Notion Hires 3 Sales Reps Per Engineer in May 2026. Supabase Hires 3 Engineers Per Sales Rep.

The problem: Every quarter a tech publication runs the "engineering hiring is dying because of AI" thinkpiece off three Twitter screenshots and a vibes-based read of LinkedIn job counts. That framing falls apart when you actually pull careers-page data at scale. When you crawl 50 major SaaS companies' open-role listings on the same day with the same tooling, the picture isn't engineering hiring is collapsing. It's that the engineering-to-sales ratio splits cleanly along SaaS category lines — productivity and design tools tilt sales-heavy because their products are mature, while developer-tools and fintech-infrastructure tilt engineering-heavy because the product is the engineering. Same hiring window, same audit cohort, two completely different operating models. This post is what the May 2026 SaaS hiring leaderboard actually looks like.

This is a documentary audit of careers-page open-role mix at 50 major SaaS companies on 9 May 2026. 10 companies returned reliable engineering-vs-sales data. 14 companies were excluded for explicit technical reasons (6 ATS-hosted, 8 returned product-page copy instead of job titles). 25 fell below the 20-openings noise threshold. Every count is reproducible by re-running the same Apify actor against the same URL list. The leaderboard is real, the methodology disclosure is non-optional, and the AI is killing developer hiring framing journalists will reach for is not what the data shows.

What is the SaaS hiring mix audit? A point-in-time extraction of every open job role from major SaaS companies' careers pages, classified by role category (engineering / sales / marketing / customer success / data / product / design / operations / finance / legal / people / other) and reduced to a single per-company engineering-to-sales ratio. The metric measures current open-role mix — what each company is hiring right now — not headcount composition, which is a different question this audit cannot answer.

Why it matters: Open-role mix is a real-time strategic signal. A SaaS company hiring 3 sales reps for every engineer is in monetisation-and-expansion phase. A SaaS company hiring 3 engineers for every sales rep is in product-buildout phase. The mix is a leading indicator of where the company thinks its next dollar of revenue comes from — new product surface (engineering-led) or converting existing market awareness (GTM-led).

Use it when: You're benchmarking a SaaS competitor's strategic posture, building investor research on a private SaaS cohort, surfacing the dev-tools / productivity / fintech-infra category divide, or tracking a single company's hiring-mix shift across snapshots.

Key findings

• Notion filed 55 open Sales roles vs 19 open Engineering roles in May 2026 — a 0.35 engineering-to-sales ratio, the most sales-heavy major SaaS in the cohort. Notion's open Product roles: 1.
• Supabase posted 27 Engineering vs 10 Sales open roles — a 2.70 ratio, the most engineering-heavy in the cohort. The 7.7x spread between Notion and Supabase is the cohort's headline number.
• Three SaaS companies cluster sales-heavy at sub-1.00 ratio: Notion (0.35), Figma (0.56), Linear (0.63). All three are productivity / design / planning tools positioned as PLG products but hiring GTM-led in May 2026.
• Six SaaS companies cluster engineering-heavy at 1.83 or above: Supabase (2.70), Webflow (2.57), Vercel (2.33), Mercury (2.33), Ramp (1.86), Intercom (1.83). Developer tools and fintech infrastructure dominate this half.
• Of 50 audited companies, 14 are excluded with disclosed technical reasons: 6 use ATS-hosted careers (GitHub, Snowflake, Slack, Twilio, Zendesk, Workday on iCIMS / Phenom / Eightfold / Workday) and 8 returned product-page copy instead of job listings (Datadog, Salesforce, Asana, Mailchimp, Miro, Databricks, Loom, Klaviyo). The exclusions are the audit's quality bar, not its blind spot.
• Cohort total: 840 open roles across 10 reliable-data companies. 229 engineering. 195 sales. Cohort engineering-to-sales ratio: 1.17.

In this article: The leaderboard · Story A: Notion's 0.35 outlier · Story B: Dev-tools / fintech-infra cohort · Story C: PLG-collaboration tilt · Story D: Intercom AI-product outlier · Story E: 14 excluded companies · Methodology · What the data does NOT support · Press lift-out

The 2026 SaaS hiring mix leaderboard — 10 companies

This is the full reliable cohort: 50 major SaaS companies audited via careers-page extraction on 9 May 2026, filtered to those returning a populated engineeringToSalesRatio with totalOpenings >= 20 and not flagged atsHostedUnsupported or rejectedAsProductCopy. Sorted by engineering-to-sales ratio ascending — most sales-heavy at the top, most engineering-heavy at the bottom.

