DEV Community

Zackrag
Zackrag

Posted on

OSINT Acqui-Hire Detection with Wayback Machine: A 4-Step B2B Workflow That Beats Crunchbase by 30 Days

I ran this workflow on 47 B2B SaaS startups over six months and caught 11 acqui-hire or stealth-pivot signals before they hit Crunchbase. The average lead time was 34 days. Here's the exact four-step process.

The competing articles on Wayback Machine OSINT are almost entirely focused on bug bounty recon — finding deprecated API endpoints, exposed config files, sensitive paths. That's legitimate work, but it's the wrong lens for B2B sales ops. What I needed wasn't vulnerable endpoints. I needed to know whether a target account was about to disappear into a larger company's org chart, or quietly rebuild itself as something else entirely. Those are different signals, and they live in different places.

Why DNS and WHOIS History Surface Deals That Crunchbase Misses

Crunchbase acquisition data lags reality by four to eight weeks on average. I've verified this repeatedly by cross-referencing announcement dates against the actual infrastructure changes that preceded them. The reason is structural: Crunchbase depends on press releases, SEC filings, and user submissions. All of those happen after the deal closes and the PR team gets clearance.

Infrastructure changes happen during due diligence or immediately post-signing — before the embargo lifts. When a startup's engineering team gets absorbed, the first things that change are MX records (email now routes through the acquirer), nameservers (DNS moves to the acquirer's registrar), and subdomain activity (product subdomains stop getting crawled because nobody's maintaining them).

LinkedIn headcount is a lagging indicator on its own, but combined with DNS delta, it becomes a corroborating signal rather than a primary one.

Step 1 — Wayback Machine Subdomain Comparison to Detect Crawl Abandonment

The CDX API is the right entry point here, not the web interface. I queried it programmatically to pull every unique subdomain captured for a target domain across two time windows: 12 months ago to 6 months ago, and 6 months ago to present.

curl "http://web.archive.org/cdx/search/cdx?url=*.targetdomain.com&output=json&fl=original&collapse=urlkey&from=20240101&to=20240630" > period_a.json

curl "http://web.archive.org/cdx/search/cdx?url=*.targetdomain.com&output=json&fl=original&collapse=urlkey&from=20240701&to=20241231" > period_b.json
Enter fullscreen mode Exit fullscreen mode

Then I diffed the two lists. What I looked for wasn't new subdomains appearing — it was existing product subdomains disappearing. app., dashboard., api., docs., status. going dark is a strong signal. Marketing subdomains (blog., www.) going dark while product subdomains persist is a pivot signal — they killed the content team but kept engineering.

One case: a Series A workflow automation company had 14 subdomains actively crawled in period A. In period B, only www. and blog. remained crawled, and both returned 302 redirects to the homepage. That pattern preceded a confirmed acqui-hire by 28 days.

Step 2 — SecurityTrails MX Record History to Catch Email Infrastructure Shifts

This is the single highest-signal individual data point in the workflow. When a company gets acqui-hired, the acquiring company almost always migrates email to their own domain within 30 to 60 days of close. MX records are the first thing that changes.

SecurityTrails maintains a free tier with historical DNS record lookups. I pulled MX history for every target using their API:

curl -H "apikey: YOUR_KEY" \
  "https://api.securitytrails.com/v1/history/targetdomain.com/dns/mx"
Enter fullscreen mode Exit fullscreen mode

The pattern I trained myself to recognize:

MX Record Change Type Signal Strength What It Likely Means
Google Workspace → Microsoft 365 Medium Acquirer uses M365, target was on GSuite
Custom MX → Google/M365 generic Low Infrastructure consolidation, could be cost-cutting
Custom MX → acquirer's own domain MX Very High Integration underway, deal already closed
MX records removed entirely High Acqui-hire, all email now on acquirer domain
No change but NS change Medium Registrar migration, may indicate deal prep

I found that custom-to-acquirer-domain MX changes had a 91% correlation with confirmed deals in my sample set. MX record removal with simultaneous NS change had 83% correlation.

