Taboola.com Ltd.

Step 01 — Business Model, Value Chain, and Unit Economics

Company: Taboola.com Ltd. (NASDAQ: TBLA)
Step: 01 of 20
Date: 2026-05-10
Thesis Impact: Mixed — the two-sided marketplace model has powerful structural properties (proprietary data flywheel, exclusive publisher relationships) but the ~63% revenue pass-through to publishers is a permanent ceiling on gross economics that requires careful framing before valuation

1. Key Findings

• Taboola is a two-sided performance advertising marketplace: publishers supply attention inventory (page space), advertisers supply demand for conversions, and Taboola earns the spread — retaining ~37% of gross ad revenue as ex-TAC gross profit [S1][S2].
• The business has two structurally distinct TAC models: (1) revenue-share RSAs (~70-80% of publisher relationships), where Taboola pays a fixed % of revenue earned on each publisher's pages; and (2) minimum-guarantee contracts (~18-19% of TAC is guarantee cost), where Taboola guarantees a minimum payment per 1,000 page views regardless of actual ad yield [S3]. This guarantee layer is the single largest structural risk to margin.
• Realize (launched Feb 2025) is a genuine evolution of the business model — from a native-only (below-article) recommendation engine to a full performance advertising platform covering display, video, in-app, and OEM formats. The strategic significance: Realize allows Taboola to sell format-diversified inventory to more advertiser types, not just content marketers [S1][S4].
• Scaled advertiser unit economics are the key flywheel: advertisers who cross $100K LTM spend show materially lower churn, higher predictability, and compounding wallet share expansion. In Q1 2026, scaled advertisers grew 3.5% and average revenue per scaled advertiser grew 5% [S5]. This cohort (~2,100+ advertisers) generates ~85% of Taboola's revenue [S3].
• Revenue is transaction-based but relationship-driven: while every impression is auctioned in real time, the long-term structure of exclusive multi-year RSAs with publishers (supply side) and growing scaled advertiser wallet share (demand side) gives the model more durability than a pure spot-market auction business.
• The analytically correct P&L starts at ex-TAC gross profit, not GAAP revenue. Management explicitly states this: "we can grow gross revenue by doing bad business — signing up a bad publisher deal — and that's not helpful. What we care about is ex-TAC" [S6].
• Step is net positive for the thesis — the business model is fundamentally sound: two-sided marketplace with exclusive supply, a proprietary data flywheel built over 18 years, and a nascent self-serve platform (Realize) with genuine expansion optionality.

2. Implications for Thesis and Valuation

Three structural realities that must anchor all downstream analysis:

1. 37% ex-TAC margin is effectively the gross margin ceiling for the core business. TAC will remain ~62-65% of revenue because publisher RSAs are the price of supply exclusivity. The only route to margin improvement above ~38% is either (a) Realize attracting direct-demand advertisers who do not require a new publisher RSA payment (unclear — the publisher still serves the ad), or (b) structural improvements in yield on existing inventory (higher RPM from better targeting → publisher gets same % of a bigger number, Taboola keeps 37% of a bigger number). Management's medium-term Adj. EBITDA margin target is 30% on ex-TAC GP — there is no ambitious margin expansion embedded in guidance.

2. Guarantee cost burden (~18-19% of TAC or ~$180-200M/yr) is the primary downside risk to the model. If ad market conditions deteriorate and Taboola cannot generate sufficient ad revenue to exceed guarantee floors, Taboola subsidizes the difference from its own economics. The Yahoo deal introduced the largest single guarantee obligation. This risk is asymmetric — upside from above-floor yield accrues to Taboola, downside below floor falls on Taboola.

3. Scaled advertiser count and spend intensity are the actual growth KPIs. Gross revenue can be gamed by adding low-yield publisher supply. ex-TAC GP is the correct economic output. But the forward indicator for ex-TAC GP growth is the count and average spend of scaled advertisers — this is how Taboola's demand engine compounds. Both grew in 2025 (6% and 2%) and Q1 2026 (3.5% and 5%), confirming the Realize flywheel is beginning to turn.

