What an agency loses when a senior BD person leaves, and how to encode it first.

01The cost no one is measuring

Ask an agency CFO what the shop spends on rent and they will answer to the dollar. Ask what it spends pursuing new business and the answer is usually a shrug, because pursuit spend hides inside salaries. It is senior people, at senior rates, spending nights and weekends on RFP responses, capabilities decks, and pitch theater.

The math is not hard, it is just never done. Run realistic mid-size agency inputs through a public model and the number lands between roughly $210,000 and $420,000 a year in senior time:

• 10 pursuits a quarter at 35 hours each, at a $1,200 loaded senior day rate, computes to $210,000 a year.
• 14 pursuits a quarter at 50 hours each, at the same rate, computes to $420,000 a year.

Those links open the Pursuit Capacity Model with the inputs pre-filled: change any number and the result recomputes in front of you. This document quotes no statistic it cannot show the arithmetic for.

That annual spend is not just a cost. It is an investment that compounds into a capability: the agency gets better at pitching with every pursuit it runs. The problem is where that capability accumulates.

02Where pitch knowledge actually lives

After a few years of pursuits, an agency's pitch capability is real and valuable. Very little of it is written down. It lives in four places, and three of them are a person:

• Proof recall. Knowing which case study wins which kind of pursuit, what the attendance figure actually was, which client will take a reference call, and which deck from 2023 had the slide that lands. The folder exists; the map of the folder is in someone's head.
• Voice. The way the agency argues: what it leads with, what it refuses to say, how confident it sounds on paper. New writers imitate it badly for months.
• Pursuit judgment. Which RFPs are winnable, which are column fodder for a decided buyer, what a realistic budget band looks like. This is the difference between a 20 percent win rate and a 12 percent win rate at the same writing quality.
• The record. Old proposals, decks, and debriefs. This is the only one that survives a departure by default, and it is the least useful of the four without the other three to interpret it.

When the person carrying the first three leaves (and agency BD tenure is short; turnover in these roles is a structural fact of the industry, not a management failure), the agency does not lose its archive. It loses the ability to use its archive.

03The half-life: how the knowledge decays

The loss is not felt on the day of the goodbye party. It surfaces on a delay, which is exactly why it goes unmeasured. The pattern agency operators describe is consistent enough to model as three phases:

We call this the pitch knowledge half-life because the loss compounds quietly: each departed senior person takes a share of the operating knowledge, and what remains is diluted through imitation by people who learned it secondhand. An agency that has lost two BD leads in four years is often running on a photocopy of a photocopy of its best thinking.

04The BD Departure Risk Calculator

Here is the exposure, priced. Three inputs, two assumptions, formulas below. Start from your annual pursuit spend (get your real figure from the Pursuit Capacity Model; the default here is the model's mid-size default).

Read it plainly: the model treats the knowledge share carried by your most senior BD person as capability you paid to build, and prices a departure as re-creating that capability over the rebuild window. It deliberately claims nothing about lost revenue or win rates. If you believe those effects exist (most operators do), the true number is higher than the one on your screen.

05A worked scenario

The point of the scenario is not the exact dollar figure (change the sliders; it is your model now). The point is that the departure cost is the same order of magnitude as a third of the annual pursuit budget, it recurs with every senior departure, and almost no agency carries it as a line item anywhere.

06The encode-first alternative

The knowledge does not have to live in a person. Each of the four kinds of pitch knowledge can be encoded into a system the whole shop shares, and the encoding is most cheaply done while the people who hold the knowledge are still in the building:

• Proof recall becomes a structured case study library: every project with its real, verified numbers, retrievable by the shape of the pursuit rather than by remembering which folder it sleeps in.
• Voice becomes an explicit register the drafting system writes in, learned from the agency's best shipped work rather than imitated from memory.
• Pursuit judgment becomes a go/no-go discipline with recorded reasons, and a win/loss loop that turns every outcome into a rule the next pursuit applies automatically.
• The record stops being an archive and becomes an engine: claims carry verification status and freshness, so nothing stale or unsupported ships, and every pursuit adds to the system instead of to one person's memory.

Encoded this way, the half-life breaks. A departure still costs you a colleague; it no longer costs you the pitch brain. The masked quarter has nothing to mask, the dip has nothing to dip into, and the new lead's first week includes a productive draft grounded in verified proof, because the knowledge was never only in the chair that emptied.

This is the design brief Pitch Box was built against: an operating system for agency new business where the knowledge base, the case study library, the voice, the review gates, and the win/loss learning live at the level of the institution. Flat pricing with unlimited seats exists for the same reason, because knowledge systems fail when adding a colleague to the system costs money.

07Method notes, and what this document does not claim

• Every dollar figure above is computed from stated inputs through public formulas (this page and the Pursuit Capacity Model). Nothing is quoted from a private dataset.
• The 9-to-12-month rebuild window and the knowledge-concentration default are labeled operator-reported assumptions, adjustable in the calculator. If your experience says different numbers, use yours; the structure of the argument survives.
• No win-rate improvement is claimed anywhere in this document. Vendor win-rate statistics without methodology have damaged the credibility of this category; we would rather publish a smaller, auditable number than a larger, naked one.
• The worked scenario is modeled and labeled as such. Measured before-and-after figures from pilot agencies will be published with methodology when the cycles complete.

08Common questions