How the index is built

Universe and inclusion

The index has three headline indices: Listed Learning, Listed Work and Private Learning & Work. A fourth series, AI Frontier, tracks AI-native companies outside learning and work for context and is not part of any index.

Listed companies are included when they

• earn at least half of their revenue from learning (platforms, content, education provision) or work (HR, payroll, talent, recruiting);
• trade on a major exchange with a market capitalisation above $50M at the monthly review;
• are valued on enterprise value, except lenders and loan servicers (Nelnet), which are valued on market capitalisation: their enterprise value would count securitised loan funding as corporate debt;
• report revenue on a net basis. Professional employer organisations such as TriNet and Insperity book their clients' payroll as revenue, so their multiples are not comparable and they are excluded.

Each company belongs to one segment. Segment medians show where value sits within each index.

Metrics

Headline multiple

The median of each index's companies. The median is used because a few outliers, such as companies with large net cash, would distort an average. Each card also shows the middle half of the index (25th to 75th percentile) and a size-weighted multiple: total enterprise value over total next-twelve-months revenue.

Listed companies

• EV / NTM revenue: enterprise value over next-twelve-months consensus revenue. NTM revenue blends current and next fiscal-year consensus, weighted by the time left in the fiscal year.
• NTM growth: NTM revenue over last-twelve-months revenue, minus one.
• FCF margin: free cash flow over the last four reported quarters, divided by last-twelve-months revenue.
• Rule of 40: NTM growth plus FCF margin.
• EV / LTM revenue and EV / EBITDA: trailing multiples on reported last-twelve-months revenue and EBITDA, for comparison with data terminals. EBITDA is as reported, not adjusted for stock compensation or one-off items, so EV / EBITDA runs higher than broker figures for some companies. It shows NM when EBITDA is zero or negative.
• 1-year change: this month's multiple against the snapshot from twelve months earlier. Until twelve months of snapshots exist, a proxy is used: EV a year ago (today's net debt, equity scaled by the share-price move) over today's trailing revenue.

What drives multiples

A straight line fitted through all listed companies relates EV / NTM revenue to growth and to Rule of 40. Companies with growth above 60% or below -40% (Rule of 40 above 90%) are left out of the fit, since mergers or collapsing revenue usually explain them. R² shows how much of the variation in multiples each measure explains.

Private marks

• Talent & AI training: expert marketplaces supplying AI training data (Mercor, micro1, AfterQuery, Invisible) often report gross run-rate before contractor payouts, and their demand depends on AI labs. They form a separate segment, and the headline card also shows the median without them.
• Confidence: high when valuation and revenue are both disclosed by the company or reported on the record; medium when either comes from anonymous sources, is a minimum or a proxy, or predates the round.
• Only primary rounds count. Secondary sales and employee tenders are noted but not used.
• The multiple is post-money valuation over ARR (or run-rate revenue) disclosed at the same round, so price and revenue refer to the same moment. Where a company discloses only statutory revenue, that is used and labelled.
• A company enters the index only when both valuation and revenue are publicly disclosed. Rounds with a valuation but no revenue are listed separately.
• Marks older than 12 months are flagged; marks older than 18 months are excluded from medians.
• Change vs previous round compares the two multiples, and is shown only when ARR at the previous round was disclosed.
• Growth is shown only where disclosed or derivable from two disclosed figures a year apart. Medians are hidden when fewer than half the companies disclose a figure.

Exits

Acquisitions and take-privates of learning and work companies since 2019: deals above about $300M, and from 2025 deals of any size with a disclosed price and revenue (including carve-outs). Distressed sales are left out. The multiple is enterprise value over trailing revenue at announcement, taken from filings and press releases. Where only equity value was disclosed, it is used and marked. Low-confidence figures are shown but left out of the medians. Premiums are measured against the unaffected share price or the stated reference average.

History

The history chart shows trailing multiples (EV / LTM revenue) are rebuilt monthly from September 2023 using share prices, share counts and reported financials. Financials are used only 60 days after their period ends, so no month uses figures the market didn't have yet. The series uses today's constituents, so companies that were acquired or delisted are missing (survivorship bias), and a month is shown only when at least 90% of an index's companies have data, so changes in coverage don't move the median.

Brighteye Public Index

The Brighteye Public Index (BPI) tracks how listed learning and work valuations move over time. It starts at 100 in January 2024. The multiples elsewhere on the page show what companies are worth today; the index shows the direction and size of re-rating since the base month.

• What moves it: each month, every company's EV / trailing revenue multiple is compared with its own multiple the month before. The index moves by the equal-weighted geometric average of those changes. A +10% and a −10% month roughly cancel out.
• What doesn't: revenue growth (the multiple already divides by revenue), company size (every company counts equally) and changes to the list. A company only counts once it has two consecutive months of data, so adding or removing one never makes the index jump.
• Eligibility: a company is left out of a month's calculation if its multiple is below 0.5x or its enterprise value is below 25% of its market cap in either month. Near-zero enterprise values (companies holding mostly cash) turn small price moves into large percentage swings.
• Cap: a single company's monthly change counts at most ±30%.
• Sub-indices: Learning and Work are calculated the same way from their own companies. An index needs at least 5 eligible companies in a month; otherwise it carries forward.
• Universe: all listed companies in the index. Companies under a takeover offer are removed from the announcement.
• Data: month-end share prices, share counts and reported financials, with financials used 60 days after their period ends. Trailing revenue updates once a quarter, so in reporting months a fast grower's multiple can fall even if its share price is flat. The current month uses the latest price and is marked "to date". History uses today's constituents (survivorship bias).