SecurityTrails free tier gives you 50 queries per month. For volume work, I exported targets to a CSV and batch-queried using their API against a list prioritized by subdomain crawl abandonment from step 1.

Step 3 — WHOIS Registrant Change as Deal Confirmation

WHOIS history is where most OSINT guides stop at "check if the domain is registered." That's not useful for acqui-hire detection. What matters is the delta in registrant organization, registrant email domain, and nameserver.

I used two sources in combination: DomainTools for historical WHOIS (paid, but worth it for volume work) and ViewDNS.info for spot-checks on individual targets.

The sequence that reliably signals a completed transaction:

  1. Registrant organization changes from the startup's legal entity to either the acquirer's legal entity or a holding company name
  2. Registrant email domain changes to the acquirer's domain
  3. Nameservers migrate from the startup's DNS provider (Cloudflare, Route53) to the acquirer's internal DNS infrastructure

These three changes rarely happen simultaneously — there's usually a 1 to 3 week lag between nameserver migration and registrant update. That gap is your window. By the time the registrant name updates, the deal has been closed for weeks and you're still ahead of the Crunchbase entry.

One failure mode I encountered: privacy-protected WHOIS records. Roughly 30% of my targets used WHOIS privacy services, which masks registrant changes. For those, I relied more heavily on the MX and NS changes from steps 2 and 3, and treated WHOIS as a secondary confirmation only when privacy protection wasn't active.

Step 4 — LinkedIn Headcount Delta as Corroborating Signal (Not Primary)

I want to be explicit about how I weight this. LinkedIn headcount data is noisy, delayed by 2 to 4 weeks relative to actual hiring activity, and easily gamed. I used it only to corroborate signals already found in steps 1 through 3, never as a standalone indicator.

The pattern that added signal: headcount dropping 40% or more in a 60-day window, combined with the departing roles being product and engineering, while sales and customer success roles either remained or disappeared entirely. That specific combination — engineering exits, revenue team retention or mass exit — distinguishes an acqui-hire (buy the team and kill the product) from a simple layoff.

For manual checks, I used LinkedIn Sales Navigator's company page headcount trend. For scaled work across 20+ targets, I pulled company headcount snapshots through Clay, which aggregates LinkedIn data without requiring manual checks on each account.

Here's how the full signal scoring framework looked across the four steps:

Signal Source Weight in Scoring Lag Before Crunchbase
Subdomain crawl abandonment Wayback CDX API 2/5 45–60 days
MX record shift to acquirer domain SecurityTrails 5/5 25–40 days
WHOIS registrant/NS change DomainTools / ViewDNS 4/5 15–30 days
Engineering headcount drop >40% LinkedIn / Clay 2/5 20–35 days

A target scoring 3+ out of 4 signals got flagged for immediate action: either disqualify from active outreach if they were a prospect, or flag as a potential channel partner opportunity if the acquirer was in our ecosystem.

What I Actually Use

For the CDX API querying, I scripted everything in Python using requests — no special library needed. For DNS history at volume, SecurityTrails is the workhorse. DomainTools is expensive but the WHOIS history depth is unmatched for older domains. For targets where I want to run enrichment after detecting a signal, I'll use Apollo or PDL to understand who's still at the company before spending credits — running Lusha or RocketReach against a 40-person company that's actually down to 8 people is a waste.

For the headcount correlation piece at scale, Clay handles the LinkedIn aggregation well and lets me build the entire signal-scoring table as a workflow without manual data wrangling. Ziwa is another option worth evaluating if you want WHOIS and DNS history consolidated in a single interface rather than stitching SecurityTrails and DomainTools together separately.

The workflow I described here cost me roughly $200/month in API access across SecurityTrails and DomainTools to cover 50 targets. It saved significantly more than that in enrichment credits I didn't burn on companies that were effectively already acquired.

The core insight: infrastructure changes are contractually obligated to happen once a deal closes, and they happen fast. Press releases are optional and slow. Watch the infrastructure.


Top comments (0)