Dev agency client discovery call playbook

A dev agency discovery call is a 30-minute structured conversation that decides whether a prospect is worth a proposal. The agenda is fixed: 5 minutes of context, 15 minutes of problem deep-dive, 5 minutes on budget and timeline, and 5 minutes proposing the next step. You never quote price on the first call, and you disqualify any prospect with no defined budget, no decision-maker present, or who wants a fixed quote before scope is written.

Most agencies treat the discovery call like a sales pitch. That is why their close rate is 8%. The shops that close 35% to 45% run discovery like a doctor runs intake: structured, time-boxed, willing to say "you are not a fit for us."

This playbook is the exact 30-minute script we recommend, plus the disqualification rules, the no-proposal rule, a comparison of good and bad discovery, and the questions that filter out tire-kickers in the first 10 minutes.

Why most discovery calls are a waste of pipeline

The default agency discovery call has no agenda. The founder explains their startup for 20 minutes. The agency owner nods and says "interesting, we can definitely help." A proposal goes out 4 days later. The prospect ghosts.

A real discovery call has three jobs: determine if the project is real (budget, authority, timeline), determine if the problem is one your agency actually solves, and determine the next step (paid discovery, scoping workshop, walk away). Note that "send a proposal" is not on that list. That is the single most important rule.

The 30-minute agenda

Print this. Keep it on your second monitor. Stick to it.

Minute	Block	Goal
0 to 5	Context	Confirm who is on the call, why they reached out, what triggered the search now
5 to 20	Problem deep-dive	What they have tried, why it failed, what success looks like in 90 days
20 to 25	Budget and timeline	Get a real range. Disqualify if absent.
25 to 30	Next-step proposal	Paid scoping, second call with engineering, or polite no

If you finish the problem deep-dive in 10 minutes, you do not have enough information. Push harder. If you are still in context at minute 15, the prospect is rambling and you have lost control of the call.

Minutes 0 to 5: context

Three questions. Time them.

"Who do we have on the call today, and who else makes decisions on this project?" This is the BANT check disguised as politeness. If the founder says "it is just me but I will need to run it by my co-founder and CFO," you have a multi-stakeholder deal and the proposal will take 6 weeks to close. Adjust accordingly.

"What made you reach out this week specifically?" The word "this week" matters. A vague "we have been thinking about it for a while" means low urgency. A specific trigger ("our lead developer quit on Friday," "we just closed our seed round," "the demo for our biggest customer is in 6 weeks") means a real project.

"What does the company do, in one sentence?" If they cannot answer this in under 20 seconds, the project will have scope problems. Founders who cannot describe their own business cannot write a useful spec.

Minutes 5 to 20: problem deep-dive

This is the meat of the call. Most agency owners spend 3 minutes here and 12 minutes pitching. Reverse it. The disciplined questions:

What have you already tried?

This single question filters more bad-fit prospects than anything else. The answers tell you three things at once: budget tolerance, technical literacy, and timeline realism. A founder who has tried two Upwork freelancers and a friend's nephew is telling you their budget is sub-$20k. A founder who has tried Toptal and hired one in-house engineer has real money and has learned that cheap does not work.

Why did the previous approach fail?

If the answer is "they were not technical enough" or "communication was bad," dig deeper. Often the real failure was a missing spec, not a bad engineer. You want to know if this prospect is chronically bad (every engineer has failed them) or just got unlucky.

What does success look like in 90 days?

Force them to picture it. "What is the screenshot you take in 90 days that proves this worked?" If they cannot answer, the project has no success criteria, which means it will never end. Walk away or charge a discovery fee to write the criteria for them.

What is the actual technical constraint?

Now you can ask the engineering questions. Stack, integrations, data volume, compliance, existing codebase. Do not try to answer technical questions live. Say "great question, I want to give you a real answer instead of a guess. Let me come back with our lead engineer on a second call." This protects you from over-promising.

Minutes 20 to 25: budget and timeline

The two questions that disqualify 60% of pipeline:

"What budget range have you set aside for this?" Not "what is your budget." Range. If they say "we do not have a number yet, we want to see proposals first," that is a hard disqualifier. Send them away. A prospect with no budget is a prospect with no project. The good script is: "We work with founders who have a defined range, even if it is wide, like $40k to $120k. The number does not need to be precise, but if it does not exist, we are not the right shop for you yet." This sounds harsh, but it saves 20 hours of unpaid scoping per week. The same discipline applies to mid-engagement: stricter discovery is also how you handle scope creep in agency projects before it ever starts.

"When does this need to be in production?" A specific date ("we are demoing to our board on August 12") means a real deadline. "As soon as possible" means no deadline, which means the project will drift. Push for a date. If they cannot give one, the project is not real.

If they pass both, you have a qualified prospect. If they fail either, you politely close the call.

Minutes 25 to 30: the next-step proposal

This is where you do not send a proposal. Read that again.

The next step on a discovery call is never a quote. The next step is one of three things:

1. Paid scoping session ($2k to $5k, 1 week). You write the actual spec, with stories, acceptance criteria, and a fixed-price quote at the end. The fee credits against the project if they sign.
2. Second call with engineering. For prospects with a real budget but unclear scope. Engineering scopes live on the call, founder gets a verbal ballpark.
3. Referral or walk-away. Their project is real but not for you. Refer them to a shop that fits (or to a booking platform like Cadence if they need 1 to 3 engineers on weekly terms).

Why no proposal on the first call? Every proposal you send without paid scoping is a guess. Guesses lose money. The shops that do unpaid proposals either inflate numbers to cover risk (kills win rate) or accept risk they cannot afford (kills margin).

