AI Stopped Being a Story. Now It's in Your Multiple

*The Uphoff Value Index tracks what buyers pay for B2B information, marketing and technology businesses, and why it changes. Each quarter, this report shows what moved and what it means if you're building, running or funding one.* What the UVI is, and who it’s for **What it is.** The Uphoff Value Index is a valuation benchmark for B2B information, marketing and technology businesses. It starts where a buyer starts, with an EBITDA multiple range for each of eight categories, then estimates where a company sits in that range based on three factors: \* **Growth and margin**, read together, so strength in one can offset the other \* **Revenue per employee**, measured against the category’s own peers \* **Protection against AI disruption**, new this quarter The eight categories: **B2B Media** · **Media & Events** · **Business Intelligence** · **Data** · **Marketing Services** · **MarTech and AdTech** · **Demand Generation and Content Syndication** · **B2B Events** **What it’s built on.** Three sources: disclosed M&A transactions; the trading multiples of a fixed set of public companies in each category; and research from the advisers who run these sale processes. This quarter that includes Collingwood’s 2026-27 Market Report, a survey of about 100 active acquirers across information, media and events, and the source of the deal counts and buyer findings below. The ranges are checked every quarter and fully reset at least once a year. A range moves only when the evidence moves it by half a turn of EBITDA or more, so a company’s number changes when the market does, not when a stock has a bad week. The UVI Review Panel, a group of operators, investors, investment bankers and industry analysts who work across these categories, reviews every update before it ships. **How you use it.** The Uphoff Advisory client tool pairs that estimate with a diagnostic of the five **Value Drivers** that build what a business is worth over time: **Revenue Quality** · **Market Position** · **Operational Leverage** · **Leadership Depth** · **Exit Readiness.** This quarter’s ranges, the new AI protection score and revenue per employee for every category go live in the client tool on uphoffadvisory.com later this month. **Who it’s for.** Operators running these businesses, investors funding them, and founders building them. As I argued in How Much Is My Business Worth Is the Wrong Question, the exit multiple is a report card on decisions made years earlier. I’ve been CEO of five businesses, each one a turnaround or a transformation. Every time, the value the next owner paid for was built or spent long before anyone talked about a sale. Most leadership teams learn what their business is worth when a buyer tells them. I built the UVI to show them while there’s still time to change it, and to line up today’s decisions with the value they want to create and the exit they’re planning for. ***Because every decision has a multiple.*** Q3 2026: six of eight categories repriced **The dividing line is AI exposure at the revenue line.** Data, B2B Media and MarTech fell because AI is already visible in their revenue. Business Intelligence held its ceiling because buyers keep paying for the data AI needs. **The deal counts agree.** Information deals rose 88% in the first half of 2026, from 51 to 96, and events deals rose 31%. Media deals fell 46%, from 54 to 29. **Size now shows up in price.** The year’s headline deals carry a scale premium most businesses in these categories won’t see, so two ranges moved down to price the lower middle market. **AI defensibility has become the gate.** Buyers test it before they look at growth or margin. On average, 60% of the sale processes buyers entered never closed, and more of them now end quietly before launch. *The UVI Index, Q3 2026* Where AI reached the revenue line, the multiple followed **Data.** In July the debate about contact and intent data was whether the market had priced AI disintermediation too early. ZoomInfo's second quarter settled it: revenue up 1.2%, a full-year outlook that still points to a decline of about 3%, a $650 million impairment, and the stock down about two-thirds in a year. Buyers have stopped discounting a future threat and started discounting revenue that no longer grows. ZoomInfo’s 35% adjusted operating margin doesn’t earn a premium on a flat top line. I looked at what that means for go-to-market data businesses in Monday’s post, B2B Marketing Is Driving Forward, Looking Backward. **B2B Media.** The category's best public reference point used to be TechTarget, with margins north of 30% before the Informa combination. Informa TechTarget now runs 10% to 13% on revenue down 3%, and it has written off nearly $1 billion of goodwill since the start of 2025, including $45 million in the first quarter of this year. Buyers value digital media businesses with $5 to $10 million of EBITDA at 6.5 to 9.5 times EBITDA. Advertising-led media lost its best argument for a higher multiple. The value hasn't left the category. It's moving to the businesses that turn an audience into data a buyer can use. Media companies with real data revenue already price that way, and they belong in Business Intelligence. **MarTech and AdTech.