TBLA-provided reporting files, not BQ.
Display-in-Native
TBLA serving display-format creatives (banner-style) within Yahoo's native ad slots. Produces higher CTR and often higher CPC than true native formats. Identified via format_category = 'Display' in TBLA network files and confirmed by TBLA's 10-K ("testing of ad formats with Yahoo in H2 2024").
Throttling (hypothesis)
The pattern where TBLA progressively routes lower-quality demand to Yahoo over time: CPC falls as cheaper advertisers fill Yahoo's slots while better advertisers are directed elsewhere. Results in declining CPM independent of inventory volume changes. Distinct from inventory throttling (restricting impression count).

Revenue & Deal Terms

Gross Revenue
Total advertiser spend before Yahoo's rev share cut. In the BQ demand_supply_daily table: spent_usd.
Net Revenue
Yahoo's share after the rev share. In fact_aggregate: billing_bill_amount / 1e9. Applied rate: 71.625% of gross (see Assumptions #1).
TAC
Traffic Acquisition Cost — what TBLA pays publishers for inventory access. TBLA's "Yahoo TAC" ($349M FY2025) includes costs paid to non-Yahoo publishers for Yahoo Omni campaigns. Cannot derive Yahoo's gross margin by subtracting Yahoo TAC from Yahoo revenue (see Assumptions #12).
billing_bill_amount
BQ column in fact_aggregate and preagg (v_preagg_oo_daily) representing Yahoo's net revenue from TBLA demand. Unit: nanonano-dollars (divide by 1×109 to get USD millions). Confirmed unit empirically by comparing BQ totals to known financial figures.
Exclusivity Agreement
30-year partnership signed ~2020 giving TBLA exclusive rights to serve native demand on Yahoo O&O properties. Generates both the Yahoo Omni revenue and the TBLA-sourced demand flows analyzed here.

Data Sources (BQ Tables)

fact_aggregate
data-revads-p.revads_taboola_vw.vw_taboola_fact_aggregate_daily — Total TBLA demand on Yahoo O&O. Partition column: local_date. Revenue = Yahoo net share. Advertiser identity fields (demand_local_advertiser_*) are all NULL.
demand_supply_daily
data-revads-p.revads_taboola_vw.vw_taboola_demand_supply_daily — Yahoo Omni demand only. Has advertiser_name. Partition column: event_dt. Covers yahooomnichannel-* and yahoodanads-* networks.
site_extended_dim
data-revads-p.revads_taboola_vw.vw_taboola_site_extended_dim — Publisher/site dimension. Joined to fact_aggregate on supply_local_publisher_id = site_id to filter to site_group = 'native' (Yahoo O&O). ~2% publisher ID coverage gap.
preagg (v_preagg_oo_daily)
Yahoo's internal O&O pre-aggregated revenue table. Used for validation cross-check. Returns $81.4M net for Q1 2026 vs fact_aggregate $80.7M (2% gap attributable to site join coverage).

Assumptions & Analytical Uncertainties

The following assumptions underpin the analysis. Confidence levels reflect how well the assumption is supported by data or contract documentation. Any finding dependent on a Low-confidence assumption should be treated as directional, not definitive.