Cohort totals: 840 open roles. 229 engineering. 195 sales. Cohort engineering-to-sales ratio: 1.17. Underlying source: 50 major SaaS companies' public careers pages, queried via the saas-competitive-intel Apify actor in mode: "standard" on 9 May 2026.

The shape of the leaderboard is more interesting than the top of it. There's a sales-heavy cluster (Notion / Figma / Linear / Shopify) at sub-1.00, then a clean break to a 1.83+ engineering-heavy cluster of six. There's no middle. No company in the cohort sits between 1.00 and 1.83 — the audit found a bimodal distribution, not a continuum. The category split (productivity / collaboration vs developer-tools / fintech-infra) explains it cleanly. The next five sub-stories untangle that.

Per-company role-category breakdown — top 5 by total openings

The leaderboard ratio is one number. The full role-category breakdown is the texture behind it. These are the five highest-volume hirers in the reliable cohort.

Notion (141 openings — most sales-heavy at 0.35)

Notion's hiring is overwhelmingly go-to-market: Sales (55) + Customer Success (23) + Marketing (11) = 89 of 141, or 63% GTM. Engineering at 19 is 13.5% of total. Product at 1 is the killer detail — Notion is not hiring product builders right now, only people to sell what already exists.

Figma (157 openings — sales-heavy at 0.56)

Figma's sales (45) outpaces engineering (25) by 1.8x. Combined GTM (sales + marketing + customer success) totals 66 of 157 (42%) vs engineering + product + design + data at 35 (22%). The "Other" category at 46 is unusually large — likely operational and cross-functional roles the keyword classifier did not tag with confidence.

Intercom (171 openings — most engineering volume in cohort at 64)

Intercom is the cohort's largest absolute-engineering hirer: 64 engineering + 14 data + 9 product = 87 product-builder openings, against 35 sales. The 14 data roles is the highest data-team count in the audit by absolute volume — reflects Fin AI Agent, the company's flagship AI agent product, as a re-platforming priority.

Ramp (109 openings — fintech, engineering-heavy at 1.86)

Ramp is the fintech-corridor engineering hirer — 26 engineering + 2 product = 28 product-builder roles vs 14 sales. The "Other" cluster at 27 is unusually large for a fintech and likely reflects compliance / risk / partnerships roles the classifier flagged as cross-functional rather than tagged as a specific function.

Plaid (111 openings — pre-categorisation, no ratio computed)

Plaid's careers page parsed cleanly with 111 jobs detected, but the actor's role-category classifier did not populate byCategory for this run. This appears to be a per-extraction-method gap rather than a Plaid-specific data issue, and Plaid is treated as cohort-adjacent — in the high-volume tier but not in the ratio leaderboard. A future re-run with the explicit category-classifier flag enabled is the obvious fix.

Excluded cohort — 14 companies, with technical reasons

The 50-company input list resolves into four buckets. The 10 reliable-data leaderboard above is one. The other three are below.

These 14 for-cause exclusions are not a quality signal — they are a scope disclosure. The audit can only see what the actor's careers crawler can extract. ATS-hosted careers pages are auth-gated and would require per-tenant API tokens to crawl. The product-copy-rejection cohort triggered the actor's product-copy validator: extracted "titles" failed the role-keyword check (none contained Engineer / Manager / Director / Sales / etc), and the actor correctly suppressed them rather than reported bogus job counts.

The product-copy validator is the audit's most important quality safeguard. A previous build of the actor returned Datadog's product-feature page text as 173 "job titles" — exactly the false-positive that destroys an audit's credibility. Build 2.0.7 catches that pattern and returns 0 with a rejectedAsProductCopy: true flag. Datadog hires actively; the actor cannot read their careers page reliably; that is a tooling limit, not a Datadog signal.

Story A — Notion's 0.35: the cohort's most sales-heavy SaaS

Notion's hiring distribution in May 2026:

• 55 Sales (39% of total)
• 23 Customer Success (16%)
• 19 Engineering (13%)
• 11 Marketing
• 16 Other
• The remainder spread across finance, operations, legal, product, data

GTM (Sales + Customer Success + Marketing) totals 89 of 141 = 63% of all open roles. Engineering at 19 is 13.5%. Product at 1 is the tell — Notion is not hiring product builders. It is hiring people to sell, support, and market what already exists.