** The ceiling moved because the company holding it up moved. The Trade Desk set the top of this range for years and now trades around six times EBITDA, with the stock down three-quarters in a year. Zeta Global, growing 44%, trades around 19 times forward EBITDA, but a single company growing at that rate is an outlier. Criteo’s falling revenue marks the floor. In this category growth is the multiple, and very few businesses have it. Intelligence held its ceiling. In events, scale is what gets paid. **Business Intelligence.** Two strategic buyers paid up for data that AI agents need: Publicis for LiveRamp at about 12.3 times forward adjusted EBITDA, and Nielsen for DoubleVerify at about $2.15 billion. DoubleVerify’s growth had slowed to under 3%, so the low end of the range widened. Proprietary data still commands a premium. Slow growth still costs you, even when the data is good. **Events.** Emerald, CloserStill and Hyve priced at an estimated 11 to 14.5 times EBITDA, and the same handful of private equity firms is now buying events businesses from each other. That's what scarcity looks like at the top of a market. Most events businesses aren't at the top. Buyers value organizers with $5 to $10 million of EBITDA at 9 to 12.5 times for tradeshows and 6 to 9.5 times for conferences, and the largest buyers are already refusing to look at anything under $2 or $3 million of EBITDA unless it fits closely with what they own. The demand for events is real. An agent can't replicate a room full of decision-makers. The UVI range moved down, from 10 to 16 times to 8 to 14, to price the business most operators actually run. In services, the spread inside the range is the story **Demand Generation and Content Syndication.** No new disclosed transaction this quarter. Public comps and private-market buyer conversations point the same way: demand generation and sponsorship businesses owned by media companies trade at 6.5 to 10x. Two things hold that pricing down: the hangover from the pandemic boom, when event budgets moved to leads and then moved back, and a flood of low-cost lead providers that has taught buyers to discount lead quality across the category. The top of the 7 to 12x range comes from the exception, and the exception is real. Intent-led businesses with recurring revenue that can show which accounts are in market and prove the return are starting to pull away from media pricing. My read: the market is two to three years from treating that revenue as subscription-like, and pricing it that way. *(Disclosure: I'm a board advisor to Pipeline360, which competes in this category.)* **Marketing Services.** The range held at 8 to 14x. What changed is how buyers place a firm inside it. The holding companies set the ceiling for generalists: Omnicom trades at about 7.8x EBITDA, which limits what the biggest buyers can pay for an agency that looks like what they already own. The premium goes to firms that don’t look like that. The average agency does about $163,000 of revenue per employee. The median public software company does about $395,000. Agentic workflows are closing that gap one firm at a time, and buyers pay for the firms that close it. Take two agencies with identical P&Ls: $11 million of revenue, $2 million of EBITDA, 15% growth. One runs with 68 people, the other with 40. Last quarter the index valued both at 9.8x. Now the 40-person firm is worth 10.3x, about $1 million more on the same earnings. Services is where the gap is widest, and where it will reprice firms fastest. Revenue per employee now moves the multiple Revenue per employee is the cleanest single read on whether technology or headcount is doing the work. Starting this quarter, the UVI measures it in all eight categories, each against its own public peers, because the bar isn't the same everywhere. Listed B2B media companies run about $260,000 to $370,000 per employee, research and advisory firms about $280,000 to $320,000, and MarTech and AdTech platforms about $390,000 to $780,000. A firm at its category's baseline gets no adjustment. A firm approaching the top of its peer range earns up to a full turn of EBITDA, and one well below takes a discount. A B2B media business doing $333,000 per employee picks up more than half a turn, about $1.7 million on $3 million of EBITDA. Growth and margin set the multiple. AI defensibility decides whether you get one. For years, growth plus margin explained most of the difference in what these businesses were worth. It explains less now. Companies with similar growth and similar margins are trading far apart, and the gap comes down to how defensible they are against AI. In Collingwood’s survey of about 100 buyers, 91% called the AI resilience of the business model essential or important. 86% said the same of the end market, and 74% of a direct relationship with the customer. Buyers now sound out businesses six to twelve months before a sale, and 60% of processes never reach a close. So starting with this quarter’s update, the UVI scores AI defensibility. AI Disruption Protection asks four questions of every business in every category: \* **Substitution.