The math: 10 discovery calls produce 6 qualified prospects, 4 buy a $3k scoping session, 3 of those buy the full project. That is a 30% close on discovery and 75% close on scoping. The "proposal to everyone" model gets 1 close from the same 10 calls and 5x the work. For agencies who want to outsource the engineering layer entirely, this is also where the white-label development services model comes in.

Disqualification rules: the hard no list

These are firm. Memorize them.

Signal	Action
No budget defined (even a range)	Disqualify on the call
Decision-maker not present and not joining a follow-up	Disqualify or reschedule with them
Wants a fixed quote before scope is written	Offer paid scoping or walk
Cannot describe the business in 60 seconds	Charge for discovery or walk
Timeline is "ASAP" with no real date	Push for a date or walk
Has been "shopping for 6 months"	Almost always tire-kicker; deprioritize
Asks for hourly rate before scope	Quote daily or weekly only, or walk

The shops that enforce these rules close at 35% to 45%. The shops that bend them close at 8% to 12% and burn 30 hours a week on dead-end proposals.

Good vs bad discovery: side-by-side

Behavior	Bad discovery	Good discovery
Talk ratio (agency vs prospect)	60/40	30/70
Agenda set at start	No	Yes, written
Budget question	Avoided as "rude"	Asked at minute 20
Technical questions answered live	Yes, often wrong	Deferred to engineering call
Proposal sent within 48 hours	Yes	No, paid scoping offered
Disqualification rate	10%	50% to 60%
Close rate on quoted projects	8% to 15%	35% to 45%
Time per won deal	25+ hours	8 to 12 hours

If your discovery looks like the left column, your sales pipeline is leaking 80% of its value before it ever gets to a proposal.

A sample script (copy this)

Opening, minute 0:

"Hey, thanks for making the time. I have us blocked for 30 minutes. The way I usually run these calls: 5 minutes on context so I understand the business, 15 on the actual problem, 5 on budget and timeline so we both know whether we are a fit, and 5 at the end to talk next steps. Does that work for you?"

Budget question, minute 20:

"Before we talk next steps, I need to ask the awkward question. What budget range have you set aside for this? It does not need to be precise, but if there is no number at all, we are probably not the right shop yet. We work best with founders who have at least a rough range."

Closing, minute 28 (if qualified):

"Based on what I have heard, this is a real project and I think we can help. I am not going to send you a proposal off this call because anything I quote right now would be a guess. What I recommend is a 1-week paid scoping engagement. $4,000, credited against the project if you sign. At the end you get a real spec, a real timeline, and a fixed quote. Want me to send the scoping agreement today?"

Closing, minute 28 (if not qualified):

"Honest answer: I do not think we are the right shop for what you need right now. You should probably either (a) define a budget range internally before talking to agencies, or (b) try a weekly-billing platform where you can book engineers without a long contract. Want me to make an intro?"

The honest "not a fit" close generates more referrals than the dishonest "let me send you a proposal" close. Founders remember agencies that told them the truth.

Where booking fits the agency model

Discovery exists because agency projects are big, multi-month commitments with custom scope. But a growing share of agency work is not actually agency-shaped. It is "we need one more engineer for 6 weeks," which is a booking problem.

When a discovery call surfaces a booking-shaped need, refer to a platform like Cadence rather than force-fit it into an agency engagement. Cadence runs on weekly billing ($500 junior, $1,000 mid, $1,500 senior, $2,000 lead), with a 48-hour free trial and no notice period. Every engineer is AI-native by default: Cursor, Claude Code, and Copilot fluency is vetted in a voice interview before they unlock bookings. Agencies in the Cadence partner program earn 10% recurring on every referred booking, for as long as the founder stays on the platform. The discovery call becomes a routing decision, not a binary win-or-lose.

What to do next

1. Print the 30-minute agenda. Run your next 10 discovery calls against it without skipping blocks.
2. Track close rate before and after. If you are not at 30%+ within 20 calls, the problem is your disqualification discipline, not the script.
3. Add paid scoping as your only first-step offering. Stop writing unpaid proposals this week.
4. Set up a referral routing rule for booking-shaped leads (1 engineer, weekly terms, short duration). Send them somewhere they fit.

For the upstream pipeline question (how to attract the right discovery calls in the first place), the playbook on dev agency niche positioning covers how vertical focus changes the close rate before the call even happens. For the contract layer (what the paid scoping agreement looks like), see the dev agency contract templates breakdown, which has the exact MSA and SOW structures we recommend. And once the project closes, the dev agency invoicing best practices guide covers how to bill scoping fees and milestones without margin leakage.

Running discovery calls but losing 40% to bookings-shaped projects you cannot serve? Route those founders through the Cadence partner program and earn 10% recurring on every booking, with no extra sales work.

FAQ

How long should a dev agency discovery call be?

30 minutes is the ceiling. Anything longer signals you do not have an agenda, and the prospect will treat the next call as free consulting. If you genuinely need more, charge for a paid discovery session.

Should I send a proposal after the first discovery call?

No. The shops that close 35% to 45% only quote after paid scoping. Every unpaid proposal is a guess, and guesses either lose money or lose the deal. Offer a paid scoping session instead.

What is a good close rate for dev agency discovery calls?

8% to 15% is typical for shops that send unpaid proposals to every lead. 30% to 45% is achievable for shops with strict disqualification rules and paid-scoping-only quotes. The difference is mostly disqualification discipline.

How do I ask about budget without losing the deal?

Frame it as fit, not interrogation. "We work best with founders who have at least a rough range. If there is no number, we are probably not the right shop yet." Prospects with real budgets respect this; prospects without budgets self-disqualify.

When should I refer a prospect to a booking platform instead?

When the need is one to three engineers, weekly or short-term, with no fixed scope. Booking platforms like Cadence are built for this shape of work, and you can earn referral commission instead of losing the lead.