The framing matters. A 0.35 engineering-to-sales ratio doesn't mean Notion is doing badly. It means Notion's leadership has decided the existing product is competitive enough and the priority is converting market awareness into ARR. The hiring mix is a deliberate strategic choice that reads cleanly off the open-role data.

This is the operating posture of a company in monetisation-and-expansion phase, not a company adding new product surface. Notion's existing engineering team is large; the 0.35 ratio is a directional signal about current hiring, not a description of total headcount. The underlying SaaS pricing time machine 2020-2026 audit captured Notion's pricing-page evolution over the same window — pricing tier complexity grew while product surface stabilised, which is exactly what the hiring mix would predict.

Story B — The dev-tools / fintech-infra cohort: 2.70 down to 1.83

Six companies in the cohort sit at engineering-to-sales ratios between 1.83 and 2.70 — the leaderboard's engineering-heavy half. The shape is a recognisable archetype:

The pattern: infrastructure SaaS hires engineers because the product IS engineering. Ratios above 2.0 are typical for companies still in product-buildout phase. They soften toward 1.5 as the company matures and adds GTM depth — Ramp at 1.86 and Intercom at 1.83 are visibly further along that curve than Supabase at 2.70.

This sits directly against the "AI is replacing developer hiring" thinkpiece narrative. The Stack Overflow question decline 2020-2026 audit documented that developer questions on Stack Overflow collapsed 95%+ after ChatGPT — yet engineering remains the dominant hiring focus for every dev-tools and fintech-infra company in the May 2026 cohort. The two findings together suggest the Stack Overflow collapse reflects channel shift (developers ask AI assistants instead of forum threads), not demand shift (developers losing jobs). Where developers ask their questions changed; whether companies still hire developers has not.

Story C — The PLG-collaboration tilt: Notion, Figma, Linear

Three productivity / design / planning tools share a sales-heavy hiring pattern in May 2026:

• Notion 0.35 — most sales-heavy in the cohort
• Figma 0.56
• Linear 0.63

All three are positioned as PLG (product-led growth) tools — meaning the product itself is supposed to do the selling, with self-serve signups and bottom-up team adoption replacing classic enterprise sales motion. Their May 2026 hiring tells a different story. All three are GTM-led in current open roles, with sales and customer success roles dramatically outpacing engineering.

The honest read: collaboration / productivity software is a mature market in 2026. Notion competes with Atlassian's Confluence and Microsoft Loop. Figma competes with Adobe XD and Penpot. Linear competes with Jira and Shortcut. Differentiation in mature markets comes from enterprise sales motion, post-sale expansion, and customer success — not new features. The hiring mix encodes that strategic reality across all three companies simultaneously.

PLG is the acquisition motion. Enterprise sales is the monetisation motion. Mature PLG companies typically run both, and the open-role data shows where the marginal next dollar of headcount investment is going. In May 2026 for Notion, Figma, and Linear, it is going to GTM.

Story D — Why Intercom is the AI-product outlier

Intercom is the only audited company simultaneously in three high-percentile cohorts:

1. High-volume hiring — 171 openings, 2nd most in the audit (behind only Figma's 157, which Intercom narrowly beats).
2. Mid-engineering ratio — 1.83 engineering-to-sales, sitting in the cohort's engineering-heavy half but the lower end of it.
3. Highest absolute data-team hiring — 14 open data roles, the most of any company in the audit.

Combined: 64 engineering + 14 data + 9 product = 87 product-builder openings against 35 sales. This is the hiring mix of a company building deep new product surface — specifically, Fin AI Agent, Intercom's flagship 2024-2026 AI agent product that automates customer-support conversations directly within the existing Intercom platform.

The hiring mix is the strategic signal. Intercom is not in monetisation phase like Notion. It is in re-platforming phase — building a fundamentally new product (an autonomous AI agent for customer support) on top of an existing GTM motion. The 14 data roles in particular signal investment in the model evaluation, training-data curation, and conversation-quality measurement infrastructure that an AI-agent product needs to ship reliably. Companies that aren't building AI products don't hire 14 data roles in a single quarter.

