Our Client Onboarding Process: Build the Brain First
Written By: Shane Clark on July 8, 2026
Most agencies start a new project by asking for a logo and a login. We start somewhere else. Before we touch a single page, we build a second brain for the client: a structured, private knowledge base that knows the business better than the website ever could. The build comes later, and it comes out better, because by the time we start it we already know where the money is, where the client wants to go, and where they are stuck. This is our exact client onboarding process, the sequence we run from the moment a client says yes, and the thinking behind each step.
The one idea underneath all of it: the website is the least informative thing in the room. A website tells you what a business says about itself. The client’s own words and their real data tell you what the business actually is. The gap between those two things is where every good decision lives, so our whole onboarding is built to close it.
- What kind of client is this?
- One folder, not forty
- The data firewall
- The discovery call
- Pour everything in
- The two lenses that change the build
- Build the brain, then build the thing
- Why we start with the brain
- Frequently asked questions
Step one: figure out what kind of client this is
The first thing we do when a deal closes is classify the engagement, because a solo operator and a multi-location company do not get the same setup. We sort every new client into one of a few buckets. A solo or one-person business. A small to mid-sized business with a team. Then a larger or multi-project company where the relationship will hold several jobs over time. Or a product, where the work is the thing we are building rather than a service business we are marketing.
That single decision tells us how to structure everything downstream: whether this is a one-off project or an ongoing relationship, how much folder structure they need, and how heavy the process should be. A solo plumber does not need the machinery an enterprise needs, and an enterprise will drown if you treat it like a solo plumber.
We also decide, right here, what we are actually being hired to do, and we never assume it is a website. Some clients need marketing and search visibility. Others need a new site designed. A few need something built or fixed. And a growing number need automation, the quiet work of taking a repetitive task off a human’s plate. Knowing which of these is in play tells us which intake questions matter and keeps us from forcing every client into a website-shaped box.
Step two: one folder, not forty
Here is a small thing that causes an enormous amount of pain if you get it wrong: where the client’s stuff lives. Most businesses have their assets scattered across a dozen places. A logo in an email from two years ago. Photos on someone’s phone. The brand colors in a PDF nobody can find. If you let that chaos into your process, your process becomes chaos.
So our client onboarding process sets up a single source of truth. One shared cloud folder, structured properly, that becomes the home for everything. If the client already has one, they share it with us. If they do not, we build it for them and hand it back, which is a quiet gift, because now their business is organized whether they hire us again or not.
And we build that folder as a real business filing system, in two zones. The first zone is the share zone: brand assets, content, reference material, the things we are allowed to see and use. The second zone is the internal zone, clearly marked do not share, for everything that should never leave the building. The two zones are color coded and labeled so nobody has to remember the rule, because the structure enforces it.
The data firewall, and why it is not optional
That second zone matters more than it looks, so let me be blunt about it. We ingest the shared folder into an AI knowledge base to build the client’s brain. That means whatever is in the share zone becomes context the AI can read. Which means certain things can never, ever go in there.
Never goes in the shared, ingested folder: customer records or personal information, patient or health information, invoicing, banking, payroll, or financial statements, and passwords, keys, or logins. That is the data firewall, and it is a hard line.
Sensitive material has a home in the internal do-not-share zone, nowhere near the part we scan. When we need a credential to do the work, it comes through a secure intake, into a password manager, and never through the shared folder. The firewall is structural, not a promise we have to keep remembering. The sensitive stuff simply has nowhere to land except the locked room.
Step three: the discovery call, and why we record all of it
Now the most valuable hour of the entire client onboarding process. We get on a call with the client and we talk about their business. Not a rigid script, because a script makes you sound like a call center and makes you miss the interesting thread the client just opened. A guide. A set of boxes we make sure get checked before we hang up, while the conversation itself is allowed to breathe.
And every call is recorded and transcribed. We tell the client up front: I take notes and transcribe our calls so nothing gets lost. Because in forty-five minutes a client will say things they would never write on a form and that will never, ever appear on their website. The transcript is gold. It becomes fuel for the brain, and it quietly steers the content, the site, the callouts, and the priorities.
