A demonstration

From the customer list you give us to a scoring model you can run against any company.

5 min 55 secSound on

What it shows, in writing

The short version, for anyone who'd rather read than watch. The complete made-up record is on the example page.

Start with your customer list

Your team gives us examples of good customers and says what makes them a fit. That's the raw material for the written profile – the companies, markets and exclusions to research.

Turn the profile into research questions

The profile names the company facts and buying roles that matter to the sale. Specialist criteria need evidence, and anything we can't confirm stays visible as a gap.

Agree the scoring

Weights and bands turn the criteria into a score and a tier. A sample gets checked before the work widens. A high score says a company looks like your best customers, not that it's buying.

This is an earlier walkthrough. Any activity, meeting or opportunity figures in it are illustrations, not measured results – the summary above is the current offer. Captions are on the player (the CC button) and the full transcript is below.

Read the full transcript (925 words)

This is the ICP Lab. In this video, we take you through the whole process from start to finish, from the customer list you give us, through to a scoring model you can run against any company. Most businesses can tell you who their good customers are, but few have a defined, scored ideal customer profile. The lab works that out from evidence in six steps. You tell us about your business, we research your customers, we describe each ICP, we test those descriptions, we build the scoring, and we set up the personas. It starts with a short survey.

You paste in the customers that make up roughly 80% of your sales, and then a few companies that look similar but are a poor fit, with an explanation on why. From there, it asks which channels you sell into and what you look at when you're deciding whether a company is worth pursuing. Once that's submitted, the rest of the work is ours. The first thing we do is research every company on that list. The good customers and the poor fits alike. For each one, the system works out what the company does. The channel it sits in, its industry, who it sells to, who owns it, and where it is.

Then the AI analysis runs across all of that evidence, and tells us how well each attribute separates your customers. In this example, it recommends segmenting by channel, and it gives the reasons along with what the poor fits have in common. You can click through the other tabs to see the same companies grouped another way, and the weaker options are labelled as weak. The poor fits show in red and they're never counted as part of an ICP. From there, we shape the groups together. Here the retail group only has one customer in it, so we merge it with the distributors.

Once the list looks right, we create the ICPs. Now we describe each one, and this is the part that matters most. The system drafts a name and two descriptions from the research on the companies in that group. What they are covers what those companies have in common. What they do, how they use the product, and how they buy. What they are not covers the companies that look similar, but that you want kept out. The draft is usually a fairly good start, but you know your customers better than the research does. That's why these next stages, the descriptions, the scoring, and the personas are done in a workshop together with your team.

We go through each description and change the wording until it reads right. The poor fits are kept together in their own group, because the definitions have to keep those companies out. The next question is whether those descriptions are defined enough, and we check that with a blind test. For each company, an AI researches it from scratch and places it in one of your ICPs using only those descriptions. The poor fits are tested as well. A poor fit matches when the AI fits it to none of your ICPs. Here, 16 of the 18 match and 2 differ.

Both are poor fits that the AI has let into the manufacturers group. The system reads every definition and suggests what to change. We open the definition with that change, read it, and save it. Then we run the test again. This time, all 18 match, and all 4 poor fits are kept out. The next step is scoring. Knowing that a company fits an ICP is useful, but you also need to know which ones to go after first. So, for each ICP, we pick the attributes that separate a top-tier customer from a marginal one, usually 3 to 5, and each one is researched for every company in the group.

Then we give each attribute a weight, and the weights have to add up to 100%. After that, we set the tier rules, which say what value earns gold, silver, and so on, for each attribute. Underneath, your own customers are scored live against those rules. And this is where you try different things. If we decide that shifts matter more than headcount, we move the weight across, and the scores change as we type. We can also move the tier thresholds. If gold is too hard to reach, we drag the marker down, and another company moves up to gold.

The other thing to do is add a company you know is a bad fit, and type its values in by hand. If the model scores it at the bottom, that's a good sign. The last piece is the personas, which are the types of people you want to reach inside each company. You give each one a name, and paste in LinkedIn profiles of real people who fit. The system pulls those profiles and drafts a description of what that person cares about, and what triggers a buying decision. Then you tick which personas apply to which ICP, because the people you'd speak to at a manufacturer are different from the ones at a reseller.

So that's the ICP Lab from start to finish. What you get is a set of ICPs described in plain language, tested against your own customers with a scoring model, and the personas for each one that not only our automated business development software will use, but the entire marketing, BD, and sales teams now have a definition they can align to.

Transcribed from the recording's audio and read through for errors. Captions on the player use the same text.

The ICP Lab is where the rest of the system starts; to see what happens once the profile exists, watch the full demonstration.

The first step is a discovery meeting – your market, what's in your CRM today and what a good customer looks like for you. Book a discovery meeting, or read the whole thing first.