Valuation implication: Taboola should be valued on EV/ex-TAC GP (economic revenue multiple) and EV/Adj. EBITDA — not EV/revenue, which includes the TAC pass-through and produces misleadingly low multiples. On ex-TAC GP basis, current EV ~$1.44B / $713.5M FY2025 ex-TAC GP = ~2.0x — the right anchor for peer comparison.

3. Objective

Map exactly how Taboola makes money: the supply chain from publisher to reader to advertiser to cash, the pricing mechanics at each stage, the unit economics that determine what scales and what doesn't, and which metrics to track.

4. Narrative Analysis

The Core Business: A Native Advertising Marketplace

Taboola was built on a single insight: publishers have valuable page real estate at the bottom of articles that goes undermonetized. After a reader finishes an article, they are in a content-receptive mindset. Taboola's recommendation widget occupies that moment — displaying a mix of editorially related content and sponsored ("native") advertisements styled to look like organic recommendations [S2].

The core mechanic is simple: a publisher embeds Taboola's JavaScript tag on their pages. When a user finishes reading content, Taboola's algorithm instantly auctions the resulting impression among competing advertisers. The winning advertiser's content appears in the recommendation widget. The reader clicks. The advertiser pays Taboola on a cost-per-click (CPC) basis. Taboola keeps its margin (~37%) and pays the publisher the RSA-negotiated share (~63%) [S1][S3].

This model has two structural strengths that have proven durable over 18 years: (1) publishers receive guaranteed revenue from page inventory that was previously unsold or filled with house ads; (2) advertisers receive clicks from readers who were already actively consuming content — higher intent than passive display.

The Value Chain

Supply side — Publishers:

Taboola maintains exclusive RSAs with ~14,000 digital publisher properties globally, including Yahoo, ESPN, USA TODAY, NBC News, The Independent, and Samsung/Xiaomi device preloads [S4][S5]. The exclusivity is central — Taboola's widget occupies the full below-article recommendation zone; publishers cannot simultaneously deploy a competitor's tag in that space.

RSA terms vary by publisher tier:

• Premium/major publishers (Yahoo, Gannett, NBCNews): Multi-year exclusive contracts with minimum guarantee provisions. Taboola guarantees a minimum revenue per 1,000 page views, providing publishers with revenue certainty regardless of market conditions. These guarantees create margin risk for Taboola but lock in premier inventory.
• Standard publishers: Revenue-share RSAs only — publisher receives a fixed % of ad revenue generated on their pages. No floor. Simpler economics, no guarantee risk.
• OEM partners (Samsung, Xiaomi): Device-level integration — Taboola News is pre-installed as a content feed on devices. Revenue model involves revenue share on ad inventory within the app. Samsung and Xiaomi represent ~$100M+ in annual revenue [S5].

The publisher network creates a proprietary data asset: Taboola has first-party access to 600M+ daily users' reading behavior across these 14,000 properties. This behavioral data — what someone reads, for how long, and what they do next — is Taboola's core targeting substrate. No third-party cookie is needed because Taboola's own JS code is embedded on each page [S4].

Demand side — Advertisers:

Advertisers access Taboola's inventory through two channels:

1. Managed/Enterprise: Larger advertisers work with Taboola account managers. Custom campaigns, performance optimization, dedicated support. Historical model — the legacy of Taboola's first 15 years.

2. Realize (self-serve, launched Feb 2025): SMB and mid-market advertisers access inventory through a self-service dashboard. Set budget, define objective (leads, sales, app installs), choose targeting parameters — Taboola's algorithms optimize in real time. Realize+ (launched Q1 2026) goes further: full automation — provide budget and objective, and Realize+ manages everything including creative generation, audience selection, bidding, and placement [S5].

