The clearest sign that an agency evaluation went wrong almost never shows up during the evaluation. It shows up four months into the engagement, when a site migration breaks conversion tracking that took two years to build, or when a compliance review kills a campaign that should never have been proposed in the first place.
Nothing in the pitch predicted it. The case studies were strong. The references checked out. The pricing was competitive. The gap between how a vendor looked on paper and what they could actually do was there the whole time, and the evaluation process was not built to find it.
That is a scoring problem more than a diligence problem. Most selection frameworks treat their criteria as roughly equal, which quietly assumes a deep portfolio can offset thin knowledge of your industry. It cannot.
Why Most Agency Scorecards Flatten the Decision
Scorecards flatten the decision because they grade every criterion on the same scale, which lets strength in the easy categories cancel out weakness in the one that actually decides the engagement. Search for an evaluation framework and the same structure repeats: eight to 10 criteria covering services offered, reputation, reporting, pricing, tools, team structure, and cultural fit.
The individual items are all reasonable. What the format hides is that they are not equally predictive, so teams score each from one to five, add up the total, and the math rewards the wrong vendor. An agency with a large portfolio, recognizable logos, and a polished proposal scores well across most categories whether or not anyone on the team has worked inside your industry. A specialist firm with a shorter client list loses points on portfolio breadth while holding the single advantage that changes outcomes.
Separating Table Stakes From Real Differentiators
Table stakes are the criteria nearly every credible agency clears, and differentiators are the ones that still vary widely among the firms that cleared them. Clear reporting, transparent pricing, defined points of contact, and baseline technical competence belong in the first group. A vendor who fails one of them exits the process. A vendor who passes has told you almost nothing about how they will perform against the next firm that also passed.
Brand-name clients are the most misread signal in the set. A logo on a case study page tells you an agency won a deal. It does not tell you who staffed the account, how wide the scope was, whether the relationship lasted, or whether that client operates anything like you do. Portfolio size carries the same flaw, because volume of work measures sales capacity more than delivery quality.
Industry knowledge behaves differently. It is not a threshold that vendors clear or fail, and the gap between two firms that look identical on a scorecard compounds across every decision made once work begins.
What Industry Expertise Actually Protects You From
It protects you from the early technical decisions that look correct at the time and turn expensive later. The cost of a generalist partner is rarely one bad campaign. It is a sequence of avoidable choices made before anyone could see the consequence.
In regulated industries, it looks like creative that clears internal review twice and fails on the third pass because nobody flagged a disclosure requirement during scoping. In organizations with long sales cycles, it looks like an attribution model built around form fills when revenue takes nine months to appear, leaving leadership with reporting that cannot answer the only question they ask. On complex websites, it looks like a migration planned without accounting for the integrations behind it, which is how a company loses rankings, tracking continuity, and lead routing in the same week.
None of those are execution failures. They are knowledge failures, and they happen before execution starts. A website migration with multiple integrations is a good example, because the risk lives entirely in what the team knows to check beforehand.
How to Test for Subject Matter Knowledge Before You Sign
Ask about failure instead of success. Every vendor prepares for questions about wins, so the fastest way to find the edge of someone’s experience is to ask what goes wrong in work like yours and when.
The standard process will not surface this on its own. Case studies are written by marketing teams, references are hand-selected by the vendor, and discovery calls are structured so the agency asks the questions and the prospect answers them. These five questions reverse that direction.
What typically goes wrong in engagements like this one, and at what stage does it usually surface?
Which constraints in our industry most often force a change of plan after work is already underway?
Which project in our vertical underperformed, and what would you do differently now?
Who specifically would staff this account, and what have they worked on in our industry?
What would you need to see in our current setup before committing to a timeline?
A generalist answers these in terms of process. A specialist names the review cycle that always runs long, the integration that always breaks, or the attribution gap that appears around month five. Precision about what goes wrong is the closest available proxy for real experience, because it cannot be researched the night before a pitch.
Telling Confidence Apart From Competence
The difference shows up in what a vendor chooses to talk about. Confidence produces language about outcomes, which no agency controls, and competence produces language about constraints, which only repetition teaches.
Outcome language is cheap to produce. A vendor who has never worked in your industry can still speak fluently about strategy, growth, and alignment, because that vocabulary is portable and commits to nothing.
Constraint language works the other way, because the only way to acquire it is repetition. When an agency pushes back on a timeline, names a dependency you had not considered, or tells you outright that a request is a bad idea while the contract is still unsigned, that carries more weight than any case study. Vendors willing to introduce friction before there is money on the table usually know where it lives.
Weighting Your Next Agency Decision
Weight it toward two things: demonstrated knowledge of how projects fail in your industry, and the experience of the people who will actually do the work rather than the team that shows up for the pitch. Everything else belongs in a pass or fail gate.
Reporting, pricing, communication structure, and technical baseline either meet your requirement or they do not, and grading them on a scale adds nothing. The two weighted factors predict more of the outcome than portfolio size, client names, and price combined. A vendor who wins under this approach is often not the one who presented best, which is precisely the point.
How Marcel Digital Approaches Industry Specific Work
Marcel Digital builds around industries rather than treating every account as a variation on one playbook. That includes financial services, healthcare, manufacturing, and other categories where compliance requirements, sales cycle length, and technical complexity change what a correct strategy looks like.
The practical effect shows up early. Our teams raise the constraints that matter during scoping instead of discovering them in month four, whether that means connecting paid media spend to actual revenue across a long buying cycle, planning a migration that runs through several integrations, or building organic visibility in a category where AI-assisted search has already changed how buyers find information.
If a previous agency left you with work that has to be redone, the cause usually was not effort. It was that nobody on the account had seen your problem before. Contact Marcel Digital to talk through what your next engagement requires, and who would be doing the work.
Frequently Asked Questions
More than any other single factor. Criteria like pricing transparency, reporting cadence, and communication structure are better treated as pass or fail requirements, because most credible agencies clear them. Once a vendor passes those thresholds, industry knowledge and the experience of the assigned team are what separate one qualified firm from another.
Not on its own. Portfolio size reflects sales capacity, and a recognizable client logo confirms only that an agency won that deal. It does not tell you the scope of the engagement, who staffed it, how long it lasted, or whether that client resembles your business. A smaller portfolio concentrated in your industry is usually the stronger signal.
Questions about failure rather than success. Ask what usually goes wrong in engagements like yours and at what stage, which industry constraints most often force a mid-project change, and what they would need to review before committing to a timeline. Vendors with real experience answer with specifics. Vendors without it answer with process.
Listen for whether they talk about outcomes or constraints. Outcome language is portable and applies to any client, so it is easy to produce without relevant experience. An agency that pushes back on a timeline, flags a dependency you had not considered, or tells you a request is a bad idea before a contract exists is demonstrating knowledge that generalists do not have.
Most of them are technical decisions made early that only become visible later. Common examples include creative that fails a compliance review late in the process, measurement frameworks built around the wrong conversion events for a long sales cycle, and website migrations planned without accounting for the integrations running behind the site. These are knowledge gaps rather than execution problems, which is why they are difficult to catch during a standard evaluation.
