Apollo Prospecting

Prospect in Apollo.io

This is the tool-specific execution layer for the B2B Prospecting Engine pack. The methodology skills decide what to do; this skill shows you how to do it in Apollo.io. It assumes you have already built your ICP and personas in [[icp-persona-builder]], decided your list strategy in [[prospect-list-builder]], chosen your enrichment and verification rules in [[lead-enrichment]], picked the signals you care about in [[buying-signal-tracker]], structured your cadence in [[outreach-sequence-designer]], hardened your sending infrastructure in [[cold-email-deliverability]], and defined your funnel math in [[prospecting-metrics]].

If you have not done that work, stop and do it first. Apollo will happily let you blast a poorly-targeted list from your primary domain and torch both your deliverability and your data budget. The tool is a power amplifier, not a strategy.

Apollo changes its UI often, so this skill stays conceptual on where things live and concrete on what to do and why. When a specific menu or label is uncertain, confirm it in-product rather than trusting a click path from memory.

When to use this skill

• You have an approved ICP and you need to translate it into Apollo People/Company Search filters.
• You are building or tiering lists in Apollo and need to suppress existing customers and open opportunities.
• You need to find and verify emails in Apollo without burning credits on garbage.
• You want Apollo to surface buying signals (job changes, hiring, funding, intent) and alert you.
• You are building an outbound cadence as an Apollo Sequence and wiring in LinkedIn/call task steps.

Do not use this skill to invent the strategy - that lives in the methodology siblings above.

The workflow

1. Translate your ICP into Apollo filters ([[icp-persona-builder]] → People/Company Search)

Build the company layer first, then the person layer on top. In Apollo's People Search and Company Search you stack filters; each filter narrows the set, so add them deliberately and watch the result count. A useful sanity band: a persona-per-tier search that returns more than ~5,000 people is almost certainly under-filtered, and one returning under ~100 may be over-constrained for a sustained sequence.

• Firmographic: industry/keywords, employee headcount band, revenue (where available), HQ geography. Map these straight from your ICP's account definition.
• Technographic: "uses technology X" filters - powerful for ICPs defined by a tech stack (e.g. targets running a specific CRM, cloud, or e-commerce platform).
• Persona (person layer): title keywords, seniority, and department/function. Prefer seniority + department over raw title strings, because titles are inconsistent across companies; layer specific title keywords only to sharpen.
• Exclusions: exclude the industries, sizes, and titles your ICP explicitly rules out. Exclusions are as important as inclusions for keeping credits honest.

Save the result as a Saved Search so it is reproducible and can drive alerts. One saved search per persona-per-tier, not one mega-search.

2. Build and tier lists, then suppress ([[prospect-list-builder]] → Apollo Lists)

Promote saved-search results into Lists that mirror your tiers (e.g. Tier A - VP Eng - fintech, Tier B - …). Keep lists small and named by ICP slice so you can run different sequences and read metrics per slice.

Suppression is non-negotiable. Before anyone enters a sequence:

• Exclude contacts already in your CRM (current customers, open opportunities, recent conversations). Use Apollo's CRM integration/enrichment so existing records are flagged, or export and dedupe against a suppression list.
• Exclude do-not-contact and previously-bounced addresses.

A prospect who is already a customer or an open opp landing in a cold sequence is an own-goal that the rep and the data both pay for.

3. Enrich and verify with credit discipline ([[lead-enrichment]] → email finder + verification)

Apollo spends credits to reveal/verify contact data and to export. Treat credits as cash:

• Reveal/verify emails only for contacts that survived filtering and suppression. Never reveal an entire broad search.
• Respect the email verification state. Send to verified/valid addresses; hold or route catch-all/unknown/guessed states differently (lower volume, or skip). Sending to unverified addresses is how you manufacture bounces - and a bounce rate above roughly 2% is the red line where mailbox providers start throttling you; past 5% you are in serious reputation damage.
• Pull "Net New" leads (not already in your data) deliberately, in batches sized to what you can actually work.
• Decide your CRM sync path up front: CSV export for one-offs, or the Apollo API / native CRM integration for repeatable syncing. Keep the field mapping consistent so downstream reporting in [[prospecting-metrics]] holds together.