** Could AI reproduce what customers pay you for? \* **Customer exposure.** How exposed is your customers’ own business to AI? \* **Relationship ownership.** Do you own the customer relationship, or rent it from search and platforms? \* **Data moat.** Is your data hard to recreate and built into how customers work? The adjustment is deliberately uneven. A well-protected business earns up to a full turn of EBITDA. An exposed one loses up to a turn and a half, because buyers treat AI defensibility as a gate, not a sliding scale. Two data businesses with identical P&Ls show the effect: $5 million of EBITDA, 5% growth, a 20% margin. The one buyers believe is protected is worth 5.6x, or $28 million. The one they think AI can route around is worth 4.0x, or $20 million, and it may never get a process started. What this means if you’re running, funding or building one **If you’re running one:** Find your category and be honest about where you sit in it. The market stopped paying for margin without growth this quarter. If your margin is strong and your growth is flat, that’s no longer a premium story. Fix the growth before you plan a process. Work through the four AI questions before anyone calls, because buyers form their view on AI early and most processes never close. Then track revenue per employee and cash conversion with the same discipline as EBITDA. Buyers already are. **If you’re funding one**: The discounts in data and media aren’t automatically opportunities. Some of these businesses are cheap for a real reason. Underwrite the revenue line and the cash it actually produces. Ask for revenue per employee before you ask for EBITDA, in every category, and benchmark it against the right peers. And watch the gap between public and private prices: listed information companies are down 20% to 30% from their late-2025 peak, private values for the best assets have held, and historically the private market follows with a lag. **If you’re building one:** Build where the multiple holds: proprietary data an agent needs and can’t scrape, experiences an agent can’t replicate, or a services model where each person carries software-level revenue. Own the customer relationship from day one, because renting it from search and platforms is now a discount. And build toward the size buyers will look at. Below $2 to $3 million of EBITDA, your value depends on fitting a specific buyer, so know who that buyer is. The market isn’t guessing about AI anymore. It’s reading the revenue line. Make sure yours says what you think it does. Sources and methodology **How the ranges are set.** Each category’s EBITDA multiple range is built from three inputs read together: disclosed M&A transactions, trading multiples for a fixed set of public companies in each category, and buyer research from the advisers who run these processes. Public multiples come from a single source, Koyfin (S&P Capital IQ data), using trailing twelve-month EV/EBITDA on reported earnings, read October 1, 2026. Ranges are checked every quarter and fully reset each spring. A range moves only when the quarter-over-quarter change in the evidence reaches half a turn of EBITDA or more. **How a company is placed in its range.** Growth and margin are combined as a single Rule of 40 score, so strength in one can offset the other. Revenue per employee adjusts the result against the category’s own public peers. An AI disruption protection score, built from four questions on business model, end market, customer relationship and data, adds a capped premium or discount. Every adjustment is capped, so no single factor can override the evidence. **Revenue per employee peers.** Trailing twelve-month revenue divided by reported headcount, from company filings, October 2026. Figures are shown in the revenue per employee chart in the Marketing Services section. **Deals and company data cited** \* ZoomInfo Q2 2026 results and impairment \* Informa TechTarget Q2 2026 results \* Zeta Global Q2 2026 results \* Criteo Q2 2026 results and guidance \* Publicis acquisition of LiveRamp \* Nielsen acquisition of DoubleVerify \* Hellman & Friedman acquisition of Hyve \* CloserStill sale \* Informa TechTarget Q1 2026 results ($45.0 million goodwill write-down) \* Informa TechTarget Q2 2025 results ($841.3 million goodwill write-down, first half of 2025) **Research and benchmarks** \* Collingwood, Market Report 2026-27 (survey of about 100 active acquirers). Coverage in A Media Operator \* Marketing agency revenue per employee (Promethean Research, via Haus Advisors) \* Public SaaS revenue per employee (Benchmarkit 2026, via The SaaS CFO) *Figures are rounded and current as of October 1, 2026. Examples in this issue were run through the UVI client tool. Nothing here is investment advice.* *The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself.* *“Uphoff on Media” is published by Tony Uphoff, Founder and Managing Partner of* Uphoff Advisory, *a strategic advisory practice for founders, CEOs, and investors in B2B information, marketing, and technology. The businesses that drive business.* This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tonyuphoff.substack.com