SEO didn't die in 2025. It split into two jobs — one pays 27% more

Executive summary

The layoffs are real, the causes are contested, and the job market is bifurcating in a way the headlines mostly miss. Chegg cut 388 employees — 45% of its workforce — on October 27, 2025, with the company stating plainly that "the new realities of AI and reduced traffic from Google to content publishers have led to a significant decline in Chegg's traffic and revenue" (Confidence: High — direct company statement, confirmed via Fox Business, corroborated by CNBC and Higher Ed Dive). Business Insider cut 21% of staff in May 2025 after organic search traffic fell 55% between April 2022 and April 2025 — a figure that traces to Wall Street Journal reporting on Similarweb data and is independently corroborated by Adweek and AdExchanger, making it the single best-sourced traffic-collapse statistic in this piece (Confidence: High).

But not every widely repeated data point in this story survives scrutiny. The HubSpot traffic collapse — cited constantly as proof that AI Overviews kill commodity content marketing — is more honestly attributed by the outlets that covered the actual event to Google's December 2024 core and spam updates and to HubSpot's own thin, off-topic "traffic at all costs" content strategy, not to AI search specifically. The oft-quoted "70-80% loss" figure traces only to SEO-blog aggregation, not to any outlet naming AI Overviews as the cause. That claim is dropped from this analysis rather than repeated.

Underneath the layoff headlines sits a quieter, better-documented split: a study of 1,175 US SEO job listings posted between April 2025 and March 2026 found that 54.9% of job descriptions now mention AI-related skills, but only 11% of job titles do — and the roles that do carry an AI-related title pay a median $113,625 versus $89,438 for those that don't, a 27% premium (Confidence: High — primary source directly fetched, every figure confirmed). The people losing jobs and the people commanding premiums increasingly look like they're in different professions, even when their job title still says "SEO."

A timeline / comparison table

Part 1 — The confirmed body count

Four data points in this story hold up under direct scrutiny. Chegg's cut is the cleanest: 388 jobs, 45% of the company, announced October 27, 2025, with an on-the-record company statement naming AI and Google traffic decline as the cause (Confidence: High). Business Insider's is the best-evidenced on the traffic side: a 21% staff reduction in May 2025, following a 55% organic-search-traffic decline over three years that comes from WSJ reporting on Similarweb data and is corroborated independently by Adweek and AdExchanger — genuine cross-outlet confirmation, not one press release repeated by five blogs (Confidence: High). CEO Barbara Peng's internal memo put it plainly: "Since returning to our roots as Business Insider, we've been building toward something new. This kind of transformation takes time — and it requires tough decisions along the way" (Confidence: High — verbatim, confirmed via Adweek).

At the holding-company level, WPP's headcount fell from 108,044 to 98,655 — roughly 8.7%, not the round "9%" some coverage implies — under CEO Cindy Rose's multi-year "Elevate28" restructuring, which targets £500 million in annualized cost savings by 2028. WPP has also flagged several hundred more job cuts, about 1% of its global workforce, by the end of 2026 (Confidence: High — figure and forward guidance corroborated across multiple trade outlets). Omnicom and IPG eliminated roughly 8,200 roles, about 6.4% of combined headcount, ahead of their merger, with roughly 4,000 more redundancies following the close around December 2025 (Confidence: Medium — confirmed via eMarketer's primary reporting, though the underlying quarter-by-quarter numbers are genuinely messy across sources; IPG alone reported ~3,200 cut in 2025 on top of ~4,000 in 2024).

Part 2 — What's murkier than reported

Two specific numbers commonly attached to this story do not hold up as presented. First, Chegg's traffic decline: the company disclosed a 49% year-over-year drop in non-subscriber traffic around January 2025 — but that figure belongs to an earlier disclosure, not the October 2025 layoff announcement itself. Chegg's actual Q3 2025 filings, made around the time of the layoffs, cite figures in the 37–50% range depending on which metric and quarter is referenced. Chegg's SEC 8-K could not be directly fetched this session (it returned a 403 error), so treat any single precise percentage tied to the layoff date itself with caution (Confidence: Medium).