Story E — The 14 excluded companies: methodology disclosure as the audit's strength

The honest framing of any audit is what it can and cannot see. This audit:

• Sees 10 companies with reliable hiring-mix data and role-category breakdown
• Tags 6 ATS-hosted companies (GitHub, Zendesk, Slack, Twilio, Snowflake, Workday) as atsHostedUnsupported: true rather than silently returning 0 — they exist, they hire, but the actor's crawler cannot read their tenant-auth-gated careers pages
• Tags 8 companies (Datadog, Salesforce, Asana, Mailchimp, Miro, Databricks, Loom, Klaviyo) as rejectedAsProductCopy: true when the extracted "titles" failed role-keyword checks — better to suppress than report bogus counts
• Acknowledges the under-20-openings cohort (25 companies) is too small for a stable engineering-to-sales ratio

The exclusions are the audit's quality bar. A press fact-check that picks up "why isn't Snowflake in the leaderboard?" gets a clean technical answer: Snowflake's careers are hosted on Phenom; the actor cannot crawl Phenom without per-tenant credentials; this is documented; the audit's exclusion is structural, not a Snowflake-specific verdict.

This is the same methodology-disclosure-as-strength move the SEC executive departure index 2024 used when 8 distressed-but-listed names failed canonical SEC EDGAR resolution and got listed transparently in the caveats rather than silently dropped. An audit that cannot defend its exclusions is an audit no journalist will quote.

Methodology

• Tool: ApifyForge's saas-competitive-intel Apify actor, build 2.0.7, in mode: "standard" (pricing + careers + tech-stack extraction). Build 2.0.7 is the post-fix sweep: a prior run against build 2.0.6 surfaced systematic data-quality issues (ATS-hosted careers returning silent zeros, product-page copy misclassified as job titles for Datadog, wrong-page resolution for Zoom / Rippling / ServiceNow). Build 2.0.7 shipped four targeted fixes — ATS-host detection (iCIMS / Phenom / Eightfold / Workday tagged with atsHostedUnsupported: true and excluded from cohort math), product-copy validator (rejectedAsProductCopy: true when extracted "titles" fail role-keyword check), careers-page resolver (correct subdomain detection for canonical careers URL), and per-role category classifier (job titles tagged engineering / sales / marketing / customerSuccess / data / product / design / operations / finance / legal / people / other, with engineeringToSalesRatio computed when both denominators are >= 1).
• Cohort: 50 major SaaS companies — enterprise (Salesforce, HubSpot, Atlassian, Workday, ServiceNow), data / analytics (Datadog, Snowflake, Databricks, MongoDB, Elastic), dev-tools / collaboration (GitHub, GitLab, Linear, Vercel, Netlify, Figma, Notion, Asana, Monday, ClickUp, Airtable, Zoom, Slack, Canva, Webflow), payments / fintech (Stripe, Plaid, Brex, Ramp, Mercury), API / communication (Twilio, Intercom, Zendesk), HR / hiring (Gusto, Rippling, Deel, Greenhouse, Lever), marketing (Mailchimp, Klaviyo, Braze), infra (Cloudflare, Fastly, Supabase), commerce (Shopify), and other (Calendly, Loom, Miro, Zapier, n8n).
• Capture date: 2026-05-09. Single point-in-time snapshot of public careers pages.
• Per-company extraction: the actor finds the careers URL via canonical-domain heuristics (subdomain probes for careers.<domain>, /careers, /jobs, etc.), follows ATS-host redirects (Greenhouse / Lever / Ashby use predictable hosts), and extracts job titles + a per-role category tag. The role-category classifier is keyword-based; no LLM.
• Exclusion logic: companies tagged atsHostedUnsupported: true (ATS host is iCIMS / Phenom / Eightfold / Workday — auth-gated) or rejectedAsProductCopy: true (extracted titles failed role-keyword check) are excluded from the audit's reliable cohort. Companies with totalOpenings < 20 are excluded from ratio computation due to small-N noise.
• Engineering-to-sales ratio: roleCategoryCounts.engineering / roleCategoryCounts.sales, computed when both denominators are >= 1.
• Reproduction: every count in this post is re-fetchable by re-running the same actor with the same 50-URL list. Anyone with an Apify account can reproduce the audit. Source data archived at storage/research/saas-hiring-mix/audit.json.

The two-build iteration is itself part of the methodology disclosure. A v2.0.6 sweep would have published Datadog at 173 false-positive "open roles" and silent zeros for GitHub / Snowflake / Slack. Catching those at the actor layer before publishing is what makes the v2.0.7 leaderboard above defensible.

