Skip to content

analyst.andrewlengly.com/studies/chegg-after-chatgpt :: chegg after chatgpt

# What should Chegg sell now that students have ChatGPT? Its revenue fell 51% in a year while four of the five peers grew. A data study by Andrew Leng Ly.

designed page ......... analyst.andrewlengly.com/studies/chegg-after-chatgpt

CHEGG AFTER CHATGPT

analyst work .......... /machine/analyst
data study ............ October 2026

question .............. What should Chegg sell now that students have ChatGPT?
headline finding ...... Chegg made $51.8 million in Q2 2026, 51% less than a year earlier.
                        From 2022 to 2025, four of the five peers I compared it with grew
                        revenue, led by Duolingo at 181%.
my recommendation ..... Run Chegg Study for cash and spend that cash on a guided tutor
                        licensed to colleges. A sale of the company or its parts stays in
                        reserve if no campus signs.
tools ................. CSV tables with a source URL on every row · JavaScript on Node for
                        the derived rates · D3.js for the first dashboard · React and SVG
                        for the charts on this page

THE QUESTION

the client ............ Chegg’s leadership. A strategic review that opened in February 2025
                        ended that October with no sale, and in August 2026 management
                        recast the company as an employability platform.
the decision they face
                        What should Chegg sell, and to whom, now that a free chatbot gives
                        students the homework answer Chegg used to charge for?
my starting hunch ..... I brought one idea in from my earlier study of grades and AI, that
                        grading pushes students toward short-term answers. I wanted to see
                        whether the edtech numbers fit it.
how the hunch held up
                        It fit in one specific way. Chegg sold the answer to graded
                        homework, and that business collapsed once a free tool gave the same
                        answer. AI use on assessed work among UK undergraduates rose from
                        53% to 94% in two survey rounds. The fit has limits. Nerdy sells
                        live tutoring and its revenue fell 4% in Q2 2026, and investors
                        value Coursera below its end-2022 level. The results are consistent
                        with the earlier study. They do not prove it.

THE DATA, 3 TABLES

[01] metrics.csv .................................................................. 213 rows
     /studies/chegg-after-chatgpt/data/metrics.csv
     One figure per row on Chegg, five peers, student AI surveys, web traffic and market
     values.
     where it came from
                        SEC filings, company earnings releases, Pew Research Center, HEPI,
                        Tyton Partners, Semrush, Similarweb, AltIndex, StockAnalysis and
                        Macrotrends.
     why i chose it ... Each figure has an ID and a series tag, so the code and the trait
                        table can point at the exact row.
     what it cannot tell us
                        Stand as independent evidence on the companies. 127 of the 213 rows
                        are a company reporting on itself. The traffic rows are third-party
                        estimates, and Chegg stopped reporting subscribers after Q2 2025.

[02] events.csv .................................................................... 16 rows
     /studies/chegg-after-chatgpt/data/events.csv
     Dated events from ChatGPT’s launch in November 2022 to the dismissal of Chegg’s suit
     against Google in September 2026.
     where it came from
                        OpenAI and Google posts, company releases, SEC filings and press
                        coverage.
     why i chose it ... It puts dates on ChatGPT and on Google’s AI Overviews, so both can
                        sit beside the revenue series.
     what it cannot tell us
                        Cause. The dates show what came first. Similarweb had Chegg’s US
                        visits down 19.1% year over year in May 2022, months before ChatGPT
                        launched.

[03] traits.csv ..................................................................... 6 rows
     /studies/chegg-after-chatgpt/data/traits.csv
     Six business-model traits, each scored 0 to 2 for Chegg and five peers.
     where it came from
                        My own coding, with the metric IDs each score rests on.
     why i chose it ... It breaks “a different business model” into six things I can compare
                        across companies.
     what it cannot tell us
                        Stand as measurement. The scores are my judgment, and the rubric
                        behind them is not among the published files. With six companies,
                        the rank correlations describe this group and nothing wider.

THE PROCESS

[01] One row per figure
     Each figure went into a row with its entity, a series tag, the period, a source name, a
     URL, the date I retrieved it and a flag for whether the company reported it. That came
     to 235 rows across three files.

[02] Pivot
     Chegg’s quarterly revenue and subscribers sit on separate rows. I pivoted them to one
     row per quarter and added the change on the same quarter a year earlier.

[03] Index
     Pearson reports in pounds and the other five in dollars, so raw revenue cannot be
     compared across companies. Each company’s revenue becomes an index in its own currency,
     with 2022 as 100.

[04] Assumptions
     The 2022 subscriber counts are Chegg Services subscribers and include Busuu language
     learners. From 2023 Chegg reported Subscription Services subscribers. Coursera’s
     release says the merged company took in “above $1.5 billion”, which enters as 1,500
     million. The four layoff rounds compound with no hiring in between.

