# The way I think about CFPs

> By Faris Aziz · published 18th Nov 2026 · topic: Community · 16 min read · tags: Conference speaking, CFPs, Talk design, Community

How I got into conference speaking, what I learned from CFPs, and how I turn real engineering problems into talks that evolve across audiences and formats.

Over the past year, I’ve had a lot of questions about CFPs. How do you get into conferences? How do you come up with a talk? What makes a proposal worth accepting? How do you know which conferences to apply to, and how many times can you give the same talk?

I’ve spent quite a bit of time thinking about this myself, both as a speaker and through organising [ZurichJS](https://zurichjs.com/). I wanted to put some of those thoughts in one place, starting with how I got into speaking and the way I think about conference programmes, then moving into how I design and evolve my talks.

For context, I attended my first conference around April 2023, gave my first public talk in February 2024, and spoke at my first conference in June 2024. I’m still learning. What follows comes from that experience, the proposals I’ve submitted, conversations with programme committees, and being on the other side of the review process.

Looking back through my submissions, the share of conferences where I received at least one acceptance went from 2 out of 11 in 2024, to 11 out of 36 in 2025, to 19 out of 41 in 2026 through 18 November. That’s roughly 18%, 31%, and 46%.

I count each conference edition once. If I submit four proposals and one gets accepted, in my opinion that counts as an acceptance.

Take these figures with a grain of salt. I tried to recover and reconcile as much data as I could from my emails, logs, and speaking history. There were also speaking opportunities I got through invitations, and those sit outside this calculation.

An acceptance still counts if I later had to decline it. Of the 32 conference acceptances I recovered, I unfortunately had to decline 13, mostly because of clashes with other conferences or logistical constraints. Sometimes, by the time an acceptance arrived, I’d already committed to being somewhere else.

## My first few opportunities

My first public talk was at a React Advanced London meetup on 22nd February 2024. I was working remotely from Switzerland for a company in London, so I wasn’t local at the time. I was on the meetup’s mailing list and received an email asking for talks. I thought I’d give it a shot.

Honestly, I’d built the whole thing up in my head a lot more than I needed to. I submitted an idea based on a little case study from my work in fintech. We’d run into backend API limitations that were making some reports slow, and I’d put a few performance optimisation hacks in place to work around them. Nothing necessarily crazy, just a problem I’d had to solve.

I didn’t even know what a CFP was at that point. I had something I could talk about, submitted the idea, and ended up flying over to deliver it.

At that meetup, I met a conference organiser and started asking what it would take to speak at a conference. I was pretty persistent about it. I wanted to figure out a way to make it happen, and there was a lot of back and forth on LinkedIn about the opportunity and the logistics.

Eventually, the deal was that I’d pay for my own flights. The hotel wasn’t originally going to be covered either, but I managed to negotiate something. It was an extremely budget hotel, but by that point I was basically thinking, you know what, let’s just make it happen.

Not everyone has the opportunity to fund something like that, and I was lucky that at the time I could. I decided that instead of going on vacation, I’d take this leap of faith on myself.

Then I was asked to deliver a full day React performance workshop alongside the talk. I’d never delivered a workshop before. I’d given one public talk, and now I had to develop enough material for a whole day. But I thought, okay, let’s take this on. If that’s what it takes to de-risk bringing me over, then so be it. I took no payment for the work.

The way I interpreted that period was almost like an intern stage. I felt like I had to pay my dues to get into that world. I was building experience, meeting people, and showing that I could deliver something useful. Later, I was able to get flights and accommodation covered and receive invitations to return.

I was fortunate that I could pursue those early opportunities. Paying for travel and taking time to prepare a workshop is a real constraint. If your company can fund it, there are good arguments for doing so. You represent the company, meet industry experts and open source maintainers, and come back with ideas and knowledge you can share with your team.

That early experience also shaped how I think about workshops. Offering a workshop alongside a talk can make the trip more viable for an organiser. The workshop can bring in revenue and help cover the cost of getting you there. Preparing one takes an enormous amount of effort, but it can be a useful way into the conference world. The great part is that a workshop built around foundational engineering practices can keep evolving for years. Even familiar topics give you plenty to explore in more depth, bring in new examples, and improve the exercises as you learn what works with different audiences. You can keep iterating on the same workshop and delivering it for several years.

## What a conference is trying to put together

Once I started applying through CFPs, one of the most useful things I learned was to look at the programme I was trying to join.

My first talk was called Data Fetching Unleashed: Next.js, React Query and their BFF. Some of the conferences I applied to already had [Dominik Dorfmeister (TKDodo)](https://bsky.app/profile/tkdodo.eu) or [Tanner Linsley](https://x.com/tannerlinsley) on the programme. In conversations with committees, the feedback was essentially that they already had a TanStack Query maintainer speaking. What would another talk with TanStack Query in the title add?

Now that we’re friends, I jokingly tell Dominik he’s the reason I got a lot of rejections. But that feedback started to make a lot of sense. A title built around a library can make two very different talks look like the same slot in the programme.

