TLDR
When I audited one client’s outreach list of 12,181 contacts, I removed 2,920 of them before a single message went out. That is about one in four records that did not match the buyer, could not be reached, or would have hurt the sending domains. The list was not the problem. The missing ideal customer profile behind it was.
The direct answer: an ideal customer profile for outbound is a written definition of the companies you should contact, built from the deals you actually won, with clear exclusions. Then you tier the accounts that fit by buying signal, with an active job posting in the role you support as Tier 1, and you record each account’s best channel and email provider so every channel can use the same list.
An ideal customer profile is the first thing I build for every outbound program, and it is the step most teams skip or write as a slogan. Most guides on the topic stop at firmographics: industry, size, location, revenue. That is enough for a marketing persona deck. It is not enough to decide who gets a message on Monday.
This guide covers what an outbound ICP needs that a marketing ICP doesn’t, how I tier accounts, and the fields I put on every record.
What is an ideal customer profile?
An ideal customer profile describes the kind of company that gets the most value from what you sell, buys at a healthy price, and stays. It describes companies. A buyer persona describes a person inside one of those companies (the title, the worries, the language). You need both, but the ICP comes first, because it decides which companies are on the list at all.
The common steps are sound: gather clean account data, define what a good customer means for you (retention, deal size, speed to close), pull the customers who meet that bar, and look for the traits they share (Clearbit). Where those guides stop is where outbound starts.
Why does an outbound ICP need more than firmographics?
An outbound ICP has to answer three questions a marketing ICP can leave open: who to exclude, who to contact first, and how to reach them.
Inbound marketing can afford a broad ICP because the buyer chooses to show up. In outbound, you choose who to interrupt. Every wrong contact costs you something: a credit to source it, a send that counts against your domain’s reputation, and a rep’s time. A broad ICP in outbound shows up as low reply rates, bounces and a domain that slowly stops reaching the inbox.
Timing matters too. Research from LinkedIn’s B2B Institute with the Ehrenberg-Bass Institute estimates that only about 5% of B2B buyers are in market at any given time; the other 95% won’t buy for months or years (Marketing Week). An ICP tells you who fits. It cannot tell you who is in that 5% this quarter. That is what tiering is for.
How do you build an ideal customer profile for outbound?
Build it from evidence in five steps: list your best and worst customers, find the shared traits, write exclusions, add the buying signals that predict timing, and test the result on a small first campaign.
Step 1: List your wins, your losses and your regrets
I ask for three short lists, not a data export:
- Best customers: the ones who bought quickly, paid without friction and would buy again.
- Lost deals: the ones who took meetings and didn’t buy, with the reason if anyone knows it.
- Regrets: customers you would not take again (slow payers, scope creep, poor fit).
Ten names in each list beat a thousand rows in a CRM nobody trusts. The regrets list is the one most teams never write, and it produces the exclusions that save the most money.
Step 2: Find the traits the best customers share
Look across the best-customer list for what repeats: industry and sub-industry, company size by headcount, the title of the person who signed, the tools they already used, how they were structured (founder-led, owned by a larger firm, regulated or not). Write down only the traits that show up in most of them. Traits that show up once are anecdotes.
Step 3: Write exclusions as hard rules
Exclusions are the most useful part of an outbound ICP, because they remove records before they cost anything. Mine usually include:
- Company types from the regrets list
- Sizes below the point where the buyer can afford the service
- Current customers and open opportunities (so outreach never collides with an active deal)
- Competitors and partners
- Contacts with no verifiable business email or profile
Step 4: Add the buying signals that predict timing
A signal is something visible from the outside that suggests a company may be ready to buy now. I keep the list short and observable:
- An active job posting for the role you support. This is my Tier 1 signal. A company hiring for that role has a budget, a named problem and a deadline.
- A leadership change in the department you sell to.
- New funding, a new location or an acquisition.
- Public complaints or reviews about the problem you solve.
- Engagement with your own content or profile.
Step 5: Test it on a small first campaign
Write version one narrow, send to a few hundred accounts, and read the results by segment. The ICP is a hypothesis until real buyers answer it.
How do you tier accounts by buying signal?
Tiering ranks the accounts that fit your ICP by how ready they look to buy right now. I use three tiers, and each tier gets a different amount of personal effort.
| Tier | What qualifies | How I treat it |
|---|---|---|
| Tier 1 | Fits the ICP and shows a strong current signal, led by an active job posting in the role you support | Researched one by one, personal first line, every channel, the senior sender |
| Tier 2 | Fits the ICP with a weaker signal (leadership change, funding, engagement) | Segment-level personalization, two channels |
| Tier 3 | Fits the ICP, no current signal | Standard sequence at safe volume, recycled into Tier 1 or 2 when a signal appears |
Tiering does two things. It puts your best effort where timing is most likely, and it keeps Tier 3 from crowding out the accounts that matter. A tier is not permanent: an account moves up the day a signal appears, which is why signals need to be checked on a schedule, not once.
