How to Find and Close More Leads Using Apollo (2026)

Most Apollo accounts fail the same way. Someone signs up, runs a search that returns 40,000 people, exports the lot, drops them into one sequence and sends from the company domain. Three weeks later the reply rate is under half a percent, the domain is landing in spam, and the verdict is "the data is bad."

The data is rarely the problem. Apollo has 240M+ contacts and 30M+ companies behind 65+ search filters. What separates a working pipeline from a burned domain is how you narrow that database, how you spend credits, and what you do after someone replies. The steps below are in the order to do them.

Every benchmark here is Apollo's own published number, taken from its product, pricing and deliverability pages in September 2026. They describe Apollo's customers, not a promise for your list.

Start with the pipeline math, not the search bar

Before you touch a filter, work out how many contacts you need. Apollo's deliverability page publishes a 2.37% cold-to-meeting rate, alongside a 45% open rate on cold audiences. Plan with those until you have your own numbers.

Two things follow. First, a realistic outbound target needs hundreds of contacts a month, not tens of thousands, so you can afford to be very picky. Second, your close rate moves the number far more than your reply rate. A team closing 15% instead of 25% needs 70% more contacts for the same result, and no sequence fixes a bad fit.

Set up sending before you search

Everyone skips this step, and it is the only one that is hard to undo. A domain with a spam reputation takes months to recover. A bad list takes an afternoon to delete.

• Send from a separate domain. Buy a lookalike of your main domain and send from that. Apollo sells domains and mailboxes inside the product, which saves a trip to a registrar.
• Authenticate it. Set up SPF, DKIM and DMARC, plus a custom tracking subdomain. Apollo configures these automatically for domains bought through it, and its AI assistant flags authentication errors that block sending.
• Warm up, then stay under the cap. Apollo's guidance is to start around 20 emails a day per mailbox and never exceed 50, with warmup taking a few weeks. That cap is the real limit on your volume.
• Do the mailbox arithmetic. At 50 a day, one mailbox sends about 1,000 emails a month. A four-email sequence spends about four emails per contact, so one mailbox covers roughly 250 new contacts a month. Need 675? That's three mailboxes, and you set them up now, not after you've started sending.
• Leave the safety nets on. Bounce and unsubscribe suppression across sequences, catch-all detection, and autopause when bounce rates spike are all on by default. Keep them that way.

One plan detail bites people on day one. Every Apollo account can send email campaigns, but non-paying accounts can only connect Gmail mailboxes. Microsoft 365 and other providers need a paid plan.

Turn your ICP into filters

An ideal customer profile (ICP) in a slide deck is a wish. An ICP saved as an Apollo search is a list you can work. Write yours as a filter set, and make each filter earn its place: every filter shrinks the list, so add only the ones that change whether you'd take the meeting.

Build the account list first and the people list second. Filter companies down to the few hundred that really fit, save that as a list, then find people inside it. Searching people first gets you the right titles at the wrong companies, the most expensive kind of wrong.

Signals beat firmographics

Firmographics tell you who could buy. Signals tell you who might buy this quarter. Apollo's signal filters (job postings, funding, job changes, Bombora intent and website visits) turn a cold list into a warm one, and they cost nothing extra to apply.

Website visitor identification is the warmest list in the product, because these people already came to you. Company-level identification works worldwide and is on the free plan. Contact-level matching is US-only and needs the Inbound add-on. A visit can trigger a Slack alert, route to a rep, or enrol the contact in a sequence automatically.

Spend credits deliberately

People get cautious with credits in the wrong places. These are the rules that matter:

• Reveals cost 1 credit. Apollo charges no more than 1 credit per contact for an email address, and its pricing page lists the same cost for a verified phone number.
• Exports cost separately. Export credits go whenever a contact leaves Apollo: a CSV, a CRM sync, or Person API enrichment into another system.
• Credits expire. They end with each billing cycle and never roll over. On an annual plan the whole year's credits arrive at the start of the term, which makes it easy to under-use them for ten months and panic in month eleven.
• Trials are small. A trial includes 50 credits and 5 mobile credits: enough to test data quality on your own ICP, and nothing more.
• "Unlimited" has a ceiling. A fair use policy caps it at 10,000 credits per account per month for non-paying accounts, or, for paying ones, the lesser of ($ paid ÷ $0.025) or 1 million credits per account per year.

In practice: reveal emails for the whole target list, reveal phone numbers only for accounts that have engaged or that you've chosen to call, and export only what is going into a sequence or your CRM. A contact sitting in a spreadsheet nobody opens cost you two credits and returned nothing.

Apollo is also moving accounts from a legacy credit system to a newer one, so check the plan page inside your own account before you budget.

If the problem is your existing CRM records rather than new prospects, use waterfall enrichment. Apollo fills each record from its own data first, then passes the gaps to 20+ partner vendors in an order you set. Apollo says beta customers saw 5–40% more records enriched on external lists and a 45% lower email bounce rate.

