James F. O’Brien is a professor of computer science at UC Berkeley and has published numerous studies on destructive modeling, earning an Academy Award for his pioneering work in movies and games.
Surgeons and military leaders use checklists to ensure consistently high levels of quality in mission-critical work. So why don’t we use them more in software?
We can. And should. Here’s how –
In my job at JetBridge, I work with some of the most experienced software development managers and some of the most demanding enterprise clients around the World. Over the years I’ve developed a checklist for releasing new website builds that will save you time and potential embarrassment. Use this consistently and your coworkers and managers will think you’re a “rock star.”
But before we jump into the checklist I assume that you have done all the development work in terms of the website functionality and its content, you have reviewed all the pages yourself or with the client and your next big step is releasing it to the world.
1. Set up a staging environment
Whether you want to test a new feature on the website, show it to the client, or run performance tests, you should have a staging environment set up. A stage environment is an environment for testing that exactly resembles a production environment.
2. Set up monitoring
Whether your hosting provider drops the ball or you accidentally publish a change that breaks one of your pages or the whole website, setting up an uptime monitoring tool will help you get notified about the issue by Text/Email/Slack message in a timely fashion. Some popular services include: sentry.io, uptimerobot.com, statuscake.com.
3. Compress images
To optimize website load time and improve its SEO, optimize the images used on the website. The primary goal of formatting your images is to find the balance between the lowest file size and acceptable quality. You can use tools like imageoptim.com for the optimization of your images.
4. Run the Google Lighthouse test and address reported issues
Lighthouse from Google is a tool that audits load times, accessibility and search engine optimization of web pages. It includes the ability to test progressive web applications for compliance with standards and best practices. Make sure to run the Lighthouse on a deployed environment (dev/stage) and not only locally.
5. Set up a 404 page
What happens if the user goes to a non-existing page accidentally or because of a broken link? Will the user see this unfriendly message?
Take some time to design a nice 404 page that will allow the user to go the main page of the website:
6. Set up a favicon
Favicon helps your website appear more professional and credible and can be found among all the other tabs more easily.
To add a favicon you can create a .png image and then use one of the following snippets between the <head> tags of your static HTML documents:
<link rel="icon" type="image/png" href="/favicon.png" />
7. Set up Meta Titles, Meta Description and Social Preview Links
Titles help search engines to serve the right results to the user. Moreover, they display the page title to the users in search results so that they can determine whether the page is something they are looking for. As with the favicon, page titles help the user navigate through the open tabs.
Meta description displays the description of your page in search results and while sharing your page on social media.
To add meta tags, describe metadata within HTML document:
<head> <title>HTML Elements Reference</title> <meta name="description" content="Free Web tutorials"> </head>
Setting up Social Preview links:
Facebook uses <meta> tags leveraging the Open Graph protocol, a classification system for web pages that extends beyond those <meta> tags already defined in HTML5. A complete list of <meta> tags available can be found at the Open Graph Web site. There are so many from which to choose that it can be somewhat intimidating, but only four are actually required:
<meta property="og:title" content="European Travel Destinations"> <meta property="og:description" content="Offering tour packages for individuals or groups."> <meta property="og:image" content="<http://euro-travel-example.com/thumbnail.jpg>"> <meta property="og:url" content="<http://euro-travel-example.com/index.htm>"
Twitter has its own <meta> tags that are similar to the Open Graph protocol, but uses the “twitter” prefix instead of “og”. As with Facebook, only a few are required:
<meta name="twitter:title" content="European Travel Destinations "> <meta name="twitter:description" content=" Offering tour packages for individuals or groups."> <meta name="twitter:image" content=" <http://euro-travel-example.com/thumbnail.jpg>"> <meta name="twitter:card" content="summary_large_image">
8. Create a robots.txt file
robots.txt contains instructions for search engine crawlers regarding the pages that are allowed to be crawled. It’s useful if you want to exclude certain pages from crawling. The presence of the robots.txt file positively influences your SEO. You should put your robots.txt at the root of the website host to which it applies. More about robots.txt: https://moz.com/learn/seo/robotstxt
9. Set up XML sitemaps/HTML sitemap
A sitemap is a file where you provide information about the pages, videos, and other files on your site, and the relationships between them. Search engines like Google read this file to more intelligently crawl your site. A sitemap tells Google which pages and files you think are important in your site, and also provides valuable information about these files: for example, for pages, when the page was last updated, how often the page is changed, and any alternate language versions of a page.
Guide on building and submitting a sitemap: https://support.google.com/webmasters/answer/183668?hl=en
10. Configure active environment variables for the production build
Make sure that you have configured the environment variables (payment service provider publishable keys, backend endpoint URL, etc.) for your build. Remember to never store any secret keys on GitHub due to security reasons. You can use a library like dotenv package to configure you environment variables.
