92% accurate.
On the list
nobody could buy.
Apollo and ZoomInfo sell phonebooks. We build the buyer set scraped from seven sources, layered with tech stack and intent, scored by ICP-fit, and hand-verified before it ships. Same total market, 2-5× the reply rate, because the timing matches the pitch.
Most lead vendors sell the same data to your competitors. The work isn’t having a name and an email, it’s having the timing. Trigger events, intent surges, hiring shifts, and stack changes are what convert; the rest is wallpaper. We just decided to build the list that converts.
Apollo sells you a phonebook.
We build the buyer set.
That difference is the entire pitch.
Intel. Not noise.
An observatory desk, a leather lead dossier laid open, then a forensic stack of firmographic, tech-stack, hiring-signal, intent, and verified-contact tiles, threaded by lime.
Four benches. One scored list.
Define, scrape, enrich, verify. Each bench owns one part of the list, run as a discipline rather than a tool license. The accuracy floor is the discipline of the fourth bench, the one most vendors skip.
the.observatory- Bench 01
Define the ICP
Most lead lists fail before scraping starts, the ICP was vague. We pin it: revenue band, headcount band, hiring posture, tech stack, expansion stage, and the trigger events that say 'they have the problem you solve right now'. Every cohort gets a written brief signed off before a single source is queried.
- Bench 02
Scrape the source
Apollo, Crunchbase, BuiltWith, LinkedIn Sales Nav, public hiring boards, news feeds, funding databases, GitHub, and proprietary triggering-event scrapers, pulled in parallel and reconciled to a single canonical record per company. No vendor lock-in, no single-source bias.
- Bench 03
Enrich the signal
Firmographic + technographic + intent. We layer: tech stack, recent funding, hiring velocity, exec changes, public pain signals, intent data (Bombora / 6sense), and the bespoke trigger feeds we keep for our own clients. The list goes from 'companies' to 'companies with the active problem'.
- Bench 04
Verify by hand
Algorithms produce 70% accuracy on a good day. We sample-verify every list at the human bench: 50-100 records pulled at random, manually checked against LinkedIn + the company site, with the failure rate held under 8%. If a list misses its accuracy floor, we don't ship it, we re-scrape.
Six layers. One verified buyer.
Every row in the list is a stacked dossier, six discrete signal layers, each from a different source, each with its own decay rate and its own QA path. That layering is the difference between “a name on a sheet” and “a buyer with the active problem”.
- /01
Firmographic core
Revenue band · headcount · industry · location · founded year.
- /02
Technographic stack
BuiltWith / Wappalyzer · current tools · stack-fit signals.
- /03
Hiring & growth
Open roles · hiring velocity · exec changes · expansion signals.
- /04
Trigger events
Funding · M&A · leadership shifts · product launches · public pain.
- /05
Intent layer
Bombora · 6sense · keyword surge · category research signals.
- /06
Verified contact
Decision-maker named · email validated · LinkedIn confirmed live.
qa · 5 checksFive checks. Zero phonebook.
Every list clears five checks before it leaves the bench. Source-dedup, email validation, sanity check, ICP-fit scoring, and a manual 50-100-row sample audit. If accuracy misses the 92% floor, the list goes back to scrape, never shipped at 87% with an apology.
- /01
Source dedup
Records reconciled across Apollo / Crunchbase / Sales Nav. One canonical row per company, no source-overlap noise.
- /02
Email verification
Hunter + ZeroBounce + SMTP-ping triple-check. Catch-all and role addresses pulled into a separate bucket, never shipped as primary.
- /03
Sanity check
Revenue / headcount / industry sanity-checked against company site. Wikipedia-aged data and ghost-company entries flagged and dropped.
- /04
Score & rank
Each row gets an ICP-fit score (0-100) blending firmographic match, trigger weight, and intent signal. Top quartile flagged for first-priority outreach.
- /05
Sample audit
50-100 random rows manually checked against LinkedIn + company site. If accuracy < 92%, the list goes back to the bench.
Five days. One scored cohort.
- MON
Brief
ICP locked, trigger filters set, source list scoped, accuracy floor agreed in writing.
- TUE
Scrape
Multi-source pull runs in parallel, Apollo, Crunchbase, Sales Nav, BuiltWith, news feeds, hiring boards.
- WED
Enrich
Tech stack, intent, hiring velocity layered. Trigger events tagged. Records reconciled to one row per company.
- THU
QA bench
Email validation, sanity-check, ICP scoring, manual sample audit. Failures sent back for re-scrape.
- FRI
Handoff
Cleaned list pushed to your CRM + outbound tool. Per-row source tag, score, and trigger reason logged for the call notes.
Anyone can buy a list. We build the dossier, seven sources, six layers, one human signing off the row.
Source reconciliation, intent overlay, manual sample audit. The 70% nobody pays for is the 70% that decides whether the next call lands.
What the lab actually returns.