The Realize product is a strategic pivot. Before Realize, Taboola competed primarily for "content marketing" budgets — advertisers running sponsored articles or listicles. Realize repositions Taboola as a performance advertising platform competing for any digital customer acquisition budget. Formats now include native (legacy), display, vertical video, and in-app/OEM placements [S4].

Ideal Customer Profile (ICP) for advertisers: Management explicitly describes the "sweet spot" as mid-to-lower-funnel, measurable-outcomes-oriented advertisers in verticals like travel, healthcare, auto, personal finance, and e-commerce [S5]. These buyers optimize on CPA (cost-per-acquisition) or CPL (cost-per-lead), making performance measurement straightforward and budget decisions data-driven. The personal finance vertical alone generated $120M in revenue in FY2025 [S5].

Connexity — The E-Commerce Layer:

Acquired in September 2021 for ~$800M, Connexity operates a performance/affiliate marketing business that connects retail brands with editorial content publishers. The model differs from native ads: retailers pay a commission on actual product sales driven by Connexity-placed product links and listings within editorial content. Connexity is integrated into the Taboola platform but has distinct economics — affiliate commissions vs. CPC/CPM. This segment adds e-commerce advertising capabilities, product listing feeds, and retail data to Taboola's platform [S3].

The Realize Flywheel and Agentic AI Angle

The Q1 2026 transcript reveals an emerging narrative that is worth tracking: Taboola has integrated its Realize platform into AI agent frameworks (MCP/tool integrations with Claude, Google, Meta) [S5]. The idea: as agentic AI systems take over media buying functions (an advertiser's AI agent autonomously allocates budget across channels), Taboola needs to be "natively callable" by those agents. Management explicitly stated that a Claude user with a $200/month subscription can now buy search, social, open web, and TV advertising through AI agents — with Taboola as one of the callable platforms [S5]. This is very early but directionally significant for how programmatic advertising could evolve.

Revenue Composition and Seasonality

Revenue is highly seasonal: Q4 is the largest quarter (holiday advertiser budgets) and Q1 is the smallest. FY2025 split: Q1 22%, Q2 24%, Q3 26%, Q4 27%. Adj. EBITDA seasonality is even more pronounced because OpEx is relatively fixed — Q4 2025 EBITDA was $86.1M (40% of full-year) vs. Q1 2025 at $35.9M (17%). This means FCF generation is heavily Q4-weighted [S6].

What Metrics Matter and What Don't

Metrics that matter:

• ex-TAC Gross Profit ($M and YoY %): The actual retained economic revenue — the correct top line
• Scaled Advertiser Count (YoY %): Demand health; proxy for platform quality
• Average Revenue per Scaled Advertiser (YoY %): Wallet share expansion — the compounding mechanism
• Adj. EBITDA and Adj. EBITDA Margin on ex-TAC GP: Operating profitability
• Adjusted Free Cash Flow: Real cash generation (after CapEx, working capital)
• Guarantee cost as % of TAC: Structural risk indicator

Metrics to watch but not overweight:

• GAAP Revenue: Includes full TAC pass-through; can grow without ex-TAC GP growing (management explicitly said this)
• Publisher partner count: Can go down as Taboola optimizes its network (down from 15,000 to 11,000 to 14,000 over FY2022-FY2025) without hurting economics
• Daily Active Users (600M): Reach metric, not a monetization driver on its own

Metrics that are irrelevant or misleading:

• GAAP Gross Margin (30%): Includes other cost of revenue (amortization of intangibles, hosting) beyond TAC; creates false comparison to software peers
• Revenue per user or ARPU: Not how this business monetizes; no direct user relationship
• Subscriber count, MRR, ARR: Pure SaaS metrics; inapplicable to a transaction-based marketplace

5. Evidence and Sources

The business model description is grounded primarily in management's own explanations across the Q1 2026 and Q4 2025 transcripts (most current articulation), the FY2024 10-K (formal business description), and the Investor Day March 2025 materials. All key quotes are from management on earnings calls, not analyst characterizations.