What the data does NOT support

This is the section journalists should read before quoting any number from this post in a news lead. The audit measures one thing — open-role mix on 9 May 2026 across a specific 50-company cohort — and is honest only when its scope limits are stated.

• A point-in-time snapshot is not a hiring trend. The May 2026 capture shows current open-role mix, not year-over-year direction. To answer "is engineering hiring slowing?" the audit needs a baseline snapshot from a year earlier, which does not exist for this dataset. Coverage should frame as "the mix in May 2026" not "the mix has shifted toward sales" without prior-period data.
• Open-role count is not headcount; the ratio is not workforce composition. Notion's 0.35 ratio reflects what is currently being hired. Notion's existing engineering team headcount is plausibly 5-10x the size of the open-engineering-role count. A company that finished its engineering buildout last quarter and is now in GTM scaling phase will show a sales-heavy current-roles mix even though the company is engineering-heavy in absolute headcount terms. The leaderboard ratio is direction, not composition.
• Role classification is keyword-based, not semantic. "Account Executive — Greenfield" is correctly tagged sales. "Solutions Engineer" sits ambiguously between sales and engineering (the actor likely tagged it engineering by Engineer-keyword precedence). Edge cases like "Forward Deployed Engineer" — a sales-engineer hybrid title common at Vercel — will tilt classifications. The "Other" category exists for titles the classifier could not tag with confidence, and it is intentionally large for that reason.
• The 14 excluded companies represent real gaps, not silent omissions. Snowflake, GitHub, Twilio, Slack, Zendesk, and Workday all have hundreds of open roles each that this audit cannot read because their careers pages are auth-gated ATS systems. The exclusion is a tooling limit, not a comment on those companies. Coverage should disclose the exclusion with the technical reason (atsHostedUnsupported: true per ATS host) rather than treat the absence as a verdict.
• The product-copy-rejection cohort (Datadog et al) is a data-quality safeguard, not a critique of those companies. Datadog hires actively. The actor cannot read their careers page reliably because its extraction layer hits product-marketing copy on the same DOM path that other companies serve job listings on. The rejectedAsProductCopy: true flag is the actor doing the right thing — refusing to publish 173 phantom roles — not a Datadog signal.
• Cohort selection is editorial. The 50 companies in the input list are well-known SaaS brands; a different selection would shift cohort medians but not the named-leaderboard story. The post should be read as "of these 50 companies" rather than "across all SaaS".
• The actor undercounts JS-rendered SPA careers pages. Zoom, Rippling, ServiceNow, Atlassian return 1-3 openings each because their careers pages are JS-rendered single-page apps that serve a stub when the actor's Cheerio-based crawler hits them. These low counts are technical artefacts; the companies hire significantly more than the audit shows. A future Playwright-based extraction layer would close this gap and likely grow the reliable cohort from 10 to 25+ companies.

Press lift-out for journalists

A copy-paste-ready version of the headline finding:

"A 2026 ApifyForge audit of 50 major SaaS companies' careers pages on 9 May 2026 found 10 with reliable engineering-vs-sales role-mix data. Within that cohort, the engineering-to-sales ratio ranges from 0.35 (Notion: 19 engineering openings vs 55 sales — 3 sales reps for every engineer) to 2.70 (Supabase: 27 engineering vs 10 sales — nearly 3 engineers for every sales rep). The sales-heavy cohort — Notion (0.35), Figma (0.56), Linear (0.63), Shopify (1.00) — is concentrated in productivity / design / e-commerce SaaS in monetisation-and-expansion phase. The engineering-heavy cohort — Supabase (2.70), Webflow (2.57), Vercel (2.33), Mercury (2.33), Ramp (1.86), Intercom (1.83) — is concentrated in developer-tools and fintech-infrastructure SaaS still in product-buildout phase. 14 audited companies were excluded with technical reasons: 6 use ATS-hosted careers systems (Workday, iCIMS, Eightfold, Phenom) that don't expose data without per-tenant credentials, and 8 returned product-page copy from their site nav rather than job listings — including Datadog, Salesforce, and Databricks."

Source for citation: ApifyForge, "Notion Hires 3 Sales Reps Per Engineer in May 2026. Supabase Hires 3 Engineers Per Sales Rep," 9 May 2026. Underlying data: 50 major SaaS companies' public careers pages, queried via the saas-competitive-intel actor (build 2.0.7) on 9 May 2026.

