What you'll be able to do after this guide: write a 1 to 5 fit score that an agent can apply consistently, and use it to decide which prospects get expensive research and outreach.
ICP stands for ideal customer profile: a description of the customers you're best placed to win and serve. A fit score measures how closely a prospect matches it.
An agent researching 15 companies a day needs a consistent way to say "this one is worth a person's time and this one isn't". A five-point scale is coarse enough to apply consistently and fine enough to rank. Stored as a rating field, it can be sorted and filtered, which the outreach agents rely on to pick who goes first.
Every rubric we use builds the score from three things:
Strength of the signal. How sure are we that the company has the problem we solve? This uses the evidence tiers in Grade Your Evidence.
Commercial fit. Is the company the right size and type to buy?
Reachability. Have we found the actual decision-maker, with a confirmed way to contact them?
The score never measures whether outreach has happened. That is what status fields are for.
| Score | Meaning | Fernhill's Segment A version |
|---|---|---|
| 5 Strong | Confirmed signal, clearly qualifies, decision-maker with a confirmed email | Competitor product confirmed from the company's own pages, Direct type, named decision-maker with a confirmed work email |
| 4 Good | One step short on evidence or on contact | As 5, but the product is a strong inference rather than confirmed, or there is no confirmed email |
| 3 Plausible | A real fit, weaker on both | Plausible product, Self-serve type, in the size range, decision-maker found |
| 2 Weak | Circumstantial evidence only, or unclear who decides | Only sector assumptions; or no confident decision-maker |
| 1 Poor | Shouldn't be on the list | Wrong buyer, excluded category, or the company is not legally active |
Reserve 5 for cases where you can point at the thing you would replace. Some companies show a clear need (for example, a regulation they must comply with) but no evidence of a current tool or a current problem. Cap these at 4. A confirmed obligation tells you a need exists; it doesn't tell you there's a pain you can name in a message.
Hard disqualifiers override everything. If the company register shows the business is in liquidation, administration or dissolved, the score is 1, the status is Skipped, and research stops, however good the rest looks.
A competitor product doesn't lower the score. A poor or outgrown tool is the buying signal. Only treat a current tool as disqualifying when there is positive evidence it fits them well.
Manual adjustments need a reason. If the agent scores lower than the rubric suggests, it must say why in a note. Repeated adjustments of the same kind are a sign the rubric needs a new rule.
State the fit in words as well as a number. "Score 4: confirmed partner of a competing product, no confirmed email yet" lets a person check the reasoning quickly. The number is for sorting; the words carry the reasoning.
A company can genuinely fit two segments: a design studio that sells to businesses and also sells templates directly to the public, for example. Give each segment its own score field so the agent can set both honestly. Two rules:
Score each honestly. A thin secondary signal gets a 2 or 3, not a 5. Don't inflate a score just to fill the field.
Routing follows the primary segment. The second score helps future targeting and reporting; it doesn't change which outreach the company receives.
The score's most important job is deciding where expensive work goes. Fernhill's agents do cheap research for every candidate, then:
Scores 1 to 3: record the research, mark the company Skipped, stop. No message is drafted. A 3 is a real fit, but not a strong enough one to spend outreach on yet; revisit 3s when sourcing runs dry.
Prospect type Unknown: whatever the score, leave the company at Researching for a person to look at. It never goes to Review.
Scores 4 and 5: browse the prospect's own website and profiles, choose the value angle that matches what they found, write the value hypothesis and the draft, then set status to Review.
Gating the personalised work behind the score keeps runs fast and keeps the outreach list full of companies worth a message.
Paste the full rubric into the description of the score field in your CRM, and into the agent's spec. An agent reading the field definitions should never have to go looking for the rules.
Next: Grade Your Evidence.