Build the sequence around the account, not the send button

Apollo sequences mix automatic emails, manual emails, calls, LinkedIn steps and general tasks. For most B2B offers, the structure that works runs three weeks, with eight or nine steps across more than one channel.

Four rules keep it working:

• One sequence per persona. A VP of Engineering and a Head of Marketing don't share a first line. If you can't write a different opener for a segment, it isn't a segment.
• Test one variable at a time, and wait for a few hundred sends before you call it. Subject lines move the open rate; the first two sentences move the reply rate.
• Send in the prospect's business hours and throttle daily volume per mailbox. Four hundred emails at 3am from one mailbox is the fastest route to the spam folder.
• Let AI draft, not decide. Apollo's assistant writes from real contact and account data and rephrases or shortens in one click. Use it to polish a line you already wrote. A sequence written entirely by AI reads like every other AI sequence in that inbox.

What to write

Keep cold emails to 50–90 words, in plain text, with no images, no attachments, and at most one link after the first message. Personalise at the account level (a job posting, a funding round, a product launch, a technology they run) rather than the person level. "I saw you went to Michigan" is noise. "You're hiring three data engineers in Austin" is a reason to write.

Keep the ask small. A reply is a lower bar than a meeting, and "is this a problem worth 15 minutes?" beats a calendar link in a first message almost every time.

Call the accounts that engage

Email finds who's interested, and calls turn that into meetings. Dial the people who have opened repeatedly, clicked or visited your site, not the whole list.

Apollo's dialer includes parallel dialing (several numbers at once, connecting you to whoever answers first), a power dialer for prioritised lists, local presence numbers, voicemail drops, AI-written call scripts, automatic call logging and AI summaries sent to your CRM. Apollo reports that customers using these get 18% more meetings from cold calls and save about an hour per SDR per day.

Two cautions. Use parallel dialing on lists where you expect few pickups, because connecting late to a live person is worse than not calling. And local presence won't work forever: carriers are getting stricter about spam labels, so watch your connect rate over time instead of assuming it holds.

From reply to closed won

Teams put their effort into finding leads. Apollo accounts usually leak at closing them.

• Answer fast. Reply speed beats reply cleverness. Check the sequence inbox every morning and set alerts for replies from target accounts.
• Book in the thread. Apollo's meeting scheduler and routing put the booking link in the conversation instead of a follow-up email, which is where most meetings die.
• Open a deal as soon as a meeting is booked. Apollo Deals has stages, stalled-deal alerts and pipeline analytics. A reply that never becomes a deal record is a reply nobody follows up.
• Record the calls and use them. Conversation intelligence transcribes and summarises calls, pulls out objections and action items, and turns outcomes into CRM activities and follow-up drafts. Scorecards and keyword tracking let you coach from evidence instead of memory.
• Pick one system of record. Apollo syncs with Salesforce, HubSpot, Outreach, Salesloft and Marketo. Make one system authoritative for each object and sync in that direction. Two-way sync with no rules creates duplicates that quietly break your exclusion lists.

Measure the funnel weekly

Track the whole funnel, not just the top. Most teams watch open rate, which is now the least reliable number in the stack because privacy proxies pre-fetch images.

Read these by sequence and by segment, never in aggregate. One persona at a 6% reply rate and another at 0.5% average out to a number that looks fine and tells you nothing.

Let an AI assistant build the list, not send the email

Apollo has an MCP connector, so an AI assistant can run most of the prospecting loop directly. It can search for contacts and sort them into lists, enrich records with verified emails and phone numbers, build sequences, manage sending domains, pull open, click and reply analytics, and create or prioritise tasks.

It works with Claude, ChatGPT and Perplexity, signs in over OAuth with no API key, and runs on any plan, including free, at no extra cost. Actions follow your account's permissions and limits, so an assistant can't spend credits you don't have.

Use it for research and assembly. Have the assistant draft the ICP filters, pull the list, enrich it and prepare the sequence, then read every message yourself before one goes out. An assistant that can send email is one bad instruction away from a domain you have to replace.

Cold email rules you can't skip

Cold outreach is legal in most markets and comes with conditions everywhere. The basics: identify yourself and your company honestly, include a real postal address, make opting out one click, honour it at once across every sequence, and keep a suppression list that outlives any single campaign. Apollo is GDPR-compliant as a platform and provides the suppression tools, but whether a given send is lawful depends on where the recipient is and what you're offering. Spend twenty minutes with someone qualified before your first European campaign, not after it.

The five mistakes that cost the most

1. Sending from the primary domain. One bad month and your invoices land in spam too.
2. Exporting everything. Exports cost credits, and a list you'll never work is pure cost.
3. One sequence for every persona. Generic copy does more damage to reply rates than anything else.
4. Ignoring bounce rate until it hits 8%. By then the damage to your domain's reputation is done.
5. Treating "booked" as the finish line. Held, qualified and closed are three more numbers, and each one leaks.

A 30-day rollout

By day 30, aim for one sequence that works, not five that are half-tested. The second persona is much easier once the first has real numbers behind it.