11. Test the website in different browsers, ensuring responsive design on varied devices
Use a staging environment that you’ve set up to do testing on different devices. Make sure that all the functionality of your website is working, make sure you don’t have broken links. Keep in mind that your pages may behave differently on mobile as the interaction on mobile devices is done using touches (not clicks).
You can use a test from Google to test if a particular page is mobile-friendly: https://search.google.com/test/mobile-friendly
I strongly recommend you to use real mobile devices to test the website interactivity instead of just using browsers’ developer tools.
12. Set-up a domain/transfer a domain to your hosting provider/set the nameservers to point to your hosting provider
If your domain registrar is different from your hosting provider – you will have to perform a domain transfer. Migrating a domain to AWS Route53:
Alternatively, instead of transferring the domain, you can make your hosting provider the DNS service for a domain that’s in use. Instructions on Making Route 53 the DNS service for a domain that’s in use for AWS:
13. Set up a CDN (Content Delivery Network)
A CDN is a really fast way to deliver static content from your website or mobile application to people with regional caching. A CDN is made up of a network of servers (“points of presence” or POPs) in locations all over the world.
Setting up CDN using
AWS Cloudfront (S3): https://aws.amazon.com/cloudfront/getting-started/S3/
14. Set Up an SSL certificate
An SSL certificate encrypts the data that goes from a user’s computer to the target website and back. Every time a user enters information into your site, SSL makes sure it can securely travel from their browser to your web server.
Installing SSL on Cloudfront: https://aws.amazon.com/premiumsupport/knowledge-center/install-ssl-cloudfront/
Getting an SSL certificate on Letsencrypt: https://letsencrypt.org/getting-started/
15. Configure payment processing
If your website accepts payments, you should have a Payment Service Provider configured on your website. Make sure to set up publishable keys for your PSP before deploying the website to production.
You may want to extend this list with other items specific to the industry you work in or business processes of your company. Happy shipping 🚢!
John Kim (CEO) firstname.lastname@example.org
As many say costs for software developers have soared across the U.S. and Western Europe, outsourcing to arbitrage labor costs has not just become popular but also increasingly necessary.
To build two SaaS companies in the Bay Area, I’ve leveraged offshore development teams to a degree of success. I also had a chance to be a marketing consultant for a leading online marketplace for developers and am the CEO of an outsourcing company, so I’ve become intimately familiar with this industry over my career.
There are many talented, ethical agencies and developers across the world; however, just like in any market, extensive due diligence by the client is a must. Before hiring outsourcers to build your app or hiring recruiters to source offshore labor, here are three insider observations you might consider:
1. Outsourcers, outstaffers and recruiters commonly portray interns as juniors, juniors as mid-levels and mid-levels as seniors. Because this practice has become so widespread, I’ve found there is market pressure on many good vendors to sell their staff this way. So clients should manage their projects intensely on a daily basis to ensure consistency and quality.
2. Failure rates for apps built by outsourcers can be incredibly high. While there is no published data yet, I’ve seen very few examples of outsourcers who have built and launched a commercially successful product on behalf of a client as a turnkey service.
3. I’ve seen some recruiters start to offer research and development (R&D) center services to clients to compete with outsourcers. This new trend of not only sourcing the developers but operating the office and administrative functions on behalf of the client have additional fees. Over time, this could reduce labor costs for the clients, but clients need to make a bigger upfront commitment.
Given the state of the current market, I suggest implementing these best practices before engaging any outsourcing, outstaffing or recruiter firm:
1. For outsourcers building the entire turnkey app for you, ask for the references of at least two software applications they have launched successfully. Code is only one part of what makes a software application successful, so if the response is “we have none” but you still like the firm, you can engage your product or project manager to manage the outsourced staff directly.
Also, ask the references how the outsourcer accommodated change requests. Your app’s scope or project is likely to change multiple times. You want a partner who is adaptive, but being adaptive in IT can be as much about the outsourcer’s culture as it is about their process.
2. If you’re looking to outstaff to add to your existing product and team through recruiters or R&D centers, you may want to intensely focus on the interview questions and test tasks assigned to the candidates. For example, when I ask a candidate, “Why do you want to work for us?” and the answer is something akin to “for more money” and not “I want to work for a ‘product’ company,” that developer never lasts with us for more than a year.
I’ve also learned to ask what local conferences or Meetups the candidates belong to. Voluntary self-education is one of the leading indicators I’ve seen in how much value a developer will bring to our projects over the long run.
But by far, the “marquee” question I ask every developer is, “Have you contributed to any open-source projects?” While not every top developer contributes to open-source projects, I’ve noticed that many open-source contributors are top developers.
It may be harder to find developers who contribute to open-source projects in emerging economies, according to a GitHub open-source contribution map, but we’ve found this is the single biggest determinant (other than cultural fit) of whether a developer will produce at an elite level.
