Discovery vs. Brief: Two Agency Models and How to Choose
The most common source of failed AI agency engagements isn't a bad agency. It's hiring the wrong type. Two fundamentally different models exist, and most buyers don't know they're choosing between them.
- agency selection
- discovery
- implementation
- strategy
The single most common source of failed AI agency engagements isn't a bad agency. It's a mismatch between what the buyer actually needed and what kind of agency they hired. Two fundamentally different engagement models exist. Most buyers don't know they're choosing between them, and the agency selection process rarely forces the question before it's too late to matter.
How the two models work
The two models work differently from the first conversation.
A Type A agency, sometimes called embedded or discovery-first, comes in without a defined brief. The engagement begins with a workflow audit: they map where your hours are buried, identify which processes have the highest automation potential and ROI, quantify the opportunity, and then scope and build against what they found. The relationship tends to be ongoing because the first build creates the data conditions for the second. The value of the engagement isn't only the software delivered at the end. It's the diagnosis that determined what to build in the first place. A Type A agency is hired to figure out the question before they answer it.
A Type B agency is a delivery shop. The buyer arrives with a defined brief: "automate our invoice approval workflow," "implement Agentforce for our inbound sales team," "build a document classification system for our legal intake process." The agency scopes against that brief, prices the project, and delivers it. The engagement ends at go-live, or shortly after. The buyer knew what they wanted. The agency built it.
Neither model is better. The mismatch is the problem. A buyer who needs discovery and hires a delivery shop gets a built thing that may not be the right thing. The delivery shop executed exactly what they were asked to do. The project failed not because of poor execution but because the brief was wrong, and nobody was hired to question it. That failure mode is invisible in the agency's portfolio and references, because from the agency's perspective the project was a success.
The one question that routes you
The diagnostic question that routes correctly is direct: do you know what you want built, or do you need help figuring that out?
If you can answer the first question with a specific workflow, a named platform or tool, and a rough scope, you're a Type B buyer. A delivery shop is faster, cheaper, and better suited to your situation. Bringing in a discovery-first agency when you already have a clear brief means paying for a process that produces a recommendation you don't need, because you already have one. The discovery phase slows you down without adding value.
If you can't answer the first question specifically, if your answer to "what do you want built?" is "something in sales ops" or "we want to do more with AI" or "we need help figuring out where to start," you're a Type A buyer. Hiring a Type B agency in this position means the agency will force a brief out of you during the scoping call, build against that brief, and deliver something that may have been the wrong starting point. The project will be executed competently. The outcome will be suboptimal, because the first decision was wrong and nobody's job was to question it.
The mismatch produces a particular kind of frustration because both parties did what they said they'd do. The agency delivered the project. The buyer approved the brief. The thing that was missing was a prior conversation about whether this was the right brief.
What each model costs
What each model costs is worth understanding before the search starts.
A Type A engagement typically runs in two phases. The discovery phase, which produces the workflow audit and implementation brief, costs between $10,000 and $30,000 and takes four to six weeks. The build phase, which executes against what discovery found, runs $50,000 to $200,000 or more depending on scope and complexity. Ongoing optimization adds cost in proportion to what was built and how quickly the organization wants to move to the next iteration.
A Type B engagement is simpler to price. A well-scoped delivery project in the $15,000 to $75,000 range can be quoted on a single call, because the brief already exists. The agency knows what they're building and can estimate the work accurately. Fixed-price contracts are common. The buyer knows the cost before the project starts.
The reason Type A feels expensive is that the discovery phase produces a plan, not software. You spend $15,000 to $30,000 and receive a document: a prioritized list of automation opportunities with estimated ROI and implementation cost, a workflow brief for the first project, and an agency type recommendation. The instinct to resist paying for thinking is understandable and frequently costly. The plan is what determines whether the build phase delivers something that works for the business. Without the plan, the build phase delivers something. Whether it's the right something is a different question, and the wrong something at $80,000 is more expensive than the right plan at $20,000.
Why discovery-first firms are harder to find
The supply problem is real and worth accounting for in the search. Type A agencies (those with a documented and reproducible discovery methodology) are harder to find than Type B agencies. Most agencies claim discovery methodology. Few have a defined process that produces consistent, reviewable outputs across different clients and industries. The claim is easy to make. The methodology is harder to verify than a portfolio of delivered projects.
When evaluating a Type A firm, the question that separates methodology from a sales pitch about methodology is: "Can you describe your discovery process step by step, tell me what it produces, and show me a redacted example from a prior engagement?" A firm with real methodology answers this specifically. They can describe the role mapping phase, the ROI quantification approach, the governance documentation, and the output format. They have a redacted example they can share. A firm without real methodology describes the process conversationally, uses words like "collaborative" and "tailored," and doesn't have a template they can show you.
Type B agencies are more available, and the selection process for them is more legible. You review portfolios of delivered projects, check references on execution quality, and evaluate technical fit against your brief. Buying execution is more concrete than buying methodology, which is partly why most buyers default to it even when they need discovery.
Know your answer before you search
The mismatch is almost always avoidable. It happens because buyers search for agencies before they know which type they need, and because agencies rarely volunteer that the engagement model might not be the right fit for a buyer's situation. The intake question that prevents it is simple: ask yourself, before you start the search, whether you know what you want built. The answer determines the category. Then verify the category of any firm you're seriously considering before the proposal stage. A Type B firm will usually be honest that they're not set up for discovery if you ask directly. A Type A firm with real methodology will be able to prove it.
The question to ask is: "Do you know what you want built?" Know that answer yourself first. Then find the type of firm that matches it.
If you want that question answered for your specific situation, the Forge Playbook does it. Answer a few questions about your business and we'll put together a tailored outline of which workflows are worth automating and what a realistic budget looks like for each. Free, no obligation, takes about three minutes.