- /0192%+verified data accuracy floor, held by manual sample audit
- /02<8%ceiling on bounce + invalid records, after triple-source check
- /032-5×conversion lift vs. raw Apollo / ZoomInfo on identical motion
- /04Weeklyrefresh cycle on live trigger events, never a frozen snapshot
- /050-100ICP-fit score per row, firmo + trigger + intent blended
- /06500+minimum list size, scaling to 10k+ on the same accuracy floor
Every list, scoped & scored.
Fixed scope per list. Everything below, every cohort, with the bench discipline that lets a 1,000-row list convert at 2-5× the rate of the same rows out of an off-the-shelf tool.
- 01
ICP + trigger-event definition
A written brief, revenue, headcount, tech, hiring posture, and trigger filters, signed off before any scrape.
- 02
Multi-source scraping pipeline
Apollo, Crunchbase, Sales Nav, BuiltWith, hiring boards, news feeds, pulled in parallel, reconciled.
- 03
Firmographic + technographic enrichment
Industry, revenue, headcount, tech stack, founding stage, layered onto every row.
- 04
Trigger + intent overlay
Funding, M&A, hiring velocity, Bombora / 6sense intent surge, tagged per record.
- 05
Email verification + role split
Hunter + ZeroBounce + SMTP triple-check. Role inboxes separated, never sold as primary.
- 06
ICP-fit scoring (0-100)
Per-row score blending firmo match, trigger weight, and intent signal, first-priority quartile flagged.
- 07
CRM + outbound-tool handoff
Direct push to HubSpot / Pipedrive / Apollo / Smartlead, source-tagged, score-logged, ready to send.
Seven sources. One canonical row.
A real lead-intelligence stack is seven sources reconciled, not a single vendor login. The reconciliation is where the conversion lift lives, and that’s the part nobody wants to operate.
- FirmographicApollo · Crunchbase · ZoomInfo · OpenCorporates
- TechnographicBuiltWith · Wappalyzer · public stack signals
- Hiring & growthLinkedIn Sales Nav · Greenhouse · public boards
- Trigger feedsFunding · M&A · exec moves · product launches
- Intent layerBombora · 6sense · keyword surge feeds
- VerificationHunter · ZeroBounce · SMTP-ping · manual audit
What buyers ask on the second call.
- 01
Why not just use Apollo?
- Apollo is a phonebook. It tells you who exists; it doesn't tell you who has the problem you solve right now. We use Apollo as one source out of seven, and the work is in the layering, adding tech stack, hiring velocity, trigger events, and intent on top, then manually verifying the result. The same 1,000 rows out of Apollo and out of our pipeline behave 2-5× differently in market.
- 02
What's the actual accuracy floor?
- 92% verified, sampled at 50-100 random rows per list against LinkedIn + the company site. If a list misses, it goes back to the bench. We don't ship a list at 87% with a discount note, we re-scrape. That accuracy floor is the difference between a 4% reply rate and a 1% reply rate downstream, and we'd rather miss a Friday than miss the floor.
- 03
How do trigger events change the math?
- A 'mid-market RevOps lead' is 50,000 people. A 'mid-market RevOps lead at a company that hired 3+ ops roles in the last 30 days and just bought a competitor' is 80 people. Same total addressable market, but the second list converts 3-5× higher because the timing matches the pitch. Trigger events are the entire reason cold lists feel different than warm ones.
- 04
Is this GDPR / CCPA compliant?
- Yes, we scrape only public business-context data (firmographic, technographic, public profiles, public hiring), never personal-context data, and we run a documented LIA (legitimate-interest assessment) per cohort. EU contacts are routed through a B2B-specific opt-in flow before outreach. We've turned down work before that didn't pass; we'll turn it down again.
- 05
Can you build a list against a competitor's customer base?
- Often, depending on the competitor and how their customer signals show up publicly. BuiltWith fingerprints, public case-study mentions, integration directories, job postings that name the tool, and conference speaker lists are all fair game. We don't touch private customer lists or anything obtained by trickery; we'll be straight about which targets are reachable this way and which aren't.
Define the ICP.
Scrape the source.
Verify the signal.
Day 1: ICP brief locked. Day 3: multi-source pull complete. Day 5: enrichment + scoring done. Day 7: manual sample audit signed and the list is in your CRM.
Pairs well with.
We’re direct about how we work.
Still something missing? Email hello@markingo.io. You’ll hear back within a business day.
Somewhere sharper. Think of us as your embedded growth team. You get the senior velocity of a well-run in-house function, without having to hire 9 specialists. We live in your Slack, your Linear, your calendar.
Want a scored, enriched list against a tight ICP?
Send the ICP brief. Sample of 50 verified, manually QA'd rows back inside 72 hours.
Ready to compound?
A 30-minute intro. No deck. We’ll ask three questions, diagnose the biggest growth lever on your desk, and tell you if we’re the right people to run it.