Finding ICP Accounts is a Process
Ask someone to research an account, and they can do it. How do you build an automation to scale that process? Read: "Everything Starts Out Looking Like a Toy" #221
Hi, I’m Greg 👋! I write weekly product essays, including system “handshakes”, the expectations for workflow, and the jobs to be done for data. What is Data Operations? was the first post in the series.
This week’s toy: A kind of ASCII art generator. Like this version of Star Wars displayed only in text. A lot of people use terminal applications and it’s handy to show process documentation and display it in a terminal - delivering a graphical bridge from flow chart to terminal.
Edition 221 of this newsletter is here - it’s October 21, 2024.
If you have a comment or are interested in sponsoring, hit reply.
The Big Idea
A short long-form essay about data things
⚙️ Finding ICP Accounts is a Process
The challenge: you want accounts that match the criteria for your ideal customer.
The reality: start by making a list of companies and then find the ones that match the qualities of your ideal customers.
Also reality: spend a lot of time thinking about what data points are discoverable and which data points potentially describe a solid prospect.
It’s hard to define the combination of factors that make an arbitrary company a good prospect. Just because a prospect matches some aspects of existing customers doesn’t guarantee they are an ideal prospect for your company.
Data is often available but unstructured
In a perfect world, you’d define a data point that marks an aspect of a solid customer and track that automatically for every customer. Over a series of steps, this process lets you build factors that score accounts for nearness to “ideal”.
Here’s what’s available to research about an arbitrary company with some web domain history:
Firmographic traits (number of employees, industry, headquarters location, address, revenue estimates, year of founding)
Technographic items (what software do they use on their website, what do surveys say they use internally)
Popularity (if the site is popular enough you can get estimates of site ranking, web ad spend)
Job Listings (date stamps of available jobs from various job listing sites)
Content from a company’s web site (the public site)
For public companies, SEC information and filings give you additional information
There’s also good old web search - but it can be challenging to triangulate information about a company with the validation of that information. Simply put, there be information dragons. (LLMs don’t make this easier, as they will often confidently state information as facts without providing receipts.)
For an individual company, research is possible
Demonstrating to a seller that an account is an ideal customer is a repeatable process. You might be looking for a specific product or service on their website, an employee characteristic (they have more than a certain number of people working for them), or perhaps a competitive software that they use.
Now, scale this effort to hundreds or thousands of potential accounts.
The first thing you need is a list of these accounts and a unique ID or key to map future enrichment to these accounts. It would be nice to have consistent information about every account and key fields populated to enable you to research them.
The reality is often different. Most accounts have middling information based on what the prospect provides. The difference is often the ability of a seller to find that hidden piece of information.
Request for product: instant research
There’s a lot of news about “AI-automated” SDRs recently, or processes that take over the initial research and qualification of accounts based on a standard operating procedure.
I’m not bullish on letting AI handle tier 1 contact with a customer. However, I love the idea of having an AI agent conduct research using a very specific procedure on lots of accounts. What would make that promising result great? Being able to validate the result.
I’d love to see a data automation product that did the following:
define a hypothesis that can be proved, e.g. “does a company have a staff or “about us” page listing listing 5 or more employees”
search the available site and produce a yes or no answer to the question, including supporting information to validate the information with a date stamp
create a place to validate this observation, which when overruled removes that data point from the account
let you tune your observation to include “possibly correct” through “must be validated by a human”
With a system like this, it would be a lot easier to find information needles in a haystack of accounts at scale, even if there is a novel way to identify a data point.
What’s the takeaway? Account research is hard to do at scale. AI combined with humans can help process automation to find specific information on lots of accounts. This would make it possible to harness the intuition of sellers yet operationalize that search across all the accounts.
Links for Reading and Sharing
These are links that caught my 👀
1/ EVs are inevitable - whether you define electric vehicles as a battery-electric or full electric, the trend is away from ICE cars because the technology is older. The US isn’t moving quite as fast as some countries but the writing is on the wall: eventually, everything will at least be hybrid.
2/ Editing 2d art just got easier - Have you ever created a drawing or vector graphic and want to update it by rotating it in space? That just got easier with a new feature from Adobe. You can call it a slippery slope into “AI does everything” or identify it as a new feature that has great value sometimes.
3/ On doing the unexpected - Packy McCormack’s essay on getting rid of Playbooks. These shortcuts, he argues, prevent you from real first-principles thinking. I’d argue that knowing the expected conventions of a thing make it easier to understand when to break the rules, and that playbooks are actually incredibly valuable (if you don’t follow them all of the time).
What to do next
Hit reply if you’ve got links to share, data stories, or want to say hello.
Want to book a discovery call to talk about how we can work together?
The next big thing always starts out being dismissed as a “toy.” - Chris Dixon