There are only so many slots to fill. At a single track conference, every session is a decision about what the entire audience will spend the next half hour watching. Imagine an event costing $100,000 to run with ten speakers. There is a lot riding on every slot. A conference with several tracks may have more flexibility to try something. But then it starts to force attendees to look at the programme and understand the opportunity cost of picking one talk over another. It’s also very much the organisers’ responsibility to craft a good schedule that makes those choices easier.

From the organiser side, you’re balancing attendee interests, sponsor expectations, what employers will support, technical depth, and keeping people engaged in the same place for a whole day. You also want to avoid people leaving halfway through, which I’ve seen happen. Sometimes people attend a conference just to get a holiday out of it, so you’re doing a lot of competing for attention. I don’t think most people want eight uninterrupted hours of dense technical material. Some sessions can go deep, some can offer a different perspective, and some can bring humour and energy.

I’ve also found talks about leadership and career development harder to get accepted. At many events, there seem to be very few slots for them, and those slots often go to people with a substantial history in that area. For me, a specific technical problem and a case study I know deeply have often been an easier starting point.

It’s only now, after just over two years of speaking at events, organising my own, and building a reputation in the industry, that I’ve started getting a couple of solid acceptances for soft talks. [WhatTheStack](https://www.faziz-dev.com/events/whatthestack-2026) and [HalfStack](https://www.faziz-dev.com/events/halfstack-vienna-2026) this year were two of those opportunities. Those talks have yet to make it to bigger stages, which is eventually the goal, but that feels like a longer ramp-up.

Organising ZurichJS Conf made those programme tradeoffs much more tangible. We analysed 434 submissions and 2,489 reviews from 15 reviewers. Some very highly rated proposals still ended up marked rejected. A rejection on its own gives you very little information about what to change.

From my conversations with organisers, sponsorship feels harder to secure, and there’s a lot of pressure towards AI content. We saw a lot of AI submissions ourselves. That makes it even more valuable to explain what your talk adds and why the subject will still matter when the event comes around.

Sponsorship budgets are tighter too. Some sponsors understandably opt for packages that include a speaking slot. Those slots help fund the conference, but they also take up space in the programme that could otherwise go to CFP submissions. For organisers, that can be a necessary part of making the event happen, and it leaves fewer slots to choose from.

## We often underestimate how many experiences can become talks

I often hear things like “I don’t have a topic”, “It’s already been talked about”, “This isn’t interesting enough”, or “I’m not up to speed with the latest stuff”. Those things matter a lot less than you might think. We often underestimate how many of our own experiences could become talks.

My most reusable talks have come from problems I worked through, decisions I had to make, and constraints that made the obvious solution difficult.

Caching is hardly a new topic. You can take an existing subject and layer it with more research, explore how it connects to other areas, and bring a different perspective to it. In my case, that meant going deeper into the networking layer, data fetching, payloads, and the architectural decisions between the backend and frontend. The talk doesn’t have to map one-to-one to your own experience. Your experience can be the starting point, and research can help you take it further.

That original data fetching talk eventually evolved into Caching, Payloads, and Other Dark Arts. The emphasis moved towards improving the experience under the constraints we actually had. Later versions went further into the full stack decisions behind that experience. I keep a version history on my website because it helps show how the content and framing have evolved. [The caching talk and its earlier versions](https://www.faziz-dev.com/talks/caching-payloads-and-other-dark-arts-optimizing-ux-in-suboptimal-conditions)

Conferences have responded well when I can take a concept and show how we applied it, what went wrong, what we changed, and what impact it had. With the appropriate approval from your employer, concrete results give the audience something to understand and the committee something to assess.

I think about it a little like a CV. You can list a responsibility, or you can explain what changed because of your work. How much did performance improve? How many users were affected? What cost or failure did you remove? The numbers need context, but they can make the value much easier to see.

I also try to choose ideas that will stand the test of time. A tool that’s exciting when the CFP opens might have disappeared by the time you get on stage six months later. AI can be part of the conversation, but I want the talk to have a foundation in engineering decisions that will remain useful as the tools change.

## Give the audience a reason to choose your talk

The title and abstract need to make sense to somebody who has no idea who you are. That became especially clear to me through anonymous reviews at ZurichJS. The proposal has to give reviewers enough to understand the session on its own.

The same applies when somebody is buying a ticket or choosing a talk. At a conference with several tracks, your session is one of a few things they could spend that time on. The title needs to express something they want to learn.

I also look at who actually attends the conference. What does the audience’s experience look like? What kinds of companies do they work for? What does the local industry look like? If the organisers have an attendee survey or a breakdown from previous editions, that can help you work out what to assume and where to go deeper.

An audience with a lot of experienced engineers may recognise different problems from one with lots of people early in their careers. People working at agencies may also approach a topic differently from people maintaining the same product for years. That changes the examples I’d choose, the context I’d explain, and which tradeoffs I’d spend time on. The country and the local industry can give you useful context, but I’d still want to understand the people coming to that particular event.

