Wing and Magic built large informational content engines — and both are now losing 83–88% of their peak traffic while hiring sales and paid-search people, not SEOs. Hire Bloom skipped SEO almost entirely, tested Google Ads for one quarter, stopped, and is now betting on niche case studies, original research, and an LLM-readable facts page. The opening for Stealth Agents is not "write more blog posts." It is own the buying-stage queries and the third-party pages AI answers cite.
Monthly organic ETV (estimated traffic value), US. Log scale — both companies fell by roughly the same proportion, from very different bases.
Ad volume, price per hour, and off-site authority — the three variables that separate the models.
Click any card for the full teardown.
A content machine in free fall, pivoting to sales and AI agents.
A decade-old ad program and a content engine that just lost 84% of its traffic.
A $20–25M staffing arbitrage business with almost no digital moat — and a sharp GEO play.
| Company | Model | Price | Organic peak → now | Google / Meta ads | Ref. domains | Verdict |
|---|---|---|---|---|---|---|
| Wing Assistant | VA subscription, US + offshore | $699/80h · $999/160h | 81.8k → 9.7k −88% | 500 / 26 active | 2,239 | Content engine decaying; pivoting to AI agents & sales |
| Magic | VA subscription, 24/7 pool | $360–540/week (40h) | 29.3k → 4.8k −84% | 97 / 0 | 1,964 | Old ads keep running; adding 14 paid-search managers |
| Hire Bloom | Remote staffing, embedded CSRs | $13/hr flat | ~950/mo no SEO | 8 / 0 | 11 | Sales/referral-led; GEO & niche proof are the edge |
Organic = free visits from Google. Paid = ads. A quarter is three months. Start dates come from Wayback Machine captures, Google/Meta ad libraries, and traffic data — not marketing copy. The pattern is blunt: content was ramped after the traffic peak, not before it.
The dates you asked for, side by side.
| Company | Founded | Blog started | Content ramp | Google Ads | Meta Ads | Organic peak | Decline began |
|---|---|---|---|---|---|---|---|
| Wing Assistant | 2021 | Jan 2022 | Q4 2024 → Q2 2025 | Jul 2023 | Oct 2026 | Dec 2024 (~82k) | 2025 |
| Magic | 2015 | Jul 2020 | Q2 2022 → Q2 2023 | Mar 2022 | None | May 2026 (~29k) | Jun 2026 |
| Hire Bloom | 2021 | Mar 2023 | Q4 2025 → Q2 2026 | Jan 2026 | None | — | — |
New pages first captured per quarter. This is the clearest proxy for “when did they ramp content?”
Bars are new pages. The line is average monthly organic visits. This is where the “we published more, but traffic still fell” becomes obvious.
One rail per company. Read top to bottom.
Wing’s largest content push was Q1–Q2 2025, immediately after its Dec 2024 traffic peak. Traffic still fell. Content volume did not create a second growth curve.
Magic nearly stopped in 2025 (9 pages) and lost 84% of its peak traffic by Aug 2026 — despite a brief Feb–May 2026 spike. The blog was propping up the long tail.
Wing’s first Google ad is from Jul 2023 (one creative ran 1,017 days); Magic’s from Mar 2022 (1,563 days). Paid search is the durable channel; the blog is the volatile one.
Google Ads Jan–Mar 2026 only, 8 creatives, then nothing. It now bets on case studies, awards, and AI-search visibility — not a scaled ad or content engine.
Wing is the clearest example of the category's broken playbook: a 967-URL site where 91% of ranked volume sits in the blog, and a 1,017-day-old Google Ads program that still carries the acquisition load. The blog is not compounding anymore — it is decaying from a higher base.
The 2026 energy is going into inside sales, field events, and an AI-agent product line (13 Inside Sales Manager roles, 8 Product Manager – AI roles). Nobody is hiring SEO.
Weakness Their #1 blog page ranks 40–62 for 201k-volume terms. They buy volume they cannot convert.
2,757 backlinks / 107 domains anchored “our Jobs Website.” Every role they post gets republished by aggregators — each copy a link. Supply and links from the same motion.
“$1,000 off… $200 each month for your first 5 months.” A concrete number beats a percentage. 26 ads, one offer page, every placement at once.
22 /partners/<tool> pages (OpenPhone, Ramp, Zendesk, Relay, Pilot…) co-marketing with tools buyers already use.
No move in 2026 reversed the decline. Traffic is the two-month average after the move vs. two months before.
| Date | Channel | Move | Reach | Lift |
|---|---|---|---|---|
| 2023-07-18 | Ads | Google text creative still running 1,017 days | — | — |
| 2026-04-23 | Partnerships | Partner-page network starts appearing | 10 partners | −17.2% |
| 2026-06-13 | Referral | ReferralCandy detected live | — | −23.3% |
| 2026-07-15 | Press | Forbes Best Startup Employers #100 post | 29 reactions | −30.3% |
| 2026-07-23 | Press | Clutch Certified post | 31 reactions | −30.3% |
| 2026-10-21 | Content | YouTube offer clips ($1,000/$2,500 credit) | 0 views | +1.3% |
| 2026-10-28 | Paid social | 26-ad Meta promo, $1,000 off | 26 active | +1.3% |
Magic runs the oldest ad program in the set — a Google campaign with creatives still live after 1,563 days — and no Meta ads. Its content engine had a real step-up in early 2026, then lost 45%, 55% and 31% in three consecutive months. The city-page cluster vanished entirely.
