Forem: Igor Kravchenko

How We Launched on ProductHunt Twice: Insights and Lessons

Igor Kravchenko — Wed, 27 Nov 2024 12:31:08 +0000

Many people have launched various projects on Product Hunt, but few have launched the same project twice. Sit back, and I’ll share some indie project horror stories...

A little introduction: For those unfamiliar, Product Hunt is the most well-known (since 2013) and largest platform for launching various projects. Nowadays, even openAI and Nvidia launch there, or someone mimics them quite well...

Launch

After two years of developing ASO.dev, we decided it was time to stop postponing marketing efforts.

Since I’ve been studying startup materials for over 10 years, we didn’t dive deeply into launch guides (maybe a mistake) and proceeded with the understanding that we were aiming for an SEO link and maybe some traffic.

We prepared materials and postponed the launch for two months, mainly improving the product and onboarding. We received connection requests on LinkedIn, voted for other projects, and, knowing that 700–800 votes were unattainable for us, decided not to compete for the top spots.
We recorded a video (finished it right on the co-founder’s birthday) and scheduled our launch for the company’s two-year anniversary. I’d share a link, but... Product Hunt deleted everything (oops, spoiler).

We emailed our client base, shared the link in channels, and asked other company founders to vote for us (sweet revenge for all the upvotes we’d given).
The launch was on a Wednesday, and we got 69 votes...
Let me explain in detail why this was a failure, and it wasn’t because of the vote count—we made it into the day’s top 10 projects.

We went through the entire Product Hunt launch process, but even the day after voting ended, we could only be found via a direct link.

Mentioning the product using !
Appearing in search by product name
Leaving a review for our project
Adding it as an alternative to another project
Listing the company or product usage in your profile\ None of this worked...

Over two days, we tried to resolve these issues with support.\
On the first day—the launch day—responses were infrequent, mostly directing us to guides. It was clear they couldn’t fix their technical problems.

Second Launch

We asked support if a second launch could solve the issue, and by Friday, they deleted EVERYTHING: all progress, the product, the company, and even the links.

They replaced all links with new ones, and we redid everything in two days, scheduling a launch for Monday, November 25, 2024.
For the second launch, we were "Featured" again, asked everyone to vote, and got even more upvotes, making it into the top 10 by the time of the newsletter.

The "Product Hunt Daily" newsletter goes out at 15–20 UTC and includes the top 10 "Featured" products.

You can track voting, for example, here—it clearly shows vote deductions and spikes. It displays not only "Featured" projects, but we’re absent there—too few votes.

Takeaways:

From the first launch, we gained experience and valuable lessons:

DO NOT ask support to change links to more readable or pretty ones! (They can break everything.)
Editing the launch title or product name may change the link (also applies to already published projects).
If your project isn’t "Featured," the launch is almost useless. Nowadays, "Featured" status is crucial—even if you have thousands of votes, without "Featured," you won’t get a badge or make it into the newsletter! "Featured" projects gather slightly more upvotes (personal experience 😂). I don’t know if this is a new or old rule, BUT without "Featured," it’s mostly pointless! Whether a project is "Featured" can be determined about a day before the launch—it says "Featured on" or "Posted on" at the bottom.

Achieving "Featured" status AFTER launch is possible, but it’s difficult and time-consuming.

Their mobile app is weird, showing only "Featured" projects and usually limiting the number to 10 (+2 ads).

Some links are intercepted by the mobile app but don’t open...

Comment notifications are only sent to the project’s hunter; makers receive no notifications.

Bonus Knowledge:

You’ll receive lots of spam across all social media and email offering to support your product ~~with bots~~, ranging from 50 to 700 votes for varying sums of money, or posts about you in groups with hundreds or millions of ~~bots~~ people 🙈. You’ll also hear from interesting people launching their projects (we understand them much better now ♡), as well as some sneaky ones claiming they supported you with their 100–500-person team on launch day, which often isn’t true.
Check users’ votes if you agree on mutual support, and verify they haven’t withdrawn their votes afterward (examples exist).

We received almost NO organic traffic. We talked to founders of other projects that secured 1st and 2nd places—they also didn’t get significant user inflow. For 700–900 votes, they had about 100–180 registrations, with many bringing their audience...

From the video views, it’s clear almost no one watched it, despite TWO launches. From my observations, having a video is almost 100% mandatory to be "Featured."
Changing the product category is still buggy—we can’t change it again, and for some reason, ASO.dev is now labeled as the best SEO-tool.

So far, we’ve had ONE purchase with a 25% discount promo code from the launch.
Our launch from November 25 ranked 11th by the end of the day.

SEO performance updates will follow.

P.S.: If you need to work with App Store Connect or organic growth (like SEO but for iOS), try our product ASO.dev 👍

SEO from a Newbie for Beginners

Igor Kravchenko — Fri, 25 Oct 2024 09:51:13 +0000

So, you've got a bootstrap project, an awesome product, but few customers. My time has come to dive into SEO.\
Running a bootstrap or indie project really sharpens a lot of skills and teaches you to focus more on business (money). Plus, you realize that marketers aren't just sitting around—they've got a tough job too!

What follows is a pretty basic rundown of things I've learned and done over the last few days (or weeks).