Notable direct quotes used:

• "We can grow gross revenue by doing bad business — signing up a bad publisher deal — and that's not helpful. What we care about is ex-TAC." — Stephen Walker, Q4 2025 [S6]
• "Taboola is one of the largest performance advertising companies outside of search and social, focused on the open web." — Adam Singolda, Q4 2025 [S4]
• "Similar to how Google and Meta understand intent within their own platforms, Taboola understands intent across the open web." — Adam Singolda, Q4 2025 [S4]
• "[With Realize+] Advertisers who want full automation can simply provide a budget and objective and Realize+ will take care of the rest, including audience targeting, creative generation, placements and continuous optimization." — Adam Singolda, Q1 2026 [S5]
• "Our data is our fuel, and it is unique to Taboola. Hundreds of millions of times every year, people across our network make decisions to buy, subscribe or take action." — Adam Singolda, Q4 2025 [S4]

6. Assumption Register Updates

7. Tables and Calculations

Table 1 — Revenue Flow and Unit Economics (FY2025 illustrative)

Table 2 — Advertiser Cohort Unit Economics

*FY2025 and Q1 2026 count/avg estimates extrapolated from announced % growth rates (FY2025: +6% count, +2% avg; Q1 2026: +3.5% count, +5% avg). Actual absolute counts not disclosed.

Table 3 — Revenue Composition by Segment/Channel (estimated)

Note: These are estimates based on available disclosures. Taboola does not break out revenue by these channels in GAAP filings.

Table 4 — Key Revenue and Cost Seasonality (FY2025)

Key pattern: Q4 ex-TAC GP margin (40.8%) is ~500bps above Q1 margin (35.5%) due to holiday advertising premium. Q4 generates 40% of full-year EBITDA despite only 27% of revenue. This operating leverage characteristic is important for FCF modeling.

Table 5 — Business Model Classification

8. Open Questions and Data Gaps

1. Realize incremental economics vs. legacy native: Does a Realize-originating campaign generate a different TAC rate than a legacy native campaign? If a Realize advertiser runs display inventory on Yahoo, does Taboola pay Yahoo the same RSA percentage as for native? If yes, Realize does not improve margins structurally. If no (e.g., OEM/in-app formats have different RSA terms), then Realize could be margin-accretive. Management has not explicitly disclosed this.

2. Connexity profitability and ROI: What does Connexity contribute to ex-TAC GP and EBITDA? Is the affiliate/commerce segment profitable on its own? Has it grown since the $800M acquisition? This is critical for assessing the Connexity investment quality in Step 07.

3. Yahoo deal guarantee structure: What is the exact guarantee level embedded in the Yahoo RSA? Is Taboola currently above or below the guarantee floor on Yahoo inventory? If above, the deal is generating incremental margin; if below, Taboola is paying a subsidy. Management references "yield compression" in FY2023 — was that below the guarantee floor?

4. Agentic AI / MCP revenue potential: The Claude MCP integration for programmatic buying via AI agents is promising but management has not quantified any revenue contribution. How large is this opportunity in 2026-2027? Is this a new demand source or just an interface change for existing budgets?

5. Connexity goodwill impairment risk: With $555.9M of goodwill from Connexity and no impairment in 4 years, what are the triggering conditions? If Connexity revenue growth stalls or e-commerce affiliate margins compress, this is a non-cash risk that could distort GAAP results.

Next-Step Dependencies: Step 02 should use this business model understanding to correctly classify the competitive set. Taboola does not compete with pure software ad-tech (The Trade Desk, Magnite) — it competes for the same performance advertising budget allocation as Meta and Google, making market sizing and peer selection distinctly different from typical ad-tech.

