Would you answer two questions before sending a cold request?
I'm building Portunio, an AI-first email experience with the agent on the sender's side. The design question I'm testing is how much clarification is useful before it becomes a barrier to reaching someone.
Here's an illustrative example, not a customer result:
A sender writes: ‘I'd love to collaborate. Can we jump on a call?’
Instead of polishing that sentence, the agent could ask:
- What would you like to work on together?
- What can you offer, and what decision do you need from this person?
The sender might then choose to say: ‘I'm proposing a guest workshop on onboarding for your founder community. I can bring a 20-minute exercise and a sample agenda. Would you like to see the agenda before deciding?’
That last version only helps if the details actually came from the sender. Inventing a convincing offer would make the request worse. The sender needs to review it, and the recipient still decides whether to engage.
The tradeoff: the recipient gets more context, but the sender does more work. Someone with a valuable but half-formed idea might leave. I'm interested in measuring both the time spent clarifying and the follow-up questions still needed afterward. I don't have outcome data proving this works yet.
If you regularly send or receive collaboration requests, would you tolerate those two questions? What would make you abandon the flow?
You can preview a request at https://portunio.com/ramikhalil (please label a hypothetical example ‘pilot test’). The waiting list is at https://portunio.com. Feedback on the tradeoff is welcome even if you don't sign up.
I actually think adding a tiny amount of friction can improve the quality of interactions if the questions are chosen carefully.
The problem with completely frictionless cold outreach is that it encourages people to send low-effort messages to hundreds of people. Asking two simple questions forces the sender to think about why they’re contacting that person and what they actually want.
At the same time, I wouldn’t make the process too complicated because legitimate users will also leave if it feels like filling out an application. Two short questions sounds like an interesting middle ground between open DMs and completely closed inboxes.
The problem you're solving is real. Most cold requests fail not because the ask is wrong but because the sender never worked out what they actually wanted before hitting send. Two questions that force that clarity would filter most of the noise before it reaches the recipient.
The abandonment question is the interesting one. I'd guess the people who leave at those two clarifying questions are exactly the people whose requests weren't worth sending. That might be a feature, not a leak.
The thing I'd track is completion rate by sender type. Adding steps that feel like homework might shift usage toward power users who already have a clear ask, and away from casual senders who need the guidance most. Which group do you want to optimise for?
That is a brilliant UX experiment, Rami! Testing Portunio by placing an AI agent on the sender's side to ask clarifying questions before a cold request goes out is a sharp way to tackle vague messages like "can we jump on a call?". Forcing senders to articulate what they want to work on and what value they offer trades a bit of friction for dramatically higher signal-to-noise ratio. How are you measuring the drop-off rate against the quality of the final requests to find that sweet spot between friction and conversion?
Excellente approche, Rami ! Tester Portunio en faisant en sorte qu'un agent IA du côté de l'expéditeur pose des questions de clarification avant d'envoyer une demande de prospection est une expérimentation UX fascinante. Forcer les expéditeurs à définir ce qu'ils souhaitent accomplir ensemble et la valeur qu'ils peuvent offrir évite les messages vagues et les transforme en propositions concrètes. Comment mesurez-vous l'abandon des utilisateurs dans le parcours lorsqu'ils sont confrontés à ces questions supplémentaires ?
I'd answer two questions if they are short and I can skip. The example is the real test though. If the agent invents a workshop I never offered, that email is worse than the lazy jump on a call. I already ignore those. For me the only useful check is what the ask actually is, and what the other person gets. Past that I just close the composer.
That boundary is useful: clarify the ask and the value, then stop. In the workshop example, the workshop and agenda must come from the sender; otherwise the agent has created a commitment they never made. I’d test the shorter flow you describe: keep any answer already present, ask only for the missing piece, and let “not sure yet” remain visible instead of filling it with a polished guess. I’m treating the skip option as feedback to test, not claiming it is already implemented. Would a single combined question feel lighter than two separate steps?
From the receiving end: the two-question filter is essentially what separates outreach I read from outreach I delete. We get almost no inbound — four organic visits a month on our SEO tool — so every message matters, and the ones that work all have something specific the sender figured out before writing.
The distinction Rami made between "decided this was not a fit" and "wanted to send but the questions got in the way" is the right measurement. In community threads I have seen the same pattern: a comment that leads with "I noticed your site has X problem, here is what I would check" gets a reply. "Great product, let me know if you want to collaborate" gets nothing.
The risk is the AI making the specificity look authentic when it is not. If the agent fills in a plausible-sounding offer the sender did not actually think through, the first reply will expose it.
The first reply exposing invented specificity is a useful failure test. I’d separate three things in the review: what the sender actually observed, what they are proposing, and what remains an assumption. “Your site has X problem” should need a concrete observation; if they haven’t checked, the wording should say so rather than imply an audit happened. For Portunio’s pilot, I’d count corrections to AI-added claims alongside completion and recipient follow-ups. A smoother message that requires the sender to walk back a promise is a failure, even if it gets a reply.
The tradeoff is really send friction vs. less back-and-forth. Have you tested how often senders answer both questions versus abandoning the request?
Not with a reliable completion-versus-abandonment result yet. We also found a gap in our visit tracking today, so I can't honestly use those numbers as evidence. For a small observed pilot, I'd record where each sender stops, time spent, and whether they return; then ask the recipient how many clarifying follow-ups they still need. I'd keep 'decided this wasn't a fit' separate from 'wanted to send but the questions got in the way.' A higher completion rate alone wouldn't prove the questions helped. Have you found a useful way to distinguish those two reasons for stopping?
That distinction is exactly what I’d be interested in digging into. If you’re open to it, what’s the best email to reach you on?
Happy to dig into it. You can reach me through my Portunio page: https://portunio.com/ramikhalil — mention our Indie Hackers discussion so I have the context. If you have an anonymized example, include that too. And if the clarification step itself gets in your way, tell me here; that would be useful feedback for exactly what we're discussing.
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.
This comment was deleted 21 hours ago
I've been on both sides of this — sending outreach and receiving it.
One pattern that works without adding much friction: ask the sender to state (1) the outcome they want and (2) why this person specifically. That filters "can we jump on a call?" spam while still letting a real opportunity through.
Where it gets harmful: if the gate feels like homework before the recipient even knows the topic. Keep it to one breath — a sentence, not a form. Recruiters especially feel this tension: personalization helps, but candidates bounce when every first touch asks them to fill something out.
Curious how you're measuring whether the clarification step increases reply quality vs just reducing volume.
The recruiter example makes the cost of the extra step concrete. I don't have evidence of improved reply quality yet. The test I'd run is whether the recipient can identify the proposed outcome, why it fits them, and the next decision without another clarification message. I'd report that alongside sender time and abandonment, including people who never finish, rather than celebrate fewer messages reaching the inbox. Your 'one breath' suggestion is useful: extract what's already in the first message and only ask about a missing detail. If you're open to trying the preview linked above with an anonymized example, the most useful feedback would be the exact question that feels like unnecessary homework. No need to submit a real request.
The useful guardrail is making the sender state the outcome in one sentence before the rewrite. If they can’t say what they want and why this person, the request probably isn’t ready to send.
I like that as the first thing to clarify. I'd be careful about turning it into a hard gate, though: someone with a useful but half-formed idea might need help finding that sentence. The boundary I'd want is that the sender supplies the substance and confirms the outcome; the agent shouldn't invent a reason this person is a fit. Your suggestion gives us a simple first check before asking anything else.