The hiring signal is unambiguous: 14 Paid Search Manager roles, 7 fraud analysts, multilingual CX, and AI-literate assistants. They are doubling down on Google and delivery, not content.
Weakness Their best organic asset is off-topic (“data annotation”) and their brand is absent from the ChatGPT answer for “best VA for small business.”
Magic's five best LinkedIn posts are job listings (17–59 reactions) vs 0–3 for product posts. The format doubles as proof the roster works.
“Pay nothing until you hire” over a face — the one brand film asset that carries the offer in the image. Their newer shorts dropped this and lost all engagement.
Reddit is actively shopping alternatives after Magic's pricing change. That is a buying-stage query a competitor can own.
Same pattern as Wing: verification without lift.
| Date | Channel | Move | Reach | Lift |
|---|---|---|---|---|
| 2022-03-03 | Ads | Google text creative still running 1,563 days | — | — |
| 2024-12-09 | Creator | Instagram reel with @linguamarina | 147 likes | −19.6% |
| 2025-09-02 | Content | IG reel “#1 in the business si…” | 63 likes | −13.5% |
| 2026-10-20 | Product | Magic app inside ChatGPT / Codex | 59 reactions | −1.7% |
Hire Bloom is not a content company — it is a sales- and referral-led staffing arbitrage business. It bills $13/hour (~$27k/seat/year), pays talent a fraction of that, and keeps an estimated 40–60% gross margin after payroll, equipment, compliance, and a US-based CSM.
Its digital presence is thin: ~950 organic visits/month, 105 keywords, 11 referring domains, and the most recent links are auto-generated spam directories created 3 Nov 2026. It ran 8 Google creatives Jan–Mar 2026, then stopped. No Meta ads.
What it does well is proof and positioning: named clients with dollar-quantified savings, original home-services research, and a deliberate /ai-info page built for LLM citation. That is the play to respect.
BYU-Pathway gives it English-fluent, US-accredited candidates in 63 countries — hard for conventional offshore recruiters to replicate, and the basis of the "US-educated" claim.
Bill $13/hr, pay talent roughly $5–8/hr, cover equipment and compliance, keep 40–60%. The social-impact story is genuinely the supply-side engine.
Home services (pest, HVAC, roofing, cleaning) is the concentration — and where it publishes the most research, case studies, and industry landing pages.
Self-reported annual savings from Hire Bloom case studies — unusually specific for the category.
BYU-Pathway is both the supply chain and the brand. Any change there hits volume and the "US-educated" claim simultaneously.
No active paid ads, ~950 organic visits/month, spammy backlinks. Inbound is reputation and AI citations — both contestable.
$13/hr is mid-market. BruntWork and Teem can undercut; Near matches the embedded model; SupportNinja out-spends on brand.
Retention was published as 86% until Aug 2026, now 70.4%. The AI research is self-published. Brand confusion with CrewBloom / Bloom Talent persists.
Payroll and labor law across 63 countries on a bootstrapped balance sheet is a real operational tail risk.
The growth story leans on Inc. and UV50 rankings; exact revenue is undisclosed and the Inc. profile is not publicly verifiable.
Ordered by evidence × timing × fit. The through-line: stop copying decaying content engines; own buying-stage queries, third-party citations, and verifiable proof. Each play is scoped to ship in days, not quarters.
Magic changed pricing and Reddit is actively shopping alternatives. Both competitors' own alternatives intent is under-served. Build /alternatives/wing-assistant and /alternatives/getmagic with a 12-row price/contract/replacement/AI comparison, sourced numbers, and switching checklist.
The “best VA for small business” answer is assembled from stackademic, assistantscout, rankite, and Forbes Advisor. Wing is named #2; Magic is absent; you are on none. Pitch each with one differentiator, one hard number, and a fact sheet.
“How much does a VA cost per hour” cites only cost pages — no provider owns it. Publish hourly cost by role, country, and hours/week, with 2026 dates, methodology, and a CSV. Add FAQ/HowTo schema and internal links.
Magic's top LinkedIn posts are job listings; Wing's job syndication carries 2,757 backlinks. Post 3 roles/week in “Position Filled! …” format, add JobPosting schema, submit feeds to Indeed / LinkedIn / Google Jobs.
They rank for almost nothing commercial (~950 visits/mo, 105 keywords). Own the exact home-services queries: “remote CSR for pest control,” “outsourced dispatch for HVAC,” “remote estimator for roofing.” Publish real outcome data, not self-reported savings.
Hire Bloom's /ai-info page is the right idea executed on a weak domain. Build an “official facts for AI” page with preferred description, canonical numbers, and mischaracterizations to avoid — then earn non-spam citations (associations, directories, research pickups) so the page actually gets used.
Wing and Magic both prove paid search is the durable acquisition channel — their oldest creatives have run 1,017 and 1,563 days. Don't buy broad “staffing agency” terms; buy exact niche intent and keep the winners running.
Monthly organic ETV and keyword count for all three; Google creative count and oldest-running creative; Meta active count; referring domains (excluding spam directories); AI answer inclusion for 3 buyer prompts; and Hire Bloom's pricing/retention claims each quarter.