3. Go through a trial period without firm commitments, and if the developers are to your liking, prepare to offer a term commitment of a year or more to the vendor. In my experience, outsourcers and outstaffers often have their developers sitting “on the bench” some of the time, so term commitments mean they don’t have to suffer a margin loss — and you could negotiate some of this savings in reduced fees.
As a negotiating tactic, you can also offer to prepay their wages quarterly or every six months.
4. Look for providers that focus on both a vertical and a software language. I believe the vendors that are efficient with your time and money are so because they focus on a specific vertical — like fintech or mobile games — and have expertise in one or two languages.
I recommend avoiding shops that claim to be experts in “Java, Python, C++, and AI/ML” and that spam you with faces of attractive females representing their agency’s expertise. These are often bottom-of-the-barrel operators. I’ve found that most of the good shops specialize in a vertical and specialize in a language, and they have plenty of work from referrals — so they don’t need to spam anyone.
5. Negotiate the monthly wages and the buyout fee if you decide to make the developers a part of your permanent team. Outsourcers and outstaffers need to charge more than what they pay the developers to make reasonable net margins.
Sales costs, office space, recruiting and bench-time contribute to overhead. In my experience, average buyout fees range from one month of wages to one year, so negotiate — as the market prices range drastically.
As Bloomberg suggests, the demand for talented and experienced tech talent in the U.S. continues to rise, and the BLS reports that unemployment for the professional, scientific, and technical services sector was at just 2.5% in December 2018 — so sourcing labor from other countries may become a necessity, even for Series B- and C-funded startups. Understanding nuances of the regions where you hire and following some best practices should save you a lot of time and money — and maybe some midnight migraines as well.
Although it might be tempting to quickly file a ticket that will scope multiple work segments and give it a short description (that is clear to you and maybe a few teammates), later such tasks may affect negatively on the entire team’s productivity.
Divide Et Impera
Here’s a blogpost from guys who do project management tools for living, on why it’s important to properly size the task and break it into smaller chunks.
A summarized list of reasons:
- You can’t remember everything. So while working on big pieces of work sometimes it’s easy to lose focus and accidentally skip something important.
- Smaller tasks are way easier to estimate and describe clearly.
- Smaller tasks produces less code that is easier to review effectively. Also frequent feedbacks help us to keep ourselves on a right track and improve implementation in the fly.
- Completing a task feels good. The more often your work is delivered, the more often your brain releases dopamine.
Imagine working on some food delivery service and creating a task like “Implement ‘become a deliverer’ flow on front-end”. It might be worth breaking it down into pieces like:
- “Create base UI components”
- “Extend front-end services with ‘become deliverer’ flow API calls with unit tests”
- “Implement ‘Personal Info’ step layout with Cypress tests”
- “Implement ‘Transport’ step …”
It is also a good practice to create SPIKE tickets if you feel uncertain of possible implementation ways. Investigation results are good to be discussed with team members afterwards.
Reasonable Ticket Structure
Here’s a list of things that speed up developer’s understanding of the problem ×100 times:
- Acceptance criteria – an answer to a question “What should we have/know/be able to do when work on this task is done?”.
- Clear steps to reproduce the issue if it’s a bug-fix ticket (imagine yourself reproducing it at app’s initial state).
- Screenshots/GIFs/video attachments of known front-end bugs.
- If you have any guesses where the bug might be fixed, post this info in ticket description.
- Links to designs really come in handy for layout implementation tasks.
- Links to open/merged PRs, that would help quickly find code of changes you made.
- What the motivation is. Explain the problem that you want to solve and when appropriate, present your idea for a solution to the problem but leave room for discussion or alternate approaches.
How About Pull Requests?
Main purpose of a PR is not just to merge your changes into
master, but to give other people a good understanding of your work’s context and reasons behind your changes. Perception of all that stuff is way easier when reviewing a short PR that scopes some single logical concept (or at least not too many of them).
Short PRs are good because they:
- Speed up code review process and consequently the product development.
- Reduce bugs introduction into codebase.
- Don’t block other developers and reduce chances of conflicts between branches.
When reading through your finished work try asking yourself “Will other people easily understand what and why is happening here? Will they not lose focus and understand all the connections between changes made?”.
If not, there’s absolutely nothing wrong with breaking your PR into bunch of smaller ones that would be easier to digest.
A couple of great ways to make PRs enjoyable to read through:
- Add a link to the ticket where changes are requested.
- Give it a sufficient description of what changes were made to achieve requested result.
- Add screenshots, GIFs or videos of layout implementation results.
- Provide steps for local testing, if appropriate.
- Add comments to your code that you expect to be unclear for others for some reason (although of course we aim to write self-descriptive code needing no explanations).
Try thinking of tickets and PRs that you create as a product that you desire to sell. The more you commit to the effort of creating it – the more grateful and happy will be people working on it and reviewing it.