Cold Email Targeting: How to Find the Right People and Verify They're Real

Bad targeting costs more than bad copy. If you have the right message but send it to the wrong person, the best case is silence. The worst case is a spam report from someone who has never heard of you, who did not ask to be contacted, and who has no use for what you are selling. Repeat that enough times and your sending domain is cooked.

Cold email programs that underperform usually trace back to one of two problems: targeting that was too broad at the start, or data that was too dirty to send confidently. Getting both right is not complicated, but it requires more deliberate thinking than most teams put in before they start building lists.

How to define your ICP before you pull a single lead

An ideal customer profile is not a vague sketch of the type of company you want. It is a set of specific, filterable criteria you can punch into a lead tool and get back a list of real people. The more precisely you can define it, the cleaner your targeting will be and the more relevant your copy will read.

The most useful ICP dimensions for cold email are:

• Job title and function: What role is responsible for the problem you solve? Head of Sales, VP of Operations, Founder, CTO. These should be specific, not "senior decision maker."
• Seniority: Are you looking for the person who writes the check, the person who owns the process, or both? For enterprise deals, you often need both, and they require different messaging.
• Company size: Headcount and revenue range. A 15-person agency and a 500-person agency have entirely different buying processes, budgets, and pain points, even if their industry is the same.
• Vertical: Not just "tech" or "professional services" but specific sub-verticals where your product creates the most value. This shapes the specific language you use in copy.
• Geography: Particularly relevant for deals that involve local services, compliance requirements, or time zone considerations.
• Tech stack: What tools is the company already using? This can indicate budget, sophistication, and whether they are likely to already have a solution. Tools like Clay and Apollo allow tech stack filtering.

The best way to validate your ICP before running a campaign is to look at your five best existing customers and identify what they have in common. Those shared attributes are your highest-confidence targeting layer.

Where lead data comes from and how reliable it is

Every lead tool is making an educated guess about email addresses. There is no authoritative master list of everyone's professional email. Tools infer addresses from scraped profiles, confirmed sends, partner data, and pattern matching against known email formats. This is why accuracy varies significantly across data sources.

Apollo.io is the most widely used database for small to mid-market outreach. Its contact coverage is strong for US-based companies in tech and professional services. Accuracy tends to drop for senior titles at large enterprises, international contacts, and companies that use custom domains. Clay is not a database itself but an orchestration layer: it pulls from Apollo, LinkedIn, Clearbit, Hunter, and other sources simultaneously, then lets you run enrichment logic on top. It is the most flexible option for teams that need to work across multiple data sources. LinkedIn Sales Navigator has the most accurate profile data because people update their own profiles, but extracting emails requires third-party scrapers, which adds a step. ZoomInfo is expensive and justified primarily when you need deep account data, org charts, or buying intent signals that cheaper tools do not carry.

Email verification: what bounce rates actually do to your domain

Every email that bounces sends a signal to inbox providers that your sending behavior looks suspicious. A bounce rate above 5% starts degrading your domain reputation. Above 8 to 10%, you are likely triggering spam filters across a significant portion of your sends, meaning even valid contacts are not seeing your emails. At 15% or more you are on a path to blacklisting.

The solution is to verify every list before you load it into your sending platform. Verification tools like NeverBounce, ZeroBounce, and Millionverifier check whether an email address has a valid MX record, whether the mailbox exists, and whether the address is associated with known spam traps or disposable email services. They do not guarantee deliverability, but they remove the addresses most likely to hard bounce.

Target a bounce rate below 3% on every campaign. If you are consistently hitting 5% or above after verification, the underlying data source is too inaccurate and needs to be replaced or supplemented.

What happens when you email the wrong person

Sending to a poorly defined list does more than reduce your reply rate. It actively damages your program in ways that compound over time. A recipient who has no use for your offer and did not expect to hear from you is far more likely to mark you as spam than someone who is adjacent to your ICP. Spam reports are weighted heavily by Gmail and Microsoft when calculating sender reputation scores.

Negative replies are a separate problem. A VP who replies to tell you that you have the wrong person, or that your targeting is embarrassingly off, is giving you feedback that others are simply ignoring while quietly hitting spam. Each spam report is estimated to require roughly 100 positive engagements to offset. Getting the targeting right from the start is far cheaper than recovering from domain damage after the fact.

How to maintain personalisation at scale

Personalisation at scale is not writing individual emails to every prospect. It is using structured research signals to create copy that reads as specific even when it is templated. The approach that works best is tiered personalisation: more depth for fewer accounts, lighter touch for broader volume.

Tier 1: Deep personalisation for top accounts

For your 20 to 50 highest-value target accounts, it is worth doing actual research. Look at recent company news, hiring activity, product launches, LinkedIn posts from the prospect, or job descriptions that signal an active pain. Clay can automate much of this research gathering, but a human should be reviewing and selecting which signals make it into the copy. A first line that references something specific and recent reads completely differently from a generic opening, and reply rates on this tier are typically 2x to 3x the campaign average.

Tier 2: Variable-based personalisation for volume

For the broader list, personalisation means using structured variables: company name, job title, vertical-specific language, and a first line that speaks to the category of person rather than the individual. The copy should still read as though you knew who you were writing to, even if the specifics come from a data field. The key is using variables that are actually accurate, not pulling in stale or mismatched data that makes the personalisation obvious and clumsy.

Targeting precision vs volume: when each matters

There is a real trade-off between how tightly you target and how many people you can reach. A highly specific ICP might produce a list of 500 qualified prospects. A broader ICP might produce 10,000. Both can be valid depending on your deal size and sales capacity.

For high-ticket deals above $20,000 ACV, precision almost always wins. You need reply rates high enough to generate conversations, not just impressions, and your sales team cannot handle thousands of unqualified replies. For lower-ticket products where sales is largely self-serve, volume with lighter personalisation often produces better pipeline numbers than a small, highly curated list. The mistake is applying the wrong model to the wrong deal size, usually by running a high-volume approach on a product that requires a consultative sale.