Source Index

Step 04 — Financial Statement Quality

Company: Taboola.com Ltd. (NASDAQ: TBLA)
Step: 04 of 20
Date: 2026-05-10
Thesis Impact: Modestly positive — clean adversarial history, declining SBC, transparent non-GAAP framework, and improving cash conversion. One genuine concern: $555.9M Connexity goodwill with no segment-level P&L to independently verify impairment risk.

1. Key Findings

• Non-GAAP framework is defensible: Taboola's primary EBITDA add-backs are SBC (genuinely declining), Connexity intangible amortization (standard M&A accounting), and interest/taxes — all conventional adjustments with sound rationale. The guarantee cost is already in GAAP TAC (no sleight-of-hand). No "adjusted" revenue or gross-up fabrications [S1].
• SBC is genuinely declining: FY2021 spike ($128M, 9.3% of revenue) was a SPAC + Connexity deal artifact. Since FY2022, SBC has declined from $74.9M (5.3%) to $67.1M (3.8%) to $63.9M (3.3%) in FY2025. The absolute dollar trend is downward, not just the percentage. No red flag [S2].
• Q1 2026 FCF is significantly inflated by a one-time event: FCF of $90.3M includes a $77M legal settlement receipt in operating cash flow. Recurring FCF for Q1 2026 is approximately $13M. This was disclosed in the press release but not separated in the headline FCF figure — a minor transparency concern [S3].
• FY2025 FCF conversion is strong: FY2025 FCF of $163.4M on Adj. EBITDA of $215.5M = 75.8% conversion — above the 60-70% management target. Validates the cash earnings quality of the EBITDA base [S2, S4].
• Adversarial sweep: CLEAN. DOJ no-poach case resolved Aug 2023 (behavioral remedies, no monetary penalty). No SEC investigations. No restatements. No short seller reports specifically targeting TBLA. Content quality ("chumbox") is an ongoing reputational issue but has not resulted in regulatory action against Taboola [S5].
• Audit quality: High. EY-affiliated auditor (Kost Forer Gabbay & Kasierer, PCAOB ID 1281). Unqualified opinions across all years. No going concern language. No audit committee disputes disclosed [S1].
• Goodwill is the principal balance sheet risk: $555.9M Connexity goodwill flat for 4+ years with no impairment. Connexity has no separate segment P&L — investors cannot independently assess whether the underlying business justifies the carrying value [S1, S2].

2. Implications for Thesis and Valuation

• The EBITDA quality is real. FCF conversion of 75%+ in FY2025 confirms that Adj. EBITDA is not papering over a cash-consuming business. The primary EBITDA add-back (Connexity amortization ~$50-60M/yr) is a legitimate acquired-intangibles adjustment, not an operating cost disguised as non-recurring.
• SBC dilution has stopped being a concern. At 3.3% of revenue and $63.9M absolute, SBC is below the threshold where it materially distorts per-share economics. Importantly, buybacks ($262M in FY2025) are running far ahead of SBC issuance — net share count is declining rapidly.
• The $77M legal settlement is a one-time tailwind that should NOT recur. It inflated Q1 2026 FCF, GAAP net income, and the balance sheet cash position. Models built on Q1 2026 FCF as a recurring base will be materially overoptimistic. This is an underappreciated source of Q1 2026 earnings quality illusion.
• Connexity goodwill is the most important unmonitored balance sheet risk. If Connexity revenue is declining (which is unknowable from public disclosures), the $555.9M goodwill could require impairment. At current EBITDA levels (~$215M), a $100-200M goodwill write-down would be painful to reported equity but non-cash, non-tax, and arguably investor-irrelevant for FCF-focused holders. However, it would be a negative signal about M&A discipline.

3. Objective

Evaluate the quality of Taboola's reported financials across five dimensions: (1) GAAP vs. non-GAAP reconciliation integrity, (2) cash conversion quality and FCF sustainability, (3) revenue recognition adequacy, (4) adversarial red flag screening, and (5) balance sheet integrity. Identify any adjustments or risks that would change the economic picture from what management reports.