4. Track buying signals and wire alerts ([[buying-signal-tracker]] → filters + saved-search alerts)

Encode the signals you chose into Apollo where the data exists:

• Job changes - your champion moved to a new account; that is a warm opening.
• Hiring - open roles in the relevant function signal budget and pain.
• Funding / growth - new capital often unlocks new initiatives.
• Intent topics - where available, filter or prioritize by accounts researching your category.

Turn the highest-value saved searches into alerts so new matches surface automatically instead of you re-running searches. Route signal hits into a dedicated, higher-priority sequence.

5. Build the cadence as an Apollo Sequence ([[outreach-sequence-designer]] → Sequences)

Build the cadence you designed as an Apollo Sequence - multi-step, mixing automated email steps with manual tasks (LinkedIn touch, call). Do not let Apollo auto-send everything; manual task steps are where reps add the human judgment that earns replies.

• Write the email copy in [[cold-email-craft]] - this skill only places it into steps.
• Set sane delays between steps and cap daily send volume per mailbox (see step 6).
• A/B test at the step level (subject or first line), and read results against [[prospecting-metrics]] - not vanity opens. For calibration: a healthy cold sequence typically lands a 1-5% reply rate; if you are under 1% after ~200 sends, stop and fix targeting or copy before spending more list.

6. Protect deliverability before you press start ([[cold-email-deliverability]])

Apollo sends through mailboxes you connect - it does not fix bad DNS. Before any sequence goes live:

• Connect dedicated sending mailboxes on a separate sending domain, not your primary corporate domain. Cold-sequence reputation damage should never touch the domain your company runs its real email on.
• Make sure those domains/mailboxes are authenticated (SPF/DKIM/DMARC) and warmed - do this in [[cold-email-deliverability]], not here. Budget 2-4 weeks of warmup for a fresh mailbox before it carries cold volume.
• Respect conservative per-mailbox daily limits - roughly 20-50 cold sends per mailbox per day is the practitioner range; scale by adding mailboxes, not by pushing one past it - and let Apollo throttle.

Apollo filter recipe (ICP → stacked filters)

ICP: Mid-market fintech, Series B+, running a modern data warehouse,
     buyer = data/analytics leadership.

COMPANY LAYER (Company Search)
  Industry/keywords ........ financial services, fintech, payments
  Employee headcount ....... 201-1000
  Geography (HQ) ........... United States, Canada
  Technographic ............ uses Snowflake OR BigQuery
  Exclude .................. industry = staffing, education
                            headcount < 50
  → Save as: "Co - MM fintech - modern DWH"

PERSON LAYER (People Search, on top)
  Seniority ................ VP, Head, Director
  Department/function ...... Data / Analytics / Engineering
  Title keywords ........... "data", "analytics", "platform"
  Exclude titles ........... intern, contractor, "sales"
  → Save as: "P - Tier A - Data leaders - MM fintech"

SIGNAL OVERLAY (separate saved search + alert)
  Same as above, plus: hiring for data roles  OR  funding in last 90d
  → Save as: "P - Tier A - Data leaders - SIGNAL" (alert ON)

Apollo Sequence outline (cadence skeleton)

Sequence: "Tier A - Data leaders - MM fintech"
Mailboxes: 2 dedicated boxes on outbound.example-go.com (warmed, authed)
Daily cap: conservative per mailbox (~20-50 cold sends); split across both

Day 1  · Auto email   · Step A1 - problem-led opener (copy from cold-email-craft)
Day 2  · Manual task  · LinkedIn - view + connect, no pitch
Day 4  · Auto email   · Step A2 - reply-thread bump, new angle/proof
Day 6  · Manual task  · Call - reference the trigger/signal
Day 9  · Auto email   · Step A3 - short, specific CTA
Day 13 · Manual task  · LinkedIn - soft value share
Day 16 · Auto email   · Step A4 - breakup / permission-to-close

A/B: test Step A1 subject line only; hold everything else constant.
Read: reply rate + positive-reply rate per step (NOT open rate).