Mark Erikson’s How I Made Immer Twice as Fast: Performance Optimization in Practice is a good example. “Twice as fast” is immediately interesting, and the rest of the title tells you that you’ll hear how he approached the work. It gives you a result and a reason to listen. [Mark’s talk](https://blog.isquaredsoftware.com/presentations/2026-09-immer-perf/?view=article)

You can have fun with that too. A lot of developers enjoy games and geeking out about unusual projects. Daniel Noel-Davies’s “Hold My Beer”: From Memory Scanning to gRPC in the Pokémon World, which we had at ZurichJS Conf, combines nostalgia, curiosity, and some very concrete technical material. I want to know how those things ended up in the same talk. [The ZurichJS programme](https://conf.zurichjs.com/schedule)

That was part of the evolution of my own titles. As I got clearer on the case study and what people could take away, I could express the talk through the problem and the decisions involved. The names of the libraries became part of how I explained the solution.

## Designing an accordion talk

I’ve started calling these accordion talks. It’s a term I came up with for the way I design a talk to expand and contract around the same two or three key takeaways. I usually start with a default version of around 30 minutes, then work out how the material supports those takeaways.

For a 20 minute version, I can close the accordion a little. Which slides, examples, or detours can I remove without losing the argument? For a longer slot, I can open it up. Which decisions deserve more depth, which examples could I unpack, and where could I show more of the implementation?

It reminds me of the idea of a T shaped developer. You’re balancing breadth and depth. A longer version needs somewhere worthwhile to go deeper. A shorter version needs to keep enough substance that the audience still understands why the lesson matters.

A workshop takes that further. People need time to try things, run into problems, and ask questions. That changes what I need to prepare and how I work with the group. It also means the same area of experience can support different formats as I develop the material.

I don’t think you need a completed slide deck before submitting a proposal. I do think an outline is super important. Especially for a 20 minute slot, I want to show how I’m going to use the time and what I can realistically cover.

We saw a useful association in the ZurichJS reviews too. Among 414 proposals with at least four reviews, those with an outline averaged 2.43 out of 4, compared with 1.98 for those without one. That doesn’t establish that adding an outline caused better scores, but it matches my experience of how much easier it is to assess a proposal when I can see the planned substance.

## Build evidence that you can deliver it

Early on, doing online talks helped me build a portfolio. They gave me recordings I could share with organisers, opportunities to practise, and experience explaining the same idea to different audiences.

At the same time, my energy comes across differently remotely. I’m much more engaging in person. Getting a recording from a meetup or another accessible event gave organisers a better sense of what I’d be like on a conference stage.

There’s an entertainment element to speaking too. How you engage with the audience, explain a difficult idea, and keep people with you matters alongside the technical content. A recording can help someone make that first bet on you. After that, feedback, reputation, and an organiser’s own experience of working with you can carry forward.

I try to carry what I learn into the next version. Sometimes that means a different title. Sometimes it means cutting a section, adding a case study, or changing the depth for a particular audience. Sometimes it means taking a proposal somewhere else because the programme was already full of similar subjects.

The acceptance figures give me some context for that process. The version histories and conversations tell me more about what I actually changed. I can point to a request for a stronger React focus or deeper architectural material. I can also see where a talk was accepted and I had to decline the invitation myself. Those are different experiences to learn from.

There’s a lot of work in developing a talk and adapting proposals for different conferences. Building two or three talks you know well and submitting them to other events can make that effort go further. AI can help with a lot of the research, drafting, and adaptation, but you still need to decide what fits the audience and what you can actually deliver.

At the end of the day, the goal is not to be a [meat proxy](https://simonwillison.net/2026/Aug/3/dont-be-a-meat-proxy/), just passing AI output along. You need to read it, check it, and bring your own experience and judgement to the proposal. Simon Willison shared a useful explanation of the term.

Hope it helps! If you’ve read this far and still have questions, shoot me a message :)

## Related

- [Why it’s called ZurichJS](https://faziz-dev.com/blog/why-it-s-called-zurichjs) (On this site, 9th Mar 2026)
- [How I found a way into npmx as a React developer](https://faziz-dev.com/blog/community-open-source-and-npmx) (On this site, 3rd Mar 2026)
- [Developer Communities: What's the Secret Behind ZurichJS' Rise?](https://creators.spotify.com/pod/profile/alm74/episodes/ENG-Developer-Communities-Whats-the-Secret-Behind-ZurichJSs-Rise-e35hq84/a-ac24hqr) (Spotify, 17th Jul 2025)

Canonical: https://faziz-dev.com/blog/the-way-i-think-about-cfps
All writing: https://faziz-dev.com/blog.md

---

Faris Aziz: software engineer, speaker, ZurichJS co-founder. Geneva, Switzerland.
Site: https://faziz-dev.com · Invite to speak: https://faziz-dev.com/invite · Services: https://faziz-dev.com/services · Mentorship: https://faziz-dev.com/mentorship · Index for agents: https://faziz-dev.com/llms.txt
LinkedIn: https://www.linkedin.com/in/farisaziz12/ · GitHub: https://github.com/farisaziz12 · Bluesky: https://bsky.app/profile/farisaziz.com