4. Narrative Analysis

4.1 GAAP vs. Non-GAAP Reconciliation Analysis

Taboola reports three non-GAAP metrics as primary KPIs: (1) ex-TAC Gross Profit, (2) Adjusted EBITDA, and (3) Non-GAAP Net Income. Each adjustment is examined:

Adjustment A: Stock-Based Compensation (SBC)

Assessment: LEGITIMATE ADD-BACK. The FY2021 spike is explained by (1) SPAC merger-related equity grants; (2) Connexity acquisition holdback shares (dealt with separately as "holdback compensation" in FY2022-2024 reconciliations). The FY2022-2025 SBC trend is genuinely declining in both absolute and percentage terms. The Connexity holdback ($7.1M in FY2024, declining as the 3-year vest schedule completes) is correctly separated [S1, S2].

The argument against treating SBC as entirely non-economic: SBC represents real economic dilution. At $63.9M/yr, it is not trivial. However, the buyback ($262M in FY2025, $383.5M cumulative total) exceeds SBC by a wide margin — so net dilution is powerfully negative (shares are being retired, not issued). The SBC add-back is appropriate but should be tracked against buyback spend for net dilution accounting [S2, S4].

Adjustment B: Connexity Intangible Amortization (~$50-60M/yr)

The Connexity acquisition brought publisher relationships, technology, customer relationships, and trade names onto the balance sheet as finite-lived intangibles. GAAP requires amortization of these acquired intangibles over their estimated useful lives (typically 5-15 years). Taboola adds this back in computing Adj. EBITDA, as is standard practice.

The FY2024 D&A reconciliation shows total amortization of $103.7M in the EBITDA reconciliation, vs. only $37.3M in the cash flow statement PP&E depreciation line. The difference (~$66M) represents acquired intangible amortization — consistent with Connexity's $800M acquisition price generating ~$50-60M/yr of intangible amortization [S1].

Assessment: DEFENSIBLE but requires monitoring. The standard argument for excluding acquired intangible amortization is that it doesn't represent an ongoing economic cost — you wouldn't pay to "recreate" Connexity's publisher relationships each year. The counter-argument: M&A is a recurring part of Taboola's growth strategy; treating every acquisition's amortization as non-recurring creates an expectation that the P&L will never reflect M&A costs. For a company that paid $800M for Connexity and has $555.9M on the balance sheet, this is material.

Adjustment C: Guarantee Cost (NOT added back)

The Yahoo minimum guarantee creates above-performance TAC payments when Yahoo's supply underperforms the guarantee floor. This cost is IN GAAP TAC — it flows through the ex-TAC GP calculation and is NOT added back to EBITDA. This is the most important non-issue: Taboola is NOT hiding its guarantee cost in non-GAAP adjustments. The guarantee is fully reflected in the ex-TAC GP number investors use for margin analysis [S1].

Assessment: NO CONCERN. This is the cleaner non-GAAP framework than, for example, a company that excludes "customer acquisition costs" as non-recurring.

Adjustment D: Restructuring Charges

FY2022 and FY2023 included modest restructuring charges (~$3-15M/yr) associated with the Connexity integration and general workforce optimization. These have been zero in FY2024 and FY2025. The elimination of restructuring charges is a genuine positive — not a signal of recurring "non-recurring" costs [S1, S2].

4.2 Cash Conversion Analysis

FCF Calculation Method: Taboola reports "Adjusted Free Cash Flow" = Operating Cash Flow - CapEx - adjustments for publisher prepayments and cash interest (per Adj. FCF disclosure). GAAP FCF = Operating Cash Flow - CapEx.