Valuation tool: value ranges

The valuation tool gives illustrative values for a company from its ARR, growth, gross margin, EBITDA margin and burn multiple. Each lens shows its calculation (multiple × revenue or EBITDA × adjustment) first, then its range. If a proposed round is entered, its post-money is marked against each range. It is a conversation starter, not a valuation. It is not designed for seed rounds, where price depends on team and market more than on metrics.

Growing 25% or more a year: valued on revenue

• Listed peers: the growth line fitted across listed companies, applied to next-twelve-months revenue (ARR grown by half the stated growth rate). Growth above the listed range is capped at its top. The range is plus or minus the line's typical error.
• Acquisition: the middle half of exit multiples for the chosen market, applied to ARR.
• Late-stage private round: the middle half of Private Learning & Work marks, applied to ARR. The burn multiple (net burn divided by net new ARR) sets the point within that range: 0.75x or better at the upper quartile, 1.5x at the median, 2.5x or worse at the lower quartile. These bands follow a common investor rule of thumb and are not fitted to index data, because private companies rarely disclose burn.
• Gross margin: each range is scaled by the company's gross margin relative to a reference (the listed peers' median for listed and acquisition ranges; 75%, typical for software, for private marks), capped between 0.4x and 1.4x. Investors pay for gross profit, so a services-heavy business is worth less per dollar of revenue than software.

Growing less than 25% a year: valued on EBITDA

Slower growers are priced on profit, not revenue, so revenue multiples would overstate their value.

• Listed peers: the middle half of EV / trailing EBITDA across profitable listed companies in the chosen market, applied to the company's EBITDA (revenue × EBITDA margin). Listed companies are larger and more liquid, so this is a ceiling.
• Acquisition: 5x to 15x EBITDA (10x midpoint), the range buyers typically pay for profitable private companies growing below 25%, from Brighteye's deal experience.
• The late-stage private lens is not shown: late-stage rounds price fast growth. With zero or negative EBITDA, no value is shown.

Valuation tool: checking a proposed round

When a pre-money and round size are entered, the tool also checks whether the price holds up against published benchmarks. Nothing entered leaves the page.

• Round terms: post-money (pre-money plus round) and dilution (round divided by post-money), against the stage benchmarks: Carta for US software (post-money percentiles, dilution) and PitchBook median pre-money and round size for the selected region (European VC Valuations Report for Europe, PitchBook-NVCA Venture Monitor for the US). Euro figures are converted at the month's exchange rate.
• Implied multiples: pre- and post-money divided by current ARR and by forward 12-month ARR.
• Benchmark ladder: the round's post-money multiples against public software medians (Meritech, on current ARR), growth cohorts (Clouded Judgement, on NTM revenue), and Brighteye's private marks, exits and listed indices. Each is tagged current, forward or trailing. A forward-ARR multiple should only be compared with forward benchmarks; comparing it with current-ARR benchmarks flatters the round.
• How to justify the valuation with investors: a paragraph summarising the above with sources and dates, ready for an investor memo or deck.

Special cases

• Mergers and acquisitions: a company whose data mixes pre- and post-deal figures is excluded until its first combined quarter is reported (Coursera, after merging with Udemy).
• Takeover offers: a company under a takeover offer is excluded from medians from the announcement, because its share price reflects the bid rather than the market's view (Tribal Group, from September 2026). It returns if the bid lapses, or moves to Exits when the deal completes.
• Delistings: companies taken private leave the index the month they stop trading, and the deal is added to Exits.
• Net-cash companies: when enterprise value is below 25% of market cap, the 1-year proxy is suppressed because small moves in EV produce meaningless percentage changes. The company stays in the medians.
• Mergers: when a company has just completed a large merger, data providers can lag its first combined quarter. Until they catch up, its enterprise value and trailing revenue are taken from its latest filing (Coursera, Q3 2026, after merging with Udemy). Its growth is left out of growth medians, Rule of 40 and the fitted line until it has four combined quarters, because the merger, not the business, drives the reported growth.
• Currencies: multiples are calculated in each company's reporting currency. Where a stock trades in a different currency (Thinkific: Canadian dollars), EV is converted first. Dollar figures use the current exchange rate.
• Review: every monthly update is checked by a person before publication. Automated checks flag currency mismatches, very low multiples, extreme growth and moves above 30% in a month.

AI exposure

Listed companies are tagged as exposed to generative AI when their core product is something a general AI model can substitute: homework help, tutoring, content libraries, freelance digital work and job boards. This is Brighteye's judgement, reviewed each quarter. It describes exposure, not outcome: an exposed company can still adapt.

Sources and freshness

Every figure on the index comes from one of the sources below. Each has a refresh window; the monthly review flags any source past it, with instructions for updating it. Automatic sources refresh on the 1st of each month.

The index is published for information only and is not investment advice.

Change log