Here is roughly what the guide covers. It goes deeper than a standard intake, though it overlaps with the make-or-break questions to ask before building a website.
The questions our client onboarding process always covers
The business, in their own words. How it started, what they sell, who buys it, and a walk through their services. This opens every other door.
The money. This is the block almost nobody asks, and it is the most important one. Which service makes you the most money? Which one is the biggest headache for the smallest return? If you could do more of one thing and less of another, what is the trade? You will never learn this from a website. A website lists every service equally, smiling, with no hint of which one pays the mortgage. Ask it out loud on a recorded call and the client just tells you: the kitchen remodels are the margin, the small repairs barely break even. That one answer reshapes the whole build, and it shapes how we price the work, which is why we price on outcomes, not hours.
The goal. Not “we want a website.” The real outcome. What does a win look like a year from now? What are you actually hiring us to change? Has anyone done this for you before, and what happened? What is the number that would make this worth it? We care about the outcome, not the vanity metric, the same reason rankings do not matter without conversions.
The customer. Describe your best customer, then the one you wish you had fewer of. Where do your good leads come from now? And the quiet goldmine: what do customers always ask before they buy? That question hands you your FAQ and your objection-handling copy in the customer’s own language.
The blockers. Where do you spend the most hours? What keeps you stuck? What do you dread doing every week? We are not running a deep audit on the call. We are just listening for the time-sinks, because this is where automation lives. A client will happily describe their own bottleneck without realizing they just named the next project. The manual invoicing, the copy-paste report they build every Monday, the same five emails they answer all day. None of that is on the website either.
The landscape and the practical. Competitors they admire or resent, what makes them different in one honest sentence, and the housekeeping: service area, hours, seasonality, who signs off on decisions, and what assets they can share.
Step four: pour everything in
By this stage of the client onboarding process, the folder is set up and the call is transcribed, so we ingest. And we ingest widely, because context is the whole advantage. This is client data put to work, not just stored in a folder. We encourage clients to over-share. Give me the old flyer, the business card, the logo, all of it. There are two kinds of input, and we want both.
The first kind is documents we scan: the brand assets, the transcribed calls, the existing website copy, service sheets, photos, past proposals, and anything else in the share zone. The second kind is live data we connect: Google Search Console to see what people actually search to find them and which pages already pull, Google Analytics to see what converts (and if they are on an old setup, we handle the move to Google Analytics 4 first), the Google Business Profile and its reviews, past ad history, and the reviews and social presence that show how customers really talk about them.
These two streams answer two different questions. The documents and the call tell us who the client says they are and what they want. The data tells us what is actually happening. Sometimes the data agrees with the client. Sometimes it says “you make your money on kitchens, but every visitor to your site is looking for a cheap faucet repair,” and now you have a real strategic conversation to have. Both streams matter, and neither one alone is enough.
Where the client onboarding data comes together
There is a clear hierarchy to these inputs, and it is the opposite of how most agencies work. At the top sits the client’s own words from the discovery call: the goals and the profitability, the things you can only get from a human. Below that sits the connected data, which validates and sharpens the story. At the bottom sits the website scan, which only tells you what already exists, not what should. Most agencies build from the bottom up, starting with the site. We build from the top down, starting with the goal.
The two lenses that change the build
Once the brain is full, two lenses shape everything we make.
The first is the profitability lens. Once we know which service actually pays, the whole build bends toward it. The homepage leads with the money-maker instead of an alphabetical list. The calls to action push the high-margin job. The content targets the searches that feed it, and the internal links funnel authority to those pages. You lean the site where the profit is, on purpose, because you asked the one question nobody else asked.
The second is the blockers lens. The time-sinks the client described are not complaints, they are a project list. Every “I spend all day doing this by hand” is an automation waiting to be built, part of the same shift where AI is absorbing the repetitive work agencies used to hand off. We are not solving it on day one, but we are noting it, because a client who came for a website often has a much more valuable problem hiding in the parts of their week they dread.