Second, and more significant: HubSpot's traffic collapse is widely cited in SEO commentary as the definitive case study of AI Overviews destroying commodity content — but the reporting that actually covered the event attributes it to Google's December 2024 core and spam algorithm updates and to HubSpot's own thin, off-topic content strategy, not to AI search. The "70–80% traffic loss through mid-2025" figure circulating in SEO blogs traces only to secondary content-marketing sites, not to any outlet connecting the cause to AI Overviews. This piece treats HubSpot as a disputed, misattributed example rather than evidence for the thesis it's usually cited to support.

Similarly, the Omnicom/IPG "$1.5 billion synergy target, $1 billion from labor savings" figures that circulate in trade coverage could not be corroborated against the eMarketer primary source used here, and are excluded rather than repeated on faith.

Table to include for design: a side-by-side comparison of the "widely repeated" figure versus the "verified" figure for Chegg's traffic decline and the HubSpot case, to visually flag where recycled numbers diverge from primary sourcing.

Part 3 — Shrinking the account team, not just the payroll

The clearest evidence of role restructuring — as opposed to blunt headcount cuts — comes from inside Meta's advertiser-facing operation. Digiday reported in February 2024 that dedicated account teams, which previously ran 5 to 10 people per client, had shrunk to 2 to 3, with the gap filled by Advantage+ automation, self-serve tools, automated lift studies and chatbot support (Confidence: High — directly confirmed).

The people willing to put their names to this shift were agency and ad-tech executives, not the advertisers themselves. Elijah Schneider, CEO of the agency Modifly, described being told directly that a service tier no longer existed: "They literally said [via email] we don't offer that as a managed service anymore, but here's essentially a Meta learning platform with links and from that you can download our software" (Confidence: High — verbatim, named source). An unnamed senior marketer at a CPG business offered the sharper line — "Our account was loaded with people who we weren't sure what they did or didn't respond to requests. Now, we have a handful of execs on the account, and a chat service to get the rest of the support" — but that quote is anonymous, and it would be inaccurate to describe it as an on-the-record advertiser confirmation; only the agency-side executives spoke for attribution (Confidence: High for the quote's existence; the speaker's identity and affiliation is confidential by the outlet's own choice, not confirmed as "advertiser on record").

Part 4 — The job market is bifurcating, and the data says so plainly

This is the strongest, cleanest signal in the entire dataset. Josh Peacock's Search for Hire analyzed 1,175 unique US SEO job listings posted between April 2025 and March 2026: 54.9% of job descriptions mention AI-related terms, but only 11% of job titles do, and listings with an AI-related title carry a 27% salary premium — $113,625 median versus $89,438 (Confidence: High — primary source directly fetched, every number matches). A separate analysis by Kevin Indig found comparable proportions — 59.5% of descriptions versus 15.5% of titles mention AI, with a similar premium for AI-titled roles. This is worth a caveat the original coverage glosses over: Indig's analysis reportedly draws on 946 listings from SalaryGuide.com, the same data provider partnered with Search for Hire's study, so this is a second analytical cut of overlapping data, not an independent replication with its own data collection (Confidence: Medium).

Read together, the two studies still support the same conclusion: the job title "SEO" is being hollowed out from the inside. Employers are quietly requiring AI fluency in the job description while the title lags behind, and the market is already pricing that gap — a 27% spread is not noise.

Part 5 — The broader layoff wave, kept in its proper lane

TechCrunch's running tracker of 2026 tech layoffs where employers explicitly cited AI shows, as of July 6, 2026, company-level cuts including Oracle (21,000, 13%), Meta (8,000, 10%), Amazon (roughly 16,000 corporate roles, ~9%, January 2026), PayPal (4,500+, 20%), Cisco (4,000, 5%) and Block (4,000, roughly 50%), among others — cumulative 2026 tech-sector layoffs near 120,000, with AI cited as the single most common reason specifically in May 2026, per Challenger, Gray & Christmas data (Confidence: High — directly fetched, company table confirmed). Two things about this figure need correcting relative to earlier circulating claims: a specific "101,743 layoffs across four consecutive months, 23% of all cuts" figure could not be verified against the tracker as it currently reads, and none of the companies in this list are marketing or SEO agencies — this is broad-economy context for the climate layoffs are happening in, not direct evidence that AI search specifically is displacing marketing roles.

What everyone is missing

The layoffs and the salary premium are the same story, not two separate ones. Commentary treats "AI is killing SEO jobs" and "AI skills pay more in SEO jobs" as contradictory signals. They're not — they're the bifurcation happening in real time. Commodity-content operations (Chegg, HubSpot-style publishers, BI's old traffic-farming model) are being cut because the traffic that funded them is gone or going. Meanwhile the surviving roles are being redefined around AI fluency, and the market pays for that fluency specifically because it's still scarce relative to demand.