First Things First: The Domain

They say .com is slightly better, but overall, it doesn't really matter. If your domain isn't brand new, that's even better—domain reputation and backlinks matter! This is also why buying underperforming but promising projects (including domains) can pay off if you know what you're doing.

We snagged aso.dev for a magical $98 (or close to that).

Next, you build the site.\
Since SEO is a priority, server-side rendering (SSR) is a must. There's a lot of talk about Google being able to parse JavaScript, but that's mostly nonsense. Sometimes it can, but it generally hates doing so, so don’t risk it. You could use a heavy SSR setup for React or Angular, but we went with astro.build—super fast, simple, and elegant. They have plenty of free and complex themes, but we settled on their Starlight theme after tweaking the home page a bit (you need to copy the Hero.astro component and override Head.astro).

I'm loving it so far. We're even reworking the site for our player, meows.app, because dynamic rendering based on API responses works perfectly. We hired a junior dev to rewrite it from scratch on the same tech stack for cheap—great practice for him and a faster, simpler site for us. Angular 14 just isn't cutting it anymore, even with Universal rendering (feels like ages ago)—check out our WIP example with Astro.

It’s looking pretty good. Previously, we used tinypng for compressing images (manually or through API + GPT script), but now we're sticking to the built-in image optimization tools in astro.build.

Google Search Console

Next up: Google Search Console. Add your site to track indexing, spot errors, and unlock achievements (which you can flex on Twitter).

Google Search Console

There are plenty of tools out there, but for now, I’m sticking with a couple of free ones.

Ahrefs

User login - Ahrefs

Ahrefs is great because you get 10k site queries and a detailed analysis of your pages’ issues. You can’t fix everything, but reducing errors helps a lot.

This tool provided most of the insights I’ve used to improve my site.

Fixes and Improvements

Optimizing Page Titles and Descriptions

Titles should be 50-60 characters, and descriptions 110-160. In the meta structure of my .md files, I added them like this:

seo:
  seo_title: "All-in-One ASO Solution for iOS Developers, marketing"
  seo_description: "ASO.dev is ultimate tool for App Store Optimization (ASO) with App Store Connect integration.Manage,optimize,grow your apps effortlessly with powerful features"

Then I asked GPT to write a Bash script to check these files. After about 30 tries, it finally worked (still faster than if I did it by hand):

# Function to check the length of seo_title and seo_description
check_seo_params() {
  local file=$1
  local in_seo_block=false
  local seo_title=""
  local seo_description=""

  while IFS= read -r line
  do
    # Look for the start of the seo block
    if [[ "$line" =~ ^seo: ]]; then
      in_seo_block=true
    fi

    # If inside the seo block, search for seo_title and seo_description
    if [[ "$in_seo_block" = true ]]; then
      # Search for seo_title
      if [[ "$line" =~ seo_title:[[:space:]]*\"(.*)\" ]]; then
        seo_title="${BASH_REMATCH[1]}"
      fi
      # Search for seo_description
      if [[ "$line" =~ seo_description:[[:space:]]*\"(.*)\" ]]; then
        seo_description="${BASH_REMATCH[1]}"
      fi
    fi

    # If the seo block ends (new block or end of file), stop reading
    if [[ "$in_seo_block" = true && "$line" =~ ^[^[:space:]] && ! "$line" =~ ^seo ]]; then
      in_seo_block=false
    fi
  done < "$file"

  local have_errors=false

  # Check if seo_title is present and valid
  if [[ -z "$seo_title" ]]; then
    echo $divider
    echo $file
    echo "seo_title: Empty or not found"
    have_errors=true
  elif [[ ${#seo_title} -lt 50 || ${#seo_title} -gt 60 ]]; then
    echo $divider
    echo $file
    echo "seo_title: Length ${#seo_title} (50 <> 60): '${seo_title}'"
    have_errors=true
  fi

  # Check if seo_description is present and valid
  if [[ -z "$seo_description" ]]; then
    if [[ $have_errors = false ]]; then
      echo $divider
      echo $file
    fi
    echo "seo_description: Empty or not found"
    have_errors=true
  elif [[ ${#seo_description} -lt 110 || ${#seo_description} -gt 160 ]]; then
    if [[ $have_errors = false ]]; then
      echo $divider
      echo $file
    fi
    echo "seo_description: Length ${#seo_description} (110 <> 160): '${seo_description}'"
    have_errors=true
  fi

  # Print divider only if there are no errors
  # if [[ $have_errors = false ]]; then
  #   echo $divider
  # fi
}
echo $divider
# Recursive search for all .md and .mdx files in the src/content/docs directory
find src/content/docs -type f \( -name "*.md" -o -name "*.mdx" \) | while read file; do
  # Check if the file contains a seo block before proceeding
  if grep -q "seo:" "$file"; then
    check_seo_params "$file"
  fi
done
echo $divider

Run the script, navigate to the file, copy the text into GPT, and ask for the optimal title and description. Here’s an example prompt:

Write seo_title and seo_description, send the result in English, use best practices and length requirements for seo.
Result in the format:
`yaml
  seo_title: ""
  seo_description: ""
`
seo_title 50-60 symbols, seo_description 100-160 symbols
text is ...

Next, update all pages—super basic, I know, but better than having no metadata or duplicate content.

Favicon Fixes

We messed up the