The vendor landscape for consulting staffing tools has converged on a single rhetorical frame: automated decision intelligence. Every tool claims to be optimizing staffing allocations, recommending consultants, predicting engagement fit. The language of automation has become the default positioning for the entire category.
Kelpmont's position is different, and it's worth being precise about why. The distinction isn't marketing differentiation for its own sake. It reflects a genuine disagreement about which problem needs to be solved and what kind of tool actually solves it.
The Automation Premise and Its Problems
Tools that automate staffing decisions operate on a premise: that the staffing allocation problem is primarily computational — that given enough historical data about consultants and clients, an algorithm can identify the optimal assignment with better accuracy than a human partner can. If the premise is correct, automation is the right answer.
The premise is partially correct. Pattern recognition across historical engagement data does surface signals that are difficult for human memory to hold simultaneously. A system that has processed five hundred engagements can see patterns that a partner managing thirty active relationships can't hold in mind reliably.
But the premise fails on the part that matters most: staffing decisions at consulting firms aren't purely optimization problems. They're judgment calls that integrate multiple competing objectives simultaneously. Optimal fit for this specific engagement. Consultant development trajectory over the next two years. Client relationship dynamics that aren't captured in engagement records. Firm strategic objectives that make certain client relationships disproportionately valuable. Partner accountability to their team and to the firm's culture.
No engagement history record holds all of these. An algorithm that optimizes for measurable engagement outcome metrics while holding these variables constant is solving a simplified version of the problem — and optimizing for the simplified version can produce decisions that are wrong in the dimensions that weren't in the model.
The Problem That Actually Needs Solving
The actual problem in mid-market consulting staffing isn't that partners make the wrong decision when they have the right information. It's that they consistently lack access to a specific class of information: what does our own engagement history show about fit patterns?
Partners who have staffed three hundred engagements carry genuine expertise. Their instincts are calibrated by real experience. But their memory is selectively reliable — they remember the dramatic cases clearly, the representative cases poorly. The pattern that their engagement history would show if systematically analyzed is different from the pattern their memory presents. The gap between those two things is the problem to solve.
Evidence-based staffing doesn't replace the partner's judgment. It corrects the information asymmetry that causes the judgment to be less accurate than it could be. The partner still makes the call. They make it with better information.
Why "Intelligence" Is the Right Word and "Automation" Isn't
The distinction matters practically as well as philosophically. Firms that deploy staffing automation tools face a consistent adoption challenge: partners who view the tool as displacing their judgment resist it, sometimes actively. The resistance isn't irrational. Partners at consulting firms are selling their judgment to clients. A tool that substitutes algorithmic recommendations for partner judgment undermines the firm's own value proposition.
Tools that surface evidence and return the decision to the partner don't have this adoption problem. There's no identity conflict. The partner's judgment is being enhanced, not replaced. The tool is like a better briefing before the meeting — it doesn't run the meeting.
We've observed this pattern play out consistently. Firms that position the engagement intelligence tool as a decision-support layer — "here's what the data shows about this configuration" — see strong partner engagement and high weekly active use. Firms that have tried tools positioned as recommendation engines describe the same story: initial interest, followed by adoption drop-off as partners experience the recommendations as a challenge to their authority rather than an aid to their judgment.
What Evidence-Based Actually Means
Evidence-based staffing, done well, has three operational characteristics that distinguish it from automated staffing:
Transparency of the signal source. The practice lead should be able to see exactly what historical data the signal is derived from. "This consultant's vertical familiarity score for healthcare is high because they've completed eight engagements with hospital or health system clients, with a composite outcome score of 83/100" is a transparent signal. A black-box fit score is not. The difference is whether the partner can interrogate the signal, confirm it with their own knowledge of the consultant, and adjust their weight on it accordingly.
Explicit acknowledgment of what's not measured. A good evidence layer should be direct about the variables it doesn't capture. Personality fit, consultant development goals, client-specific political dynamics, strategic relationship value — these aren't in the engagement record. A tool that presents its signal as comprehensive is misleading. A tool that presents its signal as one input among several, with explicit acknowledgment of the variables outside its view, is calibrated.
The decision stays with the human. There's no automated staffing action in an evidence-based tool. The signal is presented. The partner reviews it. The partner makes the call. This isn't a limitation — it's the design intent. The accountability for the staffing decision belongs with the partner, and the tool should reinforce that accountability, not diffuse it.
A Practical Test
When evaluating any staffing intelligence tool, ask the vendor two questions. First: can I see exactly what data points produced this recommendation or score? Second: can I override the recommendation without the system flagging me as doing something wrong?
If the answer to the first question is no, the tool is asking for trust it hasn't earned. If the answer to the second question involves any friction or discouragement, the tool is positioned as a decision-maker, not a decision support.
The staffing call at a consulting firm is a partnership between the data and the partner. The data provides pattern evidence. The partner provides judgment. Neither should be subordinated to the other. Tools built on this principle last in production. Tools built on the other premise face an uphill adoption battle at every firm that takes its own professional judgment seriously — which is to say, most of them.