Step five: build the brain, then build the thing
Here is how our client onboarding process actually assembles the knowledge base, and the method matters. We do not wait for every scrap of data and then invent a structure around it. We build the skeleton first: a fixed, consistent set of files that every client gets, born mostly empty. A place for goals, a place for brand, a place for the business profile, a place for the content inventory. Then the discovery answers and the scanned assets and the connected data fill those slots.
We build it that way for two reasons. Consistency, because a fixed structure means we always know where the brand colors or the business goals live, for this client and every other. And a built-in checklist, because the empty slots are a gap map. A blank “brand colors” field is a to-do. A blank “primary goal” is a follow-up call. The skeleton tells us what we still need. When the load-bearing parts are full, the brain is live, and it knows the whole business.
Now, and only now, we build. The website, the content, the automation, whatever the engagement called for. And it goes faster and lands harder, because the brain already did the hard part. Every decision has an answer waiting: what should the hero say, which service leads, what do customers ask, where is the profit, where is the pain. We are not guessing. We are reading it off the brain we spent the first week building.
Why our client onboarding process starts with the brain
You can build a website from a logo and a login. Plenty of people do, and it looks fine, and it never sends a single lead, because it was built from the least informative thing in the room. We would rather spend the first stretch of a project listening, organizing, and understanding, so that everything after it is aimed at the right target.
The website is the easy part. Knowing what the website should do, that is the whole game, and you only get that from the client’s own words, their real numbers, and the quiet problems they mention when you ask the right question and hit record.
Client onboarding process: frequently asked questions
Why build a knowledge base before building the website?
Because the website is the least informative thing in the room. It tells you what a business says about itself, not what it actually is, where the profit lives, or where the owner wants to go. When you build the knowledge base first, every later decision has an answer waiting, so the site comes out sharper and faster.
What should go in a client onboarding folder?
One structured folder in the cloud holds it all, split into two zones. The share zone holds brand assets, content, reference material, and photos, the things we are allowed to use. The internal zone, clearly marked do not share, holds everything sensitive. Keeping it in one place instead of scattered across email and phones is half the battle.
What client information should never be shared with the agency?
Anything sensitive stays out of the shared, ingested folder: customer records and personal data, patient or health information, invoicing and financial statements, and any passwords or logins. We call this the data firewall. Sensitive material lives in the internal do-not-share zone, and credentials come through a secure password manager, never a shared folder.
Why record and transcribe the discovery call?
Because in forty-five minutes a client says things they would never write on a form and that never appear on their website. The transcript captures their real goals, their margins, and their frustrations word for word. That becomes fuel for the whole build, and it means nothing important gets lost or half-remembered.
What questions should you ask on a client discovery call?
We treat it as a guide, not a script. We cover the business in their own words, the money and which service is most profitable, the real goal a year out, the ideal customer, the daily blockers, and the competitive landscape. The money and blocker questions are the ones most people skip, and they are the most valuable.
How do you find out which service is most profitable?
You ask, out loud, on a recorded call. You will never learn it from a website, which lists every service as equal. So we ask which service makes the most money, which is the biggest headache for the smallest return, and what the owner wishes they did more or less of. That one answer reshapes the entire build.
What data sources feed the client knowledge base?
Two kinds. Documents we scan: brand assets, transcribed calls, existing site copy, service sheets, and photos. And live accounts we connect: Google Search Console, Google Analytics, the Google Business Profile, past ad history, reviews, and social. The documents tell us who the client says they are, and the data tells us what is actually happening.
How do you spot automation opportunities during onboarding?
We listen for the blockers. On the discovery call we ask where the owner spends the most hours and what they dread doing every week. The manual invoicing, the Monday report rebuilt by hand, the same five emails all day. Each of those is an automation waiting to be built, and clients name them without realizing it.
Should you build the knowledge base structure first or generate it from the data?
Build the structure first. We start with a fixed, consistent skeleton every client gets, then let the discovery answers and scanned data fill the slots. A fixed structure means we always know where the brand colors or the goals live, and the empty slots double as a checklist of what we still need.