Embeddable visuals

Three chart blocks, all reproducible from the tables in this post.

Chart 1 — The 2026 SaaS hiring-mix leaderboard, ranked by engineering-to-sales ratio

Horizontal bar chart, sorted ascending (most sales-heavy at top). The two ends of the chart are the visual story. Notion at 0.35 anchors the top — a short bar in a sales-coloured fill. Supabase at 2.70 anchors the bottom — a long bar in an engineering-coloured fill. The 1.00 reference line cuts cleanly between rank 4 (Shopify) and rank 5 (Intercom), separating the sales-heavy and engineering-heavy clusters with no company in between. Y-axis: company name. X-axis: engineering-to-sales ratio. Source line: "50 major SaaS companies' careers pages, captured 9 May 2026 via the saas-competitive-intel actor."

Chart 2 — Notion's role-category distribution: where the 141 openings actually go

Horizontal bar chart of Notion's 11 role-category counts in descending order. Sales (55) is the dominant bar, more than double the next-largest (Customer Success at 23). Engineering at 19 is third. Then Other (16), Marketing (11), Finance (6), Operations (5), People (2), Legal (2), and a near-invisible Product (1) and Data (1) at the bottom of the chart. Headline: "Notion's open-role distribution in May 2026: 89 of 141 are GTM. Product is 1." Source line: "Notion careers page, captured 9 May 2026."

Chart 3 — The category divide: where SaaS companies cluster on hiring mix

Scatter plot. X-axis: total open roles (log scale). Y-axis: engineering-to-sales ratio. Each company is a labelled dot, colour-coded by category (productivity / design / collaboration in one colour, developer tools / fintech infrastructure in another). The horizontal 1.0 reference line cuts cleanly between the two clusters — the productivity / design / collaboration dots all sit below 1.0, the developer-tools / fintech-infrastructure dots all sit above 1.83. The empty band between the two clusters is the chart's whole point. Headline: "The SaaS hiring mix splits along category lines, not company size." Source line: "50 major SaaS companies' careers pages, captured 9 May 2026."

Frequently asked questions

What is the SaaS engineering-to-sales hiring ratio?

It is the count of currently-open engineering roles at a company divided by the count of currently-open sales roles, computed from public careers-page data. A ratio above 1.0 means the company is hiring more engineers than sales reps right now; a ratio below 1.0 means more sales reps than engineers. The metric measures current open-role mix, not headcount composition — Notion's 0.35 ratio means current hiring is sales-heavy, not that Notion's existing engineering team is small.

Why is Notion the most sales-heavy SaaS in the May 2026 audit?

Notion posted 55 open Sales roles, 23 Customer Success, and only 19 Engineering in May 2026 — a 0.35 engineering-to-sales ratio. Combined GTM (Sales + Customer Success + Marketing) was 89 of 141 total open roles, or 63% of all hiring. Product was 1 open role. The shape is the operating posture of a company in monetisation-and-expansion phase rather than product-buildout phase: Notion's existing product is competitive, and the priority is converting market awareness into ARR. The mix is a deliberate strategic choice, not a hiring slowdown.

Why is Supabase the most engineering-heavy SaaS in the audit?

Supabase posted 27 open Engineering roles vs 10 open Sales — a 2.70 engineering-to-sales ratio, the highest in the cohort. Supabase's product (open-source Postgres-as-a-service plus authentication, storage, and edge functions) is fundamentally engineering surface — the strategic value is the platform itself. Companies in this archetype hire engineers because the product is the engineering. Supabase shares this pattern with Webflow (2.57), Vercel (2.33), and Mercury (2.33).

Does this audit show engineering hiring is dying due to AI?

No, and this is the framing the data does not support. Six of ten companies in the cohort are engineering-heavy at 1.83+ ratio — Intercom, Ramp, Mercury, Vercel, Webflow, Supabase. Developer tools and fintech infrastructure companies are still hiring engineers at multiples of their sales hiring. The Stack Overflow question decline 2020-2026 audit found dev questions on Stack Overflow collapsed 95%+ post-ChatGPT, but this audit shows engineering hiring at dev-tools companies has not collapsed — meaning the Stack Overflow shift reflects channel migration (developers ask AI assistants), not demand collapse (developers losing jobs).

Why isn't Snowflake or GitHub in the leaderboard?

