Cold Emailing Developers
At the beginning of this year, I wrote my philosophy around sending cold emails. I’ve subsequently sent out around a thousand emails and received replies from nearly 100, at around a 9% reply rate and zero unsubscribes, and most importantly, the majority thanked me for reaching out.
Is 9%…good? Yes. Developers are notoriously difficult: The most I’ve ever seen at Sentry is 3% and that required an immense amount of targeting and focused on our largest customers. This is my current process.
Write down the problem you’re looking to solve with a survey. Surveys are not good for broad exploration. Open ended questions lead to open ended conclusions. Most questions can be broken down into parts that can be answered with analysis versus conversation.
Of my three laws, the one that’s trickiest for most people is #2 State my reason. I’d actually like to clarify it: Give clear reasons on why they should reply. Who are you? Why are you reaching out? Why to them specifically? How much effort will it be for me? What are you going to do with my feedback?
Explicitly writing down your reasons will be valuable to not only your recipients, but also you. I see people ask abstract questions like, “Why does this segment have such a high churn rate?” but asking a customer, “Why did you churn?” is a terrible idea. Vague questions get vague answers and even more damning, poor reply rates. Being transparent with your users forces you to have clarity. If I ask this question, what kind of replies do I expect? Will I be able to act on their feedback in meaningful ways?
Determine the audience best suited to give you qualified answers. I send emails directly from my inbox, so I can’t cast too wide a net: the replies eat directly into my time. I write custom queries for targeting users with the exact behavior I’m interested in.
“But Eric, there are automation tools for this? Why not just blast 10,000 users with a multiple choice survey from email@example.com? Scale!”
First, marketing tools rarely have the targeting capacity I need. “Users that were active for 1-2 months and then deactivated within the last 30 days” is pretty tricky. I could just target inactive users, but suddenly, I’m sending 20,000 emails and in reality, I don’t care about most of their responses. Before asking any question, it’s always worth asking yourself, Does this person know the answer?
Second, I can’t really speak to this because I haven’t tried it myself. It’s a personal choice: I hate marketing emails, but I talking with people. Scale is a wonderful thing when you’re enhancing the lives and productivity of your users. It’s not great when you’re asking for their time.
a. Subject Line
Subject lines are basically headlines, except your goal isn’t clickthrough rates, but meaningful dialogue. Often, you’ll get replies that you don’t understand and you’ll reply back, asking for clarification. A well formed email elicits relationship-building, not metric optimization.
If you want to see your competition, go to your Gmail Promotions tab or LinkedIn messages. Don’t compete with marketers or recruiters, you’ll never win because it’s not your full time job. Offer up your unique strengths. Most of the world pays hundreds of dollars to access people like you. You’re offering that up for free.
- Don’t templatize the user’s name
- Don’t introduce yourself (the From field already has your name)
- Don’t use exclaimation marks
- Don’t use a CTA
- Be suspicious of adjectives
b. Email Body
When composing the body, your content and tone should directly match who you are (particularly your role in the company) and who the audience is (how they use the product).
This is especially important for your most valuable kind of feedback: feedback from churned customers. Unlike happy customers (who just love your product) or disgruntled customers (the quintessential “squeaky wheel”), churned customers (either from a revenue or active perspective) have real pain that you are not address and often not hearing. Surveys have a strong tendency to bias on the first two segments, and reply rates below 1% increase the likelihood your sample doesn’t not represent the overall population.
You can see my evolution from my first survey of our paid customer base…
Hi, I'm Eric, product/engineer at Sentry. We're working on our product roadmap for the next year, and we're looking to get customer feedback to help shape it. Specifically, I'm hoping to learn how you decided which plan to buy. You're currently on the Small plan. Why did you choose Small for your business? Do you feel like you're getting value for the price? Thanks for taking the time, Eric P.S. This isn't an automated email, so unless you reply, there won't be any followups.
…to the second survey of our recently inactive customer base (sidenote, only 2% of replies redeemed the promo code, so it wasn’t much of a carrot)…
Hi, I’m Eric, engineer/product at Sentry. I’m emailing because you’ve recently setup your Sentry org , but haven’t used us much in the last month. We're trying to figure out where we can improve, so I’m starting conversations with customers like you. Do you plan on still using Sentry? Where has it missed your expectations? User journeys rarely fit into boxes, so we're not doing the normal 20-question form survey—just reply directly. In appreciation, I’d be happy to grant you a $50 Sentry credit. While we can’t fix everything, I’ll be personally walking through all feedback with our founder/CEO as we build our roadmap for the second half of the year. Thank you for your time, Eric P.S. This isn't an automated email. Unless you reply, there won't be any followups.
In the second case, I’m much more clear about what problem we’re trying to address and what we plan on doing with their replies.
“But Eric, I don’t have access to the CEO”, you might point out, “I can’t promise them the CEOs eyes on their response.” That’s not the point. I’ve reached out to people about individual features I built asking for feedback and gotten even higher response rates. People love helping startups build better products. People hate providing feedback that ends up in some dusty database, to be GROUP BY’d some disinterested analyst months later. Show that their feedback is part of the former, not the latter. Ideally, you’d even have some public track record of all the product features that have come from user feedback, but I don’t know any company that has that.
I use YAMM. It’s simple, sends from my true email, and costs $2 a month. You’re limited to sending 1,500 emails a day but…if you’re sending more than that, go back to step 2.
This part is tedious (and the main reason people go with marketing automation). So far, my best idea is extracting responses into a database. This is especially true because it’s useful to give a set of summary statistics alongside each user response. I plan on tinkering with this more in the future.