How my start-up saves $100+ a month with home-grown solutions
Hi there, my name is Jen and I am building Lunch Money, a multicurrency personal finance tool for the modern-day spender.
Frugality is in my blood. Early on in life, I learned the value of a dollar from my parents and have been particularly frugal (or money-conscious) since I quit my full-time job 4 years ago.
This has permeated all aspects of my life now, as a self-proclaimed “froodie”– a frugal foodie (sure you can pay top dollar for the best food, but the real gems are when you find unfathomable value for the taste) and as the founder of a budgeting app. So it’s no surprise that in building my app, I found myself evaluating if paying for a service was worth saving the time it would take to just roll my own solution.
I’m a big believer that time is money, so if you value your time at a certain rate, then you can easily figure out if spending 1 to 2 hours engineering your own solution is worth yet another recurring monthly charge for potentially the lifetime of your product. Remember, it all adds up!
I have outlined 3 home-grown solutions which are currently saving me in total over $100 per month and I am confident these solutions can sustain Lunch Money well through the first 1000 users. As volumes increase, I will surely re-evaluate these solutions but until then, here they are!
Knowledge Base
The problem: Lots of repeated information in support requests
Since launching, I’ve been getting a steady stream of support requests and questions from users. That along with the fact that I was pushing out new features and improvements on the daily means a lot of information I was conveying would quickly be outdated.
I was also noticing a lot of the same questions. While this was a strong indicator that certain parts of the product could be more intuitive or have a walkthrough, I felt it was nearing the time for Lunch Money to have a centralized knowledge base.
The hunt for a solution
Something like Discourse would be cool to have eventually if I build a community around Lunch Money, but for now it didn’t make sense. Since our features are still evolving, I didn’t want to worry so much about outdated answers. Also, at $100 per month, the price point is unjustifiable.
I found that knowledge bases are typically a feature as part of a larger set of tools for customer support. The most popular option I could find was Zendesk, whose support docs system is an add-on, so you have to subscribe to a base plan first which in total would cost you at least $10 per month.
I also found other solutions such as Helpscout, offered at $25/month and packed with features, as well as startups using Notion as a knowledge base. Though they lack analytics and the general look and feel of a knowledge base, the price comes in lower at $4 up to $8 per month.
It was looking like it would cost on average about $10/month for a decently-featured hosted version. From afar, it was enticing given that these services were offering beautiful templates that were already fully hooked up and all you had to do was provide the content.
Deciding to host my own
I realized that all I really needed for my knowledge base was a good-looking template. All the other features that were offered were not so useful to me and would only be distractions (e.g. comments, live chat, etc). I also didn’t want to spend too much time beyond writing content to get this off the ground.
This led me to browsing Themeforest for some Jekyll or Hugo themes since I was familiar with these static-site generators and I wasn’t looking for the bloat that came with a Wordpress or Ghost site.
Lunch Money’s knowledge base, powered by Jekyll (free) template from Themeforest for one-time payment of $49
It took me 10 minutes to find a beautiful, simple template that met my basic needs and it costed a one-time fee of $49. So this template would pay for itself after just a couple of months and considering a docs site is something I would want to have for as long as the product is around, this will end up saving me a lot. I bought the template, added some basic content, pushed it to Github and deployed it for free with Netlify.
Final Cost and Savings
Set up time cost: 10 minutes to find the template, 5 minutes to deploy to Github and Netlify (excludes content-writing time)
Paid solutions:
- Notion offered at $4 - $8 per month
- Freshdesk offered at $15/month
- Helpscout offered at $25/month
Estimated savings: $15 per month or $180 per year
Cost: $49
Pays for itself in: just over 3 months
Support Requests
While researching customer support tools, I noticed they also handle support tickets that come in through email. However, at those price points, I simply couldn’t justify it. I used a service like Freshdesk exclusively at my first start-up and I can’t exactly pinpoint the value it brought from separating support requests from my normal inbox.