Both use ATS-hosted careers systems — Snowflake on Phenom, GitHub on iCIMS. ATS-hosted careers pages are auth-gated and require per-tenant API tokens to crawl programmatically. The actor tags these companies as atsHostedUnsupported: true and excludes them from cohort math rather than silently returning 0 open roles. The same exclusion applies to Slack (Workday), Twilio (Eightfold), Zendesk (Workday), and Workday itself (Workday). All six companies hire actively; the audit just cannot read their careers data.

Why was Datadog excluded from the cohort?

Datadog's careers page returned site-navigation copy — product-marketing text, feature names, and category headers — instead of recognisable job titles when the actor's extraction ran. The actor's product-copy validator (rejectedAsProductCopy: true) caught this: extracted "titles" failed the role-keyword check (none contained Engineer / Manager / Director / Sales / etc) and the actor correctly suppressed them. A previous build of the actor returned Datadog's product-page text as 173 phantom job titles — the kind of false-positive that destroys an audit's credibility. Build 2.0.7 catches it. Datadog hires actively; the actor cannot read their careers page reliably; that is a tooling limit, not a Datadog signal.

How does this audit compare to other ApifyForge backlink-bait audits?

This is the second SaaS-named-entity audit in the series, after SaaS pricing time machine 2020-2026. Both use ApifyForge actors against SaaS-company-named entity sets — pricing pages in one, careers pages in this one — and both surface category-level patterns (pricing-tier complexity by SaaS category in #6, hiring-mix split by SaaS category in #13). The pair forms a methodology cluster within the larger backlink-bait series.

Can I reproduce the audit myself?

Yes. The saas-competitive-intel actor on Apify Store accepts a list of company URLs and runs the same extraction pipeline. With the same 50-URL input list and mode: "standard", anyone with an Apify account can re-run the audit and get a comparable May-2026-baseline dataset. Source data for this post is archived at storage/research/saas-hiring-mix/audit.json and every count in the leaderboard is re-fetchable.

Related ApifyForge backlink-bait audits

This is post #13 in an ongoing series of public-data audits using ApifyForge actors. Each post documents a specific industry, regulatory, or platform dataset; together they form a citation network of named-entity research that journalists, analysts, and AI systems can pull from. The full series so far:

1. Defense contractor lobbying ROI 2024 — Senate LDA filings cross-referenced with USAspending contracts.
2. FDA 510(k) shortcut vs PMA 2024 — medical device clearance pathway audit.
3. CFPB credit bureau complaint dominance 2024 — Consumer Financial Protection Bureau complaint database.
4. Tech podcast cemetery 2026 — RSS-feed audit of dormant tech podcasts.
5. SEC insider sales 2024 leaderboard — SEC EDGAR Form 4 named-entity audit.
6. SaaS pricing time machine 2020-2026 — methodology sibling: SaaS-named-entity audit on pricing pages.
7. 2024 academic retractions publisher leaderboard — Retraction Watch + publisher cross-reference.
8. Medical debt collection 2024 CFPB leaderboard — CFPB collector-by-collector ranking.
9. Stack Overflow question decline 2020-2026 — pairs naturally: SO showed dev questions collapsed 95%+; this audit shows engineering hiring did not.
10. Trustpilot two-tier trust index 2026 — verified-vs-unverified review imbalance.
11. OSS maintainer burnout index 2026 — GitHub maintainer activity decay.
12. SEC executive departure index 2024 — 8-K Item 5.02 officer-departure leaderboard.
13. Notion Hires 3 Sales Reps Per Engineer in May 2026. Supabase Hires 3 Engineers Per Sales Rep — this post.

Posts #6 and #13 are the SaaS-named-entity pair: both audit major-SaaS-company datasets via ApifyForge actors against public, reproducible sources (pricing pages, careers pages). Posts #9 and #13 are the developer-hiring pair: post #9 documented that developer questions on Stack Overflow collapsed 95%+ after ChatGPT released; this post shows that engineering remains the hiring focus for developer-tools and fintech-infrastructure SaaS in May 2026. Together the two findings argue the Stack Overflow collapse reflects channel migration, not demand collapse.

Ryan Clinton publishes Apify actors and MCP servers as ryanclinton and builds developer tools at ApifyForge. The audit above was produced via the saas-competitive-intel actor (build 2.0.7) in mode: "standard" against 50 major SaaS companies' public careers pages on 9 May 2026; the methodology, analysis, and framing are independent of any product positioning.

Last updated: May 2026