The FY2024 and FY2025 conversion of 74-76% is above management's stated 60-70% target [S4]. This reflects:

• Working capital is not consuming cash (receivables growing proportionally with revenue)
• CapEx is modest ($35-45M/yr, ~2-2.5% of revenue) — this is a software-heavy business, not capital-intensive
• Debt repayments were front-loaded in FY2023 ($79.3M), reducing interest burden going forward

Q1 2026 FCF anomaly: OCF was $108.7M; CapEx $18.4M; GAAP FCF = $90.3M. This includes a $77M legal settlement received in operating cash flow (settlement income classified as "other income" in P&L, but cash flows as operating). Recurring Q1 2026 FCF = ~$13M. This is in line with Q1 2025 FCF of $36.1M when you exclude the settlement (Q1 has structurally lower EBITDA, hence lower OCF) [S3].

DSO (Days Sales Outstanding): At Q1 2026, Trade receivables were approximately $350-380M (implied from balance sheet current assets minus cash and other items). Quarterly revenue $466.4M → implied DSO ~25-30 days. This is normal for an advertising technology company that invoices monthly post-delivery. No receivables quality concern detected [S6].

Working capital dynamics: FY2025 balance sheet shows Current Liabilities of $521M vs. Current Assets of $558M — net current assets of $37M. The most significant current liability item is "Publisher payables" (TAC owed to publishers, typically on 30-45 day net terms). This creates a natural float: Taboola collects from advertisers before paying publishers. This is a minor structural working capital benefit [S6].

4.3 Revenue Recognition Quality

CPC/CPM model: Revenue is recognized at the moment a click or impression occurs. This is the simplest possible revenue recognition model — no allocation, no variable consideration estimates, no contract modification complexity. ASC 606 applies in a straightforward manner.

Gross vs. net reporting: Taboola reports GROSS revenue (total advertiser spend on Taboola's network) and discloses TAC as a separate line. The ex-TAC GP convention is equivalent to net revenue presentation. GAAP requires gross presentation when Taboola is the principal in the arrangement — which it is, since Taboola controls the advertising inventory before it is sold to advertisers and bears the performance risk [S1].

Yahoo accounting adjustment (Q2 2024): In Q2 2024, Taboola adjusted its revenue recognition methodology for certain Yahoo-related inventory — some Yahoo advertising revenue previously reported gross was reclassified to net (only ex-TAC GP counted). This reduced reported gross revenue without affecting ex-TAC GP or EBITDA. Management disclosed this in the Q2 2024 press release commentary. This was a one-time accounting reclassification, not a change in business economics. The adjustment was not restated retroactively [S3].

Connexity revenue recognition: Connexity operates a comparison-shopping marketplace — revenue is recognized when advertisers' products receive clicks (CPC model, same as native). There is no evidence of more aggressive revenue recognition in the Connexity model [S1].

4.4 Adversarial Sweep Results

Results of adversarial research conducted across regulatory databases, legal databases, and short seller reports:

The DOJ no-poach case (additional context): In August 2023, the DOJ filed and simultaneously resolved a no-poach consent agreement against Taboola for agreements with Outbrain not to solicit each other's employees. This was a relatively common enforcement action in the 2020-2024 period against HR practices in tech (Apple, Google, Adobe faced similar actions). The resolution: behavioral remedies (cease and desist from no-poach agreements), zero fine, zero criminal referral. This is immaterial to Taboola's financial quality assessment [S5].

4.5 Audit Quality Assessment

Assessment: HIGH QUALITY. EY-affiliate, unqualified opinions, no material weaknesses. No concerns [S1].

4.6 Balance Sheet Integrity

Goodwill ($555.9M — Connexity, flat since FY2022):

The $555.9M goodwill balance represents the premium paid above fair value of Connexity's net identifiable assets at acquisition. It is tested annually for impairment under ASC 350.