Assumption Confidence Detail & Risk
Revenue share rate: 71.625% HIGH The 71.625% rate is derived from Yahoo Omni contract terms visible in demand_supply_daily SQL (spent_usd * 0.71625 = net_usd). Validated against TBLA's own forecast file (Y! Forecast & Budget 2026.02.21): FY25 actuals show TAC/Gross = 71.5% for TBLA marketplace demand ($217M TAC / $303M gross) and 70.9% for Omni ($118M TAC / $166M gross). Both demand types use the same ~71.6% rev share. Gross estimates for TBLA-sourced demand are reliable approximations.
TBLA-sourced demand derivation: fact_aggregate minus Yahoo Omni MEDIUM The gap methodology assumes (a) demand_supply_daily is a complete, non-overlapping subset of fact_aggregate, and (b) Yahoo Omni is the full set of Yahoo-sold demand. If demand_supply_daily under-counts Yahoo Omni — e.g., missing some network IDs — the TBLA-sourced figure would be overstated. Weeks where omni data is missing are treated as zero Yahoo Omni activity (fillna(0)), which would also overstate TBLA-sourced demand in those weeks.
site_group = 'native' join coverage HIGH The left join from fact_aggregate to site_extended_dim (publisher_id → site_id) has ~2% missing coverage. Publisher IDs with no match in site_extended_dim are excluded. Validated by comparing fact_aggregate ($80.7M net Q1 2026) vs Yahoo preagg ($81.4M net) — the 0.9% gap is consistent with this exclusion and considered immaterial.
billing_use_case filter: 'TABOOLA', 'TABOOLA_NATIVE_IO' HIGH These two values were confirmed via schema exploration as the relevant TBLA demand billing use cases. Cannot rule out that additional billing_use_case values exist which also represent TBLA demand and are excluded. Validated against known revenue totals.
billing_bill_amount unit: divide by 1×109 HIGH The BQ column billing_bill_amount is stored in nano-nano-dollars (confirmed empirically by comparing summed BQ totals against known financial figures). Dividing by 1e9 converts to USD millions. This unit conversion is not documented in BQ schema metadata.
Week alignment: +1 day shift on TBLA network files MEDIUM TBLA network reporting files use Sunday-start weeks (BigQuery DATE_TRUNC(WEEK) default). BQ queries use WEEK(MONDAY). A +1 day shift is applied to network file week_start values before merging. For weeks near period boundaries, this may introduce a 1-week misalignment in comparisons. The 2025/2026 network CPM comparisons should be treated as directional rather than exact week-over-week matches.
ad_opportunities field meaning MEDIUM Assumed to represent all TBLA-eligible ad slots on Yahoo O&O pages. Unclear whether TBLA pre-filters slots before logging ad_opportunities (e.g., excluding slots that don't meet TBLA's own inventory quality thresholds). If pre-filtered, the fill rate denominator understates raw supply and actual fill rate against all available slots is lower than reported.
Network CPM as a routing proxy LOW The non-Yahoo TBLA network CPM (from TBLA-provided files) and Yahoo O&O TBLA-sourced CPM (derived from BQ) are drawn from different advertiser pools. Network files reflect Yahoo Omni advertisers running on non-Yahoo sites; Yahoo O&O TBLA-sourced demand reflects TBLA marketplace advertisers. These are structurally different demand pools — a CPM differential is not direct evidence of deliberate routing. Included as circumstantial ecosystem context only.
2025 vs 2026 Q1 network file format comparability LOW The 2025 TBLA network file has no format_category split (all rows appear as 'Other'). The 2026 Q1 file provides format_category, enabling native-only filtering. For the 2025 comparison, all-format network CPM is used; for 2026 Q1, native-only. These are not fully comparable. The 2025 network CPM includes display-format demand (higher CPM), so the all-in 2025 network CPM may overstate the like-for-like native comparison.
TBLA "Yahoo revenue" definition HIGH Confirmed by TBLA 10-K verbatim: "Revenues from the related party are derived from Yahoo's advertiser spend on the Company's network, for which Yahoo is the billing entity." TBLA marketplace demand on Yahoo pages is booked as non-Yahoo revenue. The $202M FY2025 figure understates the total TBLA demand flowing through Yahoo O&O by a significant margin (est. ~2.3× based on BQ demand split).
Unaffiliated TAC in Yahoo TAC figure HIGH Confirmed by TBLA 10-K footnote: "Certain traffic acquisition costs noted herein are unaffiliated with the Yahoo revenues recorded for this period." The $348.9M FY2025 Yahoo TAC includes costs paid to non-Yahoo publishers for Yahoo Omni campaigns running across TBLA's network. Subtracting Yahoo TAC from Yahoo revenue does not yield Yahoo's gross margin contribution.
Throttling = deliberate demand routing LOW The throttling hypothesis — that TBLA deliberately routes lower-quality advertisers to Yahoo — is inferred from CPC and CPM trends. We cannot observe TBLA's internal auction mechanics or bid routing logic. The same CPC pattern could result from: (a) organic advertiser budget shifts, (b) TBLA's yield algorithm optimizing for volume over price, (c) seasonal effects, or (d) deliberate routing. Data is consistent with throttling but does not prove intent.
Fill rate = TBLA pass rate MEDIUM Below-100% fill rate means TBLA didn't fill every available slot. Root cause is ambiguous: (a) insufficient TBLA demand to fill Yahoo's scale, (b) floor price enforcement (TBLA bids below Yahoo's floor are rejected), (c) deliberate allocation of demand to higher-margin non-Yahoo publishers, or (d) technical latency/timeout rejections. The fill rate data alone cannot distinguish among these causes.
Amendment numbering: TBLA "No. 4" vs Yahoo "No. 6" UNRESOLVED TBLA's FY2025 10-K references "Amendment No. 4" to the Yahoo agreement dated July 25, 2025. Yahoo internally tracks this as "Amendment No. 6." Possible explanations: different counting conventions (including or excluding the original agreement and certain side letters), or some amendments not required to be filed with the SEC. Not material to the financial analysis but relevant when interpreting deal term language.
Advertiser identity: no leak exists in Yahoo BQ HIGH Exhaustive schema exploration of fact_aggregate, demand_campaign_daily, dsp_adv_mdm_mapping_dim, and related tables confirmed that all demand_local_advertiser_* fields are NULL for TBLA-sourced demand. A single master SFDC deal order (O-672198) covers all TBLA marketplace demand in Q1 2026. Yahoo receives aggregated totals only — TBLA has structured data feeds to be opaque on advertiser identity for marketplace demand.