Here’s the thing– your first few customers are so important. They will be your champions later on. It’s imperative to give top-notch service from the very beginning. For me, this means support requests have a high priority and deserve to go to my work inbox which I check most frequently. I also don’t want to install yet another app on my phone or have yet another tab always open in Chrome.
Setting up my support email for free
To allow users to email me at support@lunchmoney.app, I hooked up the domain to Mailgun (free) to receive emails and route them appropriately to my inbox. This initially saved me the $6 it would cost to get on GSuite and have Google manage my work domain.
While I was happy with this solution for the first few months, I quickly noticed issues with relying on a free tier for something as important as my email. I was experiencing almost no deliverability to outlook.com and hotmail.com email addresses. This is because spammers also tend to use these services (Mailgun, Sendgrid) and they end up “polluting” the shared IPs that are used in these free tiers, causing them to be blacklisted by email service providers.
Eventually, I grew frustrated enough with the deliverability issues that I upgraded from Mailgun to GSuite.
Having every form of communication from my users arrive in my work inbox has actually simplified my workflow and hasn’t felt overwhelming yet. Furthermore, I use filters, tags and the snooze feature religiously, so if you have a good system, this can also help to improve your workflow. My work email is essentially my TODO list so I like to keep it clean and as close to empty as possible!
Final Cost and Savings
Set up time cost: 15 minutes to set up Mailgun and/or GSuite
Paid solutions:
- Zendesk offered at $5/month
- Front offered at $12 per month
- HiverHQ offered at $15/month
- Helpscout offered at $25/month
Estimated savings: $15 per month or $180 per year
Cost: $6/month
Pays for itself: immediately
Drip Campaigns
Let’s talk about drip campaigns. They are crucial to customer engagement and guiding your users to eventual conversion. However, many tools out there designed to help you manage your drip campaigns are overloaded with features and therefore really expensive. Furthermore, they require set up and integration since you’ll want to update the service with your customer’s traits so they can be segmented and targeted properly. Overall, I felt these tools brought a lot of extra overhead for a cost that was way too high.
So I started thinking about what it would take to implement my own simple drip campaign. I figured I could start off pretty simple and just email a user over the course of their trial 3 times: when they sign up, one week before their trial ends to offer a trial extension, and on the last day of their trial.
These were all straightforward queries to the database. I store the join dates for all users as well as a type so I know how long their trial is. I also have to hit the Stripe API to make sure that I wasn’t asking already-converted users to convert. This was another nice thing– I could query all the information directly when I needed it, instead of playing telephone between multiple services via webhooks.
Using Redis queues (free as a Heroku add-on) with Bull, I created a daily repeating worker which would, for each email template, query the database and retrieve all eligible users, double check their Stripe status and double check that they haven’t already been sent this email (safeguarding in case the queue hiccups and the job retries itself) and eventually sends the email.
I use Postmark as my email service provider. I’m a big fan of Postmark after experiencing the lowest lows with Sendgrid and Mailgun. Postmark has a rigid selection process to ensure spammers do not get on their platform and so they maintain high deliverability rates. The best part is they offer a $75 credit to bootstrapped startups which gets you 7 months of free service if you send under 10,000 emails a month.
My solution is working great for me so far as I have no complaints. I have all the open rate and CTR stats I need from Postmark and I can corroborate the data easily by querying my database to see how many actually extended their trial. I have also since added a 4th email: if you extend your trial, we send an email to follow up on how the extension is going.
The key is to make your homegrown solution robust and extensible from the beginning. I can’t see a reason yet why I would switch to a paid full-service tool anytime soon!
Final Cost and Savings
Set up time cost: 3 hours to set up queues once and write the code to power your campaign and write tests
Paid solutions:
- Drip offered at $49/month
- Intercom offered at $49/month
- Customer.io offered at $150/month
Estimated savings: $80 per month or $960 per year
Cost: 3 hours at $75/hour = $150 in engineering cost
Pays for itself in: 3 months