Key data:

• Goodwill has been flat for 4 consecutive years (FY2022-FY2025) — no impairment charges
• Taboola uses a single-step impairment test (or likely a 2-step): compare reporting unit fair value to carrying amount
• The reporting unit is likely a single segment (Taboola does not disclose multiple operating segments)
• No impairment test assumptions (discount rate, terminal growth) are publicly disclosed
• Connexity has no separate revenue or EBITDA disclosure — investors cannot perform independent goodwill impairment assessment

Risk scenario: If Connexity's e-commerce affiliate revenue is declining (Amazon, Walmart, Google competing aggressively), the Connexity reporting unit's fair value could fall below $555.9M. At Taboola's current stock price ($5.25/share, ~$1.44B market cap), the entire company's equity market cap is only ~2.6x the Connexity goodwill balance. A goodwill impairment would not affect cash flow but would create a large GAAP loss and might trigger investor concern.

Management's implicit defense: Connexity management has cited "double-digit growth" in e-commerce revenue in 2023 press releases. No 2024 or 2025 specific Connexity disclosure. Annual impairment tests passed [S1, S2].

Long-Term Debt ($122.7M at Dec 2024, reduced to ~$119M at Q1 2026): Term loan with voluntary prepayment option. No financial covenant breach disclosed. Interest coverage (EBITDA / interest expense) is comfortably above 10x at current EBITDA levels. No debt maturity cliff concerns [S6].

Publisher Payables (largest current liability ~$300-350M est.): These are amounts owed to publisher RSA partners, typically on 30-45 day payment terms. This is normal working capital for an ad-tech company. No concerns.

Restricted Stock and Options Overhang: RSUs outstanding at FY2024: $134.8M unvested fair value at grant date. These will vest over 4 years (standard quarterly vesting). At current SBC run rate of ~$64M/yr, dilution from new RSU grants is approximately $64M/yr / $5.25 per share = ~12.2M new shares/yr. Buyback of ~261.8M shares in cumulative program at avg ~$3.49/share is retiring far more shares than RSUs create.

5. Evidence and Sources

Primary: FY2024 10-K GAAP/non-GAAP reconciliation tables and MD&A [S1]; XBRL quarterly SBC, CapEx, and OCF data [S2]; Q1 2026 press release FCF anomaly disclosure [S3]; Management FCF conversion target statements [S4]; DOJ/regulatory research [S5]; Balance sheet quarterly snapshots [S6].

6. Assumption Register Updates

7. Tables and Calculations

Table 1 — SBC Trend (FY2019-FY2025)

Table 2 — GAAP vs. Non-GAAP Reconciliation Components (FY2024)

Table 3 — FCF Conversion Quality (FY2022-FY2025)

*Q1 2026 FCF of $90.3M includes $77M legal settlement; recurring FCF ≈ $13M; recurring conversion ≈ 49% annualized

Table 4 — Adversarial Sweep Summary

8. Open Questions and Data Gaps

1. Q1 2026 legal settlement counterparty: The $77M settlement was disclosed as Taboola being the plaintiff. The counterparty and nature of the dispute were not publicly disclosed. While the settlement amount is real (cash received and auditable), the opacity of the underlying dispute is unusual. Unlikely to be material to the investment thesis, but should be tracked for any subsequent disclosures.

2. Connexity impairment test discount rate: What discount rate does Taboola use in its annual goodwill impairment test? WACC sensitivity matters: at 10% WACC, the implied Connexity enterprise value at $555.9M goodwill requires a certain EBITDA level — if Connexity is generating only $50-70M of EBITDA, a 10% WACC test may be passing narrowly. No disclosure provided.

3. FY2024 SBC vs. FY2023 SBC (+4.3% abs): The slight absolute uptick in FY2024 SBC ($67.1M vs. $64.3M in FY2023) bears watching. If SBC reaccelerates as Realize requires new engineering talent, the declining trend could reverse. Monitor FY2026 SBC disclosures.

4. DSO trend: More granular data on receivables days outstanding would confirm working capital dynamics. Current estimate of 25-30 days is approximated from balance sheet and revenue data — actual DSO should be confirmed from future 10-Q receivables disclosures.

Source Index