[05] Judgment
     I scored six traits from 0 to 2 for each company and wrote the evidence beside each
     score. Semrush gave no traffic split for Udemy or Nerdy, so both score 1 on own traffic
     by judgment.

[06] Check
     I summed the four quarters of each year against the full-year row and the two segments
     against the quarter’s total, and compared Chegg’s own full-year figures with the
     StockAnalysis rows the index uses. None of the 10 comparisons was off by more than $0.1
     million.

  # Two companies before and after the index.
  - Before, four rows of metrics.csv (seven of twelve columns shown)
  - After, one row per company
  Pearson’s index sits below 100 because its headline sales carry disposals and currency
  moves. Its underlying sales grew 7% across the same years.

  # The code for the key steps
  - The pivot from long rows to one row per quarter, with the change on a year earlier.
    derive.mjs ........ /studies/chegg-after-chatgpt/derive.mjs
  - The revenue index and the market value change for each company.
    derive.mjs ........ /studies/chegg-after-chatgpt/derive.mjs
  - The rank correlation between each trait score and revenue change.
    derive.mjs ........ /studies/chegg-after-chatgpt/derive.mjs

  # Dead ends
  google trends ....... I wanted a Google Trends series for brand interest. Google Trends
                        blocked automated access, so brand interest comes from published
                        traffic estimates.
  the subscriber line
                        Chegg stopped reporting subscribers after Q2 2025. The count ends at
                        2.6 million, and the last four quarters have revenue only.
  a traffic history ... Only three months of chegg.com visits had a published estimate, and
                        they come from two providers that count visits differently. The 95%
                        drop from April 2021 to September 2026 mixes the two, so I kept it
                        out of the findings.
  trait scores and institutions
                        The trait table could not carry the case for selling to
                        institutions. Across the six companies an institutional buyer shows
                        no link to revenue growth, with a rank correlation of -0.03. The
                        evidence I used for that option is Udemy’s own split, where business
                        revenue grew 6% in 2025 and consumer revenue fell 9%.

THE 3 FINDINGS THAT CHANGE THE DECISION

[01] 51% · Q2 2026 revenue against Q2 2025, $105.1 million to $51.8 million
     Chegg’s revenue fell 4% to 8% a year in each quarter of 2023, and by Q2 2026 the drop
     was 51%.
     dataset .......... metrics.csv
     query ............ derive.mjs, cheggChanges()

     # Chegg’s quarterly revenue fell 51% in a year, to $51.8 million
     Every quarter from Q1 2022, from Chegg’s earnings releases. ChatGPT launched in Q4
     2022, and Google began rolling AI Overviews out to all US searchers in Q2 2024.
     quarter .......... Revenue ($ millions)
     total net revenue
                        Q1 2022 202.2, Q2 2022 194.7, Q3 2022 164.7, Q4 2022 205.2,
                        Q1 2023 187.6, Q2 2023 182.9, Q3 2023 157.9, Q4 2023 188.0,
                        Q1 2024 174.4, Q2 2024 163.1, Q3 2024 136.6, Q4 2024 143.5,
                        Q1 2025 121.4, Q2 2025 105.1, Q3 2025 77.7, Q4 2025 72.7,
                        Q1 2026 63.3, Q2 2026 51.8
     source ........... metrics.csv, derive.mjs, revenueByQuarter(),
                        derive.mjs, cheggChanges()

[02] 49 · Chegg’s 2025 revenue with 2022 set to 100. The lowest peer stands at 93.
     From 2022 to 2025 Chegg’s revenue fell 51% while four of the five peers grew theirs,
     from Nerdy at 10% to Duolingo at 181%.
     dataset .......... metrics.csv
     query ............ derive.mjs, peerChanges()

     # Chegg’s revenue fell to 49% of its 2022 level by 2025 while four of the five peers
       grew
     Full-year revenue in each company’s reporting currency, with 2022 as 100. Pearson
     reports in pounds and its headline sales include disposals, while its underlying sales
     grew 7%. Duolingo’s 2025 figure is the rounded $1.04 billion its source shows.
     fiscal year ...... Revenue index (2022 = 100)
     duolingo ......... 2022 100, 2023 144, 2024 202, 2025 281
     coursera ......... 2022 100, 2023 121, 2024 133, 2025 145
     udemy ............ 2022 100, 2023 116, 2024 125, 2025 126
     nerdy ............ 2022 100, 2023 119, 2024 117, 2025 110
     pearson .......... 2022 100, 2023 96, 2024 92, 2025 93
     chegg ............ 2022 100, 2023 93, 2024 81, 2025 49
     source ........... metrics.csv, derive.mjs, peerChanges()