What fields should every ICP record hold?
Every record should carry the fit data, the tier and its reason, the best channel, and the email provider, so each channel can work from the same list without re-researching it.
This is the part I have found missing from almost every ICP template. The ICP lives in a slide; the list lives in a spreadsheet; nothing connects them. I put the ICP into the record itself:
| Field | Why it is there |
|---|---|
| Fit criteria met (yes or no, per criterion) | So anyone can see why the account is on the list |
| Exclusion check passed | So an excluded account never slips back in |
| Tier and the signal behind it, with the date seen | So the tier can expire when the signal does |
| Best channel (LinkedIn, email, phone) | Different buyers answer on different channels |
| Email provider (Google, Microsoft, other), from the MX record | Each provider filters cold email differently, so I send to each from matched infrastructure |
| Verification status | Unverified emails raise bounce rates and damage sending domains |
I added the email provider field after a cold email campaign where the same message reached Microsoft inboxes and mostly missed Google ones. Since then every list I build is tagged by provider before the first send, and that tag is part of the ICP record, not an afterthought in the email tool. I wrote up the deliverability side in the cold email deliverability setup guide.
What does a good ICP change in the results?
A good ICP shows up in three places: a smaller list, a higher share of conversations that turn into opportunities, and a clear answer on which channel to lead with.
The list gets smaller. The 12,181-contact audit I mentioned removed 2,920 records, about 24%. Fewer sends to the wrong people means fewer bounces and complaints, and that protects the domains for the people who do fit.
The channel becomes obvious. In one client program for a fractional executive firm, LinkedIn produced 52 qualified leads, against 3 from email (9,240 emails delivered) and about 5 from phone (2,283 dials). The ICP was the same across all three channels. What differed was where that buyer answers. Without a channel field on each record, the team would have kept splitting effort evenly. The LinkedIn side of that program is written up in the fractional executive firm case study.
The scoreboard moves past replies. A broad ICP can still produce replies. What it doesn’t produce is meetings that turn into opportunities, because many of the people replying could never buy. I judge an ICP by conversions: meetings held that became opportunities, and deals closed.
Common mistakes I see in ICPs
- Writing it from the product, not from the deals. “Companies that need X” describes everyone. “Companies like our ten best customers” describes someone.
- No exclusions. The list keeps every record that might fit, and the domain pays for it.
- Treating the ICP as finished. Markets move and signals expire. Review after every 20 to 30 meetings or each quarter.
- Too many ICPs. Past three, the profile stops filtering anything.
- Keeping it out of the data. If the tier and the reason aren’t on the record, the rep can’t use them.
FAQ
What is an ideal customer profile in B2B? An ideal customer profile (ICP) describes the type of company that gets the most value from what you sell and is most likely to buy, stay and pay. It describes companies, not people. Buyer personas describe the people inside those companies.
What is the difference between an ICP and ICP tiering? The ICP says which companies fit. Tiering ranks the companies that fit by how ready they look to buy right now, so the most timely accounts get the most personal outreach first.
Why is an active job posting a Tier 1 signal? A company hiring for the role your product or service supports has a budget, a named problem and a deadline. That combination is rare and visible, so I treat it as the strongest signal and contact those accounts first.
How many ICPs should a company have? Usually one to three. If you write more than three, the list stops being a filter. If two profiles need the same message and the same channel, merge them.
How often should I update my ideal customer profile? Review it after every 20 to 30 meetings or every quarter, whichever comes first. Compare the companies that turned into opportunities with the ones that only replied, and tighten the criteria toward the first group.
Can I build an ICP without many customers? Yes. Start from the deals you have, the deals you lost and the customers you never want again, write a narrow first version, and let the first campaign results correct it.
Get the foundation right first
If you want the ICP, the tiering and the verified list built for you and handed to every channel, that is our lead generation and ICP tiering service. If you are not sure your website and sending domains are ready for buyers to look you up, start with the free SEO and domain scan.
Sources
- Clearbit, How to create an ideal customer profile for B2B leads
- Marketing Week, The 95:5 rule is the new 60:40 rule (Peter Weinberg and Jon Lombardo, LinkedIn B2B Institute)
- Our own program data: a 12,181-contact list audit and channel results from a live client program (anonymized)