Every partner can tell you which consultants they consider "good with clients." The assessment is real and usually accurate — partners who have watched a lot of engagements develop pattern recognition for consultant-client fit that is genuinely predictive. The problem is that this knowledge doesn't transfer. It lives in one person's head, surfaces only when that person is in the room, and leaves with them when they move on. Formalizing client chemistry as a scoring input is not about replacing that partner judgment — it's about making it available to the whole allocation process rather than just the meetings that specific partner attends.
What "chemistry" is actually composed of
When partners describe a consultant as a "good fit" for a client, they are typically synthesizing several different types of signals, often without decomposing them. Some of these are about technical communication style: the consultant presents clearly, adapts their register to the client's sophistication level, and doesn't over-jargon their outputs. Some are about working relationship dynamics: the consultant manages client expectations well, surfaces problems early, and doesn't generate the kind of escalations that make partner oversight taxing. Some are about prior history: there was a specific project that went well, a relationship was built, and trust was established through demonstrated delivery.
Each of these components is measurable to varying degrees. Communication style and working relationship qualities are visible in post-engagement ratings, particularly when the feedback collection instrument asks specific behavioral questions rather than just "how satisfied were you." Prior history is directly observable in engagement records. The challenge is that most firms don't systematically collect feedback at a level of specificity that captures these dimensions, so the operationalized version of chemistry has to start from whatever data actually exists.
The three-signal framework
Crewpath's client chemistry dimension is built from three signal types, applied in priority order when multiple signals are available for a given consultant-client pairing.
Post-engagement ratings are the primary quantitative signal. These are the scores and qualitative comments collected after project close. A consultant who has delivered multiple engagements for a client and consistently received strong feedback has a demonstrably positive chemistry signal for that client. The rating data is time-stamped, so it ages like any other historical signal — a strong engagement from 24 months ago contributes positively but with less weight than a strong engagement from last quarter.
Relationship flags are the primary manual override mechanism. Partners can flag a consultant-client pairing as preferred — based on informal feedback, a particularly strong working relationship, or explicit client requests — or as flagged, based on documented friction, mismatch in working style, or known client preference against deployment. Both flag types require a reason code. The reason is visible to any decision-maker reviewing the scored output, which preserves the human context that the flag is representing. A flag without a reason is not accepted by the system.
Sector proximity signals are the fallback for pairings with no direct history. A consultant who has never worked with a given client but who has strong positive ratings from clients in the same industry sector and revenue band receives a modest positive chemistry indicator. The rationale is that client engagement patterns tend to cluster by sector: the expectations, working styles, and communication norms of a mid-market private equity client are more similar to other mid-market private equity clients than to a public sector transformation client. The sector proximity signal is weak — it doesn't drive the ranking by itself — but it's more informative than treating a new pairing as having no chemistry data at all.
Where intuition outperforms formalized signals
We're not arguing that a chemistry score replaces partner judgment. There are categories of chemistry knowledge that simply do not make it into any scoring system. A partner who spent 18 months in a pre-sales relationship with a potential new client has fine-grained knowledge about that client's CEO's communication preferences, their internal political dynamics, and which consultants from past projects they've spoken warmly about. None of that is in the database. When that partner overrides the model's top recommendation because they know a specific consultant will click with the client's culture in a way the data doesn't capture, that override is correct.
The formalized signal matters most in the cases where the relevant partner is not in the room. When the originating partner delegates a staffing decision to the resource manager, or when a committee is making a decision about a client relationship primarily managed by someone who couldn't attend, the chemistry data available in the system is the only chemistry information the decision-makers have access to. Formalizing that data means it is present in every decision, not just the ones that happen to include the right partner.
Building the data over time
The chemistry dimension is the most data-dependent part of the scoring model, and it is the most likely to be weak at implementation time. Most firms discover, when they run a retroactive pilot, that they have decent feedback data for their top 15-20 clients and very sparse or inconsistent data for the remaining 60-70% of their client relationships. Consultants who worked primarily with smaller clients or on shorter engagements often have almost no feedback records at all.
This is not a reason to deprioritize the dimension. It is a reason to treat the implementation period as an opportunity to upgrade feedback collection practices in parallel with deploying the scoring model. Within 6-9 months of systematic feedback capture, the chemistry signal quality improves substantially. The retroactive pilot is useful partly as a scoring exercise and partly as a diagnostic for exactly this data gap: firms can see which client relationships have rich chemistry data and which ones are essentially invisible to the model, and can target their data collection improvements accordingly.