[03] 22% · Share of chegg.com desktop visits that arrive direct, September 2026
     Just 22% of chegg.com’s desktop visits arrive direct and 57% come from Google organic
     search, while the three other education sites I pulled each get 55% or more direct.
     dataset .......... metrics.csv
     query ............ derive.mjs, trafficSources()

     # 22% of chegg.com’s desktop visits arrive direct. The other sites get 55% or more
     Share of desktop visits that are typed in or bookmarked, September 2026, from Semrush
     estimates. Chegg’s Google figure is organic search and the others are referrals from
     google.com, so the bars show the direct share, which is measured one way for all four.
     Site ............................................... Direct share of desktop visits (%)
     chegg.com ....................................................................... 22.2%
       From Google organic search: 56.9% · Monthly visits: 6.41 million
     Quizlet ......................................................................... 55.3%
       Referrals from google.com: 30.4% · Monthly visits: 66.66 million
     Duolingo ........................................................................ 65.1%
       Referrals from google.com: 14.9% · Monthly visits: 72.15 million
     Coursera ........................................................................ 65.6%
       Referrals from google.com: 11.8% · Monthly visits: 65.57 million
     source ........... metrics.csv, derive.mjs, trafficSources()

MY RECOMMENDATION

  Run Chegg Study for cash and put that cash into a tutor for colleges.

[A] Run Chegg Study for cash
     Keep serving the subscribers who remain. Stop paying to win new homework users, and use
     the cash flow to clear debt and pay for the college tutor.
     evidence ......... Finding 1. Chegg earned $9.1 million in adjusted EBITDA, a measure
                        of operating profit, in Q2 2026. Its adjusted operating costs were
                        $32.3 million, which Chegg’s release puts at nearly half the
                        prior-year level.
     main risk ........ The cash lasts only as long as the subscribers do. Academic Services
                        revenue fell 61% in a year, and at that pace it brings in about
                        $13.4 million in Q2 2027, a straight-line projection.

[B] Sell a guided tutor and the Q&A library to institutions
     License a hint-first Chegg tutor to colleges as a sanctioned study tool. Keep licensing
     the answer library to AI developers.
     evidence ......... Finding 2. 94% of UK undergraduates use AI on assessed work and 38%
                        say their university provides AI tools. Udemy’s business revenue
                        grew 6% in 2025 while its consumer revenue fell 9%. Pearson grew
                        underlying sales 4% in the first half of 2026, and Chegg’s content
                        licensing brought in about $7 million in Q2 2025.
     main risk ........ Nerdy just quit selling tutoring to schools. It said in August 2026
                        that it will shut its schools business.

  - Running Chegg Study for cash needs no new money. Today’s cash flow already covers it.
  - The tutor goes to colleges because institutions are the buyers whose spending held up
    best among the peers.

LIMITS AND NEXT STEPS

  # What the data cannot support
  - 127 of the 213 figures in metrics.csv come from a company reporting on itself.
  - Nothing here shows why any one student left Chegg. The dates of ChatGPT and AI Overviews
    sit beside a revenue series, and Chegg’s US visits were already falling before either.
  - The trait scores are my judgment across six companies. Use beyond deadlines lines up
    best with revenue growth, at a rank correlation of 0.85, and that is a pattern to test.
  - These data cannot show that students would choose learning over answers if grades
    counted less.
  - Quizlet and Course Hero are private. They appear in the traffic rows and neither is
    scored.
  - chatgpt.com is left out of the traffic chart. Semrush’s page for that site has since
    moved on to October and no longer shows the September figures in the table.
  - The data stops at Q2 2026. Chegg’s Q3 results were due in early November, after the
    study closed, and its guidance midpoint of $43.5 million sits 44% below Q3 2025.

  # What I would test with more access
  - With Chegg’s own data I would follow subscribers by sign-up month, to separate students
    who left for a chatbot from students who never arrived from Google.
  - I would pilot the guided tutor with a few colleges and treat a paying campus as the test
    of option B.
  - I would pull the Google Trends series I could not get.
  - I would add Q3 2026 from the November release and rerun the build.

TL;DR

studied ............... What Chegg should sell now that students have ChatGPT, set against
                        five peers.
found ................. Chegg’s Q2 2026 revenue fell 51% from a year earlier, to $51.8
                        million.
  # by the numbers
  51% ................. Revenue drop, Q2 2025 to Q2 2026
  49 .................. 2025 revenue with 2022 set to 100
  22% ................. Desktop visits that arrive direct
the catch ............. Nothing here shows why any one student left. Chegg’s US visits were
                        falling before ChatGPT arrived.
next .................. With Chegg’s own data, follow subscribers by sign-up month to
                        separate chatbot losses from lost Google traffic.

# end of /machine/chegg-after-chatgpt. andrew.leng.ly@gmail.com · leave machine mode