External Validation — TBLA Forecast File (Y! Forecast & Budget 2026.02.21)

TBLA provided Yahoo with a forecast and budget file. Key figures from this file were compared against dashboard calculations to validate methodology. File dated Feb 21, 2026 — Q1 2026 actuals were not yet final.

Dashboard Claim Status Excel Evidence
Rev share ~71.625% applies to TBLA marketplace demand (not just Omni) CONFIRMED FY25 Supply sheet actuals: TBLA Adv TAC/Gross = $217M/$303M = 71.5%. Omni TAC/Gross = $118M/$166M = 70.9%. Both demand types use essentially the same rev share rate. Upgrades Assumption #1 from MEDIUM to HIGH.
Yahoo preagg net Q1 2026 = $81.4M CONFIRMED TBLA's own Q1 2026 budget for total TAC to Yahoo (both demand types) = $91.2M. Our actual = $81.4M — ~10% below budget ($9.8M shortfall). Omni slightly beat (+2%); TBLA marketplace missed budget by 18% ($47.5M actual vs $58.0M budget).
Yahoo Omni net Q1 2026 = $33.9M CONFIRMED TBLA's Omni Channel budget gross = $46.7M. $46.7M x 71.625% = $33.45M ≈ $33.9M. Actual slight beat vs budget. Validation is exact within rounding.
TBLA “Yahoo Revenue” = $201.6M FY2025 (from 10-K) CONFIRMED Excel file shows “Rob Sales Revenue” FY2025 = $201.623M — exact match to TBLA's 10-K figure of $201.6M. “Rob Sales” = Yahoo's gross billings = what TBLA counts as “Yahoo revenue.”
Display-in-native format testing on Yahoo O&O CONFIRMED FY25 Y! Demand Breakdown sheet shows “Yahoo DSP RTB (Display)” = $2.0M FY25 (Q1=$0.54M, declining to Q4=$0.39M). TBLA's own tracking confirms display-format activity on Yahoo O&O, consistent with 10-K disclosure and our Advertiser Daily file.
TBLA marketplace demand missed Q1 2026 budget NEW FINDING Budget net = $58.0M, actual = $47.5M — $10.5M miss (18% below budget). Yahoo Omni beat by $0.7M (+2%). Total shortfall driven entirely by TBLA marketplace underdelivery, not Yahoo Omni. Not reflected elsewhere in this dashboard; flagged here for context.

Data Sources

BigQuery — Yahoo Internal

  • data-revads-p.revads_taboola_vw.vw_taboola_fact_aggregate_daily — Total TBLA demand, Yahoo net revenue. Date range: 2025-01-01 to 2026-03-31.
  • data-revads-p.revads_taboola_vw.vw_taboola_demand_supply_daily — Yahoo Omni demand by advertiser. Date range: 2025-01-01 to 2026-03-31.
  • data-revads-p.revads_taboola_vw.vw_taboola_site_extended_dim — Publisher dimension (site_group filter).
  • media-monetizationdotpix-d project, impersonated via dataflow-sa service account.

TBLA-Provided Reporting Files

  • TBLA_2025_01b_Core_Weekly_TBLANetwork.csv — Full-year 2025 non-Yahoo network performance. No format_category split.
  • TBLA_2026_01b_Core_Weekly_TBLANetwork.csv — Q1 2026 non-Yahoo network with format_category (Native / Display / Other).

SEC Filings (TBLA Public)

  • 10-K: FY2023, FY2024, FY2025 (filed Feb/Mar each year)
  • 10-Q: Q1–Q3 2024
  • 8-K: Earnings press releases FY2024, Q1–Q3 2024
  • Stored locally: 02_TBLA_Financials\

Derived CSVs (this analysis)

  • TBLA_Throttling_Weekly.csv — 69 weeks; demand split, CPM/CPC/CTR by source
  • TBLA_FillRate_Weekly.csv — Fill rate, RPM, CPM by week
  • TBLA_NetworkCPM_vs_YahooOO_2025/2026Q1.csv — Network vs Yahoo CPM comparison
  • TBLA_Q1_2026_YahooOmni_Advertisers.csv — 271 Yahoo Omni advertiser profiles
Yahoo FP&A · Data as of Q1 2026 · Sources: BQ (data-revads-p.revads_taboola_vw), Yahoo preagg (v_preagg_oo_daily), TBLA SEC filings (10-K FY2023–FY2025)