Worked example (end to end, one ICP)

ICP (from [[icp-persona-builder]]): data/analytics leaders at US/Canada mid-market fintechs (201-1000) running Snowflake or BigQuery.

1. Filters. In Company Search I stack the firmographic + technographic filters above and exclude staffing/education and sub-50 headcount. ~600 accounts. I save it. On the person layer I add seniority VP/Head/Director + Data/Analytics/Engineering function, exclude sales/contractor titles. ~900 people. Saved as the Tier A person search.
2. List + suppress. I promote the results into a Tier A - Data leaders - MM fintech list. Using the CRM integration, I drop everyone already a customer or in an open opp, plus prior bounces and do-not-contact. ~720 remain.
3. Enrich + verify. I reveal/verify emails only for that suppressed list. I keep verified/valid, route catch-all to a lower-volume track, and skip unknown/guessed. ~540 send-ready. I export the verified set to the CRM via the API with consistent field mapping.
4. Signals. I clone the search into a SIGNAL variant (hiring for data roles OR funding in last 90d), turn the alert on, and these hits feed a higher-priority sequence.
5. Sequence. I build the 7-step cadence above across two warmed mailboxes on a dedicated outbound domain, paste copy from [[cold-email-craft]], set conservative caps, and A/B the opener subject.
6. Read. A week in, I judge by reply and positive-reply rate per step in [[prospecting-metrics]] - and ignore open rate entirely (see below).

Deliverable

Produce a working Apollo setup plus a one-page run sheet documenting it: the saved searches (one per persona-per-tier, plus signal variants with alerts on), the tiered and suppressed Lists with their counts at each stage (raw → suppressed → verified send-ready), the verification routing rule (verified send / catch-all low-volume / unknown skip), the CRM sync path and field mapping, and the live Sequence outline with mailboxes, daily caps, step schedule, and the single A/B variable. The run sheet is what lets a second rep - or you in a month - reproduce and audit the setup.

Quality bar

• Every saved search maps to exactly one persona-per-tier from the ICP; none is an untierable mega-search.
• No contact enters a sequence without passing CRM suppression and holding a verified email state.
• All sending mailboxes are on a dedicated outbound domain, authenticated and warmed, with daily caps inside the 20-50 range.
• The sequence mixes auto emails with manual LinkedIn/call tasks; success is read in replies and positive replies, never opens.

Common failure modes

• Blasting sequences from your primary domain. The fastest way to poison the email your whole company depends on. Always send from dedicated, authenticated, warmed domains/mailboxes. Apollo connects the mailbox; it does not fix your DNS - that is [[cold-email-deliverability]].
• Ignoring verification states. Treating catch-all/unknown/guessed like verified manufactures bounces, which wrecks reputation and corrupts your metrics. Send to verified; route or skip the rest. Watch the 2% bounce red line.
• Over-broad filters that burn credits. Revealing a 50k-row search "to see who's there" spends your data budget on contacts you will never work. Filter and suppress before you reveal/export; reveal only what you can actually sequence.
• Not suppressing customers and open opps. Cold-emailing a current customer or an active deal embarrasses the rep and pollutes reporting. Suppress against the CRM every time, before sequencing.
• Relying on open rate. Apple Mail Privacy Protection and image proxies inflate opens into noise. Optimize for replies, positive replies, and meetings - the funnel math in [[prospecting-metrics]] - not opens.
• One mega saved search. A single giant search you can't tier or alert on becomes unmaintainable. Keep one saved search per persona-per-tier, plus signal variants.
• Letting Apollo auto-send every step. Removing the manual LinkedIn/call tasks turns a thoughtful cadence into spam. Keep human task steps in the sequence.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: SkillMedev
• Source: SkillMedev/b2b-prospecting-engine
• License: MIT
• Homepage: https://skillme.dev/pack/b2b-prospecting-engine

Install and usage instructions live in the source repository linked above.