Nobody has actually shown AI search caused the HubSpot collapse, and repeating it anyway is a symptom of the problem it claims to diagnose. An industry anxious about AI search has adopted a convenient origin story faster than it has fact-checked it — the same pattern of confident, unsourced repetition that AI-generated content is accused of. The correction here isn't pedantic: if HubSpot's collapse was mostly algorithmic-update and content-quality driven, then "write better core content, stop optimizing for volume" remains sound advice independent of AI search, which changes the practical takeaway entirely.

Agency holding companies are cutting headcount for reasons that predate and extend beyond AI. WPP's "Elevate28" and the Omnicom-IPG merger were both underway before generative AI dominated earnings calls; AI is the language used to describe cuts that are also about merger synergies, margin pressure, and multi-year cost programs. The 41-vs-8 word-count comparison in Omnicom's annual report is a genuinely interesting signal of rhetorical priority, but a keyword tally is not proof that AI, rather than merger economics, is driving the actual reduction.

The account-team compression at Meta is the most underrated data point in this set, because it describes de-skilling of a service tier, not elimination of a headcount line. Nobody at Meta was laid off to make room for Advantage+; advertisers simply stopped being sold a human-staffed tier that used to exist. That's a template other platforms and agencies are likely to follow — not a mass layoff announcement, but a quiet discontinuation of a service level, which shows up nowhere in employment statistics.

Future predictions

• Salary-premium data for AI-titled SEO roles will keep widening before it narrows, as the labeling lag (54.9% of descriptions vs. 11% of titles) closes and titles catch up to actual job content (Confidence: Medium — based on directly-verified current gap, extrapolated forward).
• More holding-company layoff announcements will explicitly cite AI in company language even where the underlying driver is merger integration or margin targets, because it is now the industry's default explanatory frame (Confidence: Medium — pattern observed in Omnicom/WPP language, not a verified causal claim).
• Expect more disputed attribution cases like HubSpot, where a real traffic collapse gets retroactively pinned on AI Overviews by secondary commentary even when the primary reporting blames algorithm updates or content strategy (Confidence: Medium — pattern-based, not itself a specific forecast with a named source).
• Publisher-side organic traffic data (the Define Media Group 42% figure) will likely be updated or contested within the next year as more sites report Q1–Q2 2026 numbers; treat the 42% figure as a snapshot, not a settled endpoint (Confidence: Low — forecast, no verified follow-up data yet).
• Account-team compression on the Meta model (fewer humans, more self-serve automation) is plausible to spread to other major ad platforms, though no verified reporting yet confirms this occurring elsewhere (Confidence: Low — extrapolation from a single documented case).

Practical takeaways

1. Treat job titles as a lagging indicator: 89% of AI-relevant SEO roles still carry old titles, so hiring managers evaluating candidates — and candidates evaluating postings — should read job descriptions, not titles, for actual AI-skill requirements.
2. Before attributing any specific traffic collapse to AI Overviews, check whether the outlet covering it names AI Overviews explicitly or whether (as with HubSpot) the real cause is an algorithm update or content-quality issue; misdiagnosis leads to the wrong fix.
3. Commodity, traffic-volume content strategies are the demonstrated failure mode across Chegg, Business Insider, and the disputed HubSpot case — shifting budget toward defensible topical depth and entity clarity is better supported by the evidence than chasing AI-visibility hacks. See Semantic Authority and Entity & Knowledge Architecture for how that shift gets operationalized.
4. Agencies restructuring account-service tiers (per the Meta case) should audit which client-facing roles are genuinely irreplaceable versus which have already been quietly automated elsewhere — the compression may already be underway before headcount reports reflect it.
5. Because AI-titled roles carry a real, measured premium, teams should invest in monitoring how their brand and content actually surface in AI-driven answers, not just traditional rankings — a capability gap the salary data confirms employers are already paying for. See LLM Visibility Monitoring and Generative Engine Optimization.
6. Don't over-index on tech-sector-wide AI layoff totals (TechCrunch's ~120,000 figure) as marketing-specific evidence — none of the named companies in that tracker are agencies or SEO teams, and using it as such overstates the case beyond what's verified.