The first teaming call: what to cover in twenty minutes

Twenty minutes is enough if you spend them on the right six things

The standard first teaming call burns ten minutes on company history and five on scheduling the next call. Nothing gets decided. Two weeks later someone discovers the work share does not close, or the data rights positions are incompatible, or the engineer whose resume is already pasted into the technical volume has been committed to another bid since November. The call was pleasant. It was also a waste of two calendars. A first call has exactly one job: produce six answers that let both firms either start writing immediately or walk away with enough runway to find someone else.

We run this agenda on every first call, in both directions, whether we are the prime bringing on a specialist or the AI and data team a prime is bringing on. It takes twenty minutes because each item has a right answer format. Scope boundary is an interface, not an ambition. Work share is a number. Key personnel are names with hours. Data rights is a position, stated before anyone drafts. Exclusivity is a shape with an expiry date. Timeline is a calendar date for the sub's material, not the government's close date. Six answers, six formats, twenty minutes.

The reason to compress it is not politeness. It is that the answers are correlated. If the work share cannot close, the scope boundary has to move, which changes who the key personnel are, which changes the data rights position, which changes whether exclusivity makes sense. Discovering those dependencies one email at a time takes three weeks. Discovering them in one sitting takes twenty minutes, and either firm can say no while there is still time to bid with someone else.

The agenda, with a clock on it

Put the clock on the screen. Every item below has a target duration, and the discipline of moving on is what makes the call finish. If an item cannot be answered in its window, that is itself the answer: it goes on a short written list and gets closed by email within forty-eight hours or the teaming does not proceed.

Minutes 0 to 3: the scope boundary

The failure mode here is talking about capability. "We do AI/ML." "We do systems integration." Two firms can trade capability statements for an hour and still not know who writes which section. The fix is to force the conversation down to an interface: the specific artifact that crosses from one firm to the other, and the environment it lands in.

Good boundary statements sound like this. "You deliver a container image that exposes a REST endpoint, takes a JSON record, returns a scored record with a confidence value and a reason code, and runs in our GovCloud account under our IAM roles." Or: "We own everything from the sensor bus to the message queue. You own everything from the queue to the operator display, including the model and the evaluation suite." Or: "We write section 3.2 and 3.4 of the technical volume, you write 3.1, 3.3, and the management approach, and we integrate."

Two boundary questions get skipped and cause most of the later pain. First, who owns the evaluation? A model handed across a boundary with no agreed acceptance metric turns into an argument at delivery. Name the metric and the threshold on the first call, even roughly. Second, whose environment does integration happen in, and who pays for it? Cloud accounts, classified spaces, a lab bench with the actual hardware, a government-furnished dataset that has not been released yet. If the environment does not exist on the date the work starts, that is a schedule risk both firms need to price now.

Minutes 3 to 7: work share, as a number

Work share is where teaming conversations get vague, and it is the one item that is often not negotiable, because a statute or a regulation sets the floor. Say the number out loud on the first call. If the prime wants a sub to carry 60% of an SBIR Phase I, the deal is dead before it starts, and both sides should know that in minute five rather than week three.

Program type	Who must hold the prime position	Minimum the prime performs	Where the rule lives
SBIR Phase I	The small business concern	At least two-thirds of the research work	SBIR/STTR Policy Directive; 13 CFR 121.702
SBIR Phase II	The small business concern	At least one-half of the research work	SBIR/STTR Policy Directive; 13 CFR 121.702
STTR Phase I and II	The small business concern	At least 40%, with the research institution at 30% or more	SBIR/STTR Policy Directive; 13 CFR 121.702
Small business set-aside, services	A qualifying small business	50% of the amount paid goes to its own employees or similarly situated entities	FAR 52.219-14; 13 CFR 125.6
Unrestricted prime contract	Anyone eligible	Set by the subcontracting plan, required above $750,000	FAR 19.702; FAR 52.219-9
Commercial, state, or local work	By agreement	By agreement, written into the teaming document	The teaming agreement itself

Two things to settle alongside the percentage. Work share is measured in dollars of the research effort, not in pages of the proposal or in enthusiasm, so both firms need a rough labor split before either writes a cost volume. And on a set-aside, the similarly situated entity rule in 13 CFR 125.6 can change the arithmetic in a small business's favor, which is worth checking before anyone concludes the split does not work.

Minutes 7 to 10: key personnel, named

Ask for names, roles, percentage of time, and the window of availability. Not "our senior ML team." Not "a principal engineer to be assigned." A seat with no human in it is the single clearest signal that a firm is bidding capacity it does not have, and evaluators read it the same way. On SBIR work there is a hard rule behind this: the principal investigator must have primary employment with the small business during the project, which under the SBIR/STTR Policy Directive means more than half of the PI's time is with that firm. A prime that expects to supply the PI on an SBIR has misread the program.

Three follow-ups take thirty seconds each and are worth more than the resumes. Is this person committed to any other bid closing in the same window? What happens to the schedule if they are unavailable in month two, and who is the named backup? And are they willing to be named in the proposal, with employer and role, rather than appearing as an anonymous "senior engineer"? Our bench answers all three on the first call, because named people with stated hours are what a source selection actually scores.

If a clearance is in play, get it on the table now. Facility clearance, personnel clearance level, and whether a DD Form 254 will flow down to the subcontract. Those have lead times measured in months, and a teaming plan that assumes a clearance will materialize during the period of performance is a plan with a hole in it.

Minutes 10 to 13: data rights, before anyone drafts

Data rights get postponed because they feel like a lawyer's problem. They are an engineering problem that a lawyer writes down. The question is simple: on the day this contract ends, who can do what with the code, the models, the weights, and the data?

The vocabulary that matters on a DoD bid is in DFARS 252.227-7013 for noncommercial technical data, 252.227-7014 for noncommercial computer software, and 252.227-7018 for the SBIR data rights clause, which carries a twenty-year protection period from award under the current SBIR/STTR Policy Directive. Anything a firm developed exclusively at private expense before the contract is background intellectual property, and it stays that way only if it is identified and marked. Anything developed under the contract with government funds carries broader government rights.

The three positions to state. What each firm is bringing as background IP, listed by name, so it can go in the assertions table rather than being discovered at delivery. What each firm expects to be jointly developed, and who holds it. And whether the prime is asking for a license from the sub that is broader than the license the government is getting, which is a legitimate ask but has a price and needs to be in the teaming agreement, not assumed.

The flow-down check. A prime cannot grant the government rights it does not hold, and a sub cannot quietly deliver a component whose license forbids government use. On AI work this bites hardest on model weights and training data. If a fine-tuned model rests on a base model whose terms restrict redistribution, or on a dataset with a non-commercial license, the whole delivery is compromised. Ask the question on call one. It takes forty seconds and it has ended more teaming arrangements than price ever has.

Minutes 13 to 16: exclusivity

Exclusivity has three honest shapes, and any of them is fine as long as both sides say which one they are in. Two-way exclusive: neither firm teams with anyone else on this solicitation. One-way: the sub is exclusive to this prime, the prime may bid other configurations, or the reverse. Non-exclusive: either firm may appear on more than one proposal for the same requirement, which some agencies allow and some do not.

Three terms make an exclusivity clause fair rather than a trap. It is scoped to one named solicitation, not to a technology area or a customer for all time. It has an expiry, normally at award announcement or a fixed calendar date, whichever comes first. And it dies if the prime does not bid. A firm that holds a sub exclusive and then no-bids has taken a competitor off the field for free, and any clause without that release should be redlined on the spot.

The organizational conflict of interest question belongs here too, under FAR subpart 9.5. If either firm holds advisory or systems engineering work touching this requirement, or wrote any part of the requirement, that has to surface on the first call. It is easier to design around a conflict at the teaming stage than to explain one after a protest.

Minutes 16 to 20: timeline, backward from close

The last four minutes are pure arithmetic. Take the solicitation close date and count backward. Integration and final read: three days. Cost volume assembly with the sub's pricing loaded: five days. Technical volume integration: five days. The sub's draft material: due before that. The teaming agreement and any non-disclosure agreement signed: before the sub writes a word that matters.

On a DoD SBIR closing through the DSIP portal, that arithmetic usually puts the sub's material due about two to three weeks before the government's deadline. Say that date out loud and get agreement on it. The most common cause of a weak volume is not a weak partner. It is a good partner who was given the real deadline instead of the internal one and delivered on time to the wrong date.

Two other dates belong on the call. The question window, because technical questions to the contracting officer or the topic author close well before the solicitation does, and the answer often changes the scope boundary. And the date any letters of support, research institution commitments, or facility agreements have to be in hand, since those depend on a third party's calendar and cannot be compressed at the end.

The three questions that reveal whether a partner is real

These work in both directions. A small business should ask them of a prime; a prime should ask them of a small business. Each one is hard to answer with marketing language.

1. Show me something you built that is running today, and tell me who runs it. Not a deck, not a pilot that ended, not a demo video. A system in use, and the name of the person responsible for it now. Firms that build things answer this in one sentence. Firms that assemble slides change the subject to methodology. If the work is under an agreement that limits what can be described, the shape of the answer still tells you plenty: a real builder will describe the architecture, the data volume, and what broke in month three without naming the customer.

2. Who does the work, what fraction of their time, and will they be named? This is the personnel question asked as a character test. A firm with a real bench gives names, hours, and a yes on naming. A firm without one gives titles. Follow up with: what else are those people committed to between now and the period of performance? The answer separates a bench from a directory.

3. What happens if we do not win? The most useful question on the call, because it has no rehearsed answer. Listen for two things: whether the firm expects to be paid for pre-award work or treats it as its own investment, and whether it intends to bid the next cycle together. Firms that plan for the loss are planning a relationship. Firms that have no answer are planning a transaction, which is fine, as long as everyone knows which one this is.

What to send in the forty-eight hours after

The call produces a one-page written record, sent by whichever firm called the meeting, and confirmed by reply. Not minutes. Six answers, on one page, in the same order.

• Scope boundary stated as an interface and an environment, one paragraph.
• Work share as a percentage split, with the governing rule named.
• Key personnel by name, role, percent of time, and named backup.
• Data rights position: background IP listed, joint work identified, flow-downs confirmed.
• Exclusivity shape, scope, and expiry date.
• Timeline with the sub's material date, the question window, and third-party dates.

If all six are answered, the teaming agreement and non-disclosure agreement can be executed the same week and writing starts. If two or more are open, the honest move is to say so and set a hard date to close them. A teaming arrangement that starts with three open items rarely closes them under deadline pressure. It just carries them into the proposal, where evaluators find them.

What a bad first call looks like

It is worth naming the pattern so it is recognizable in the moment. The call opens with a capability briefing from each side. Nobody states a percentage. Personnel are described by title. Data rights are deferred to "our contracts people." Exclusivity is implied by tone and never written. The close date is mentioned once, and the sub's date never is. Everyone agrees the fit is strong and schedules a follow-up in ten days.

Nothing in that call is hostile. That is what makes it dangerous. It feels like progress and produces no decisions, and the second call has to start over because there is no written record of the first. Twenty structured minutes beat two friendly hours, and both firms get their time back.

Common questions on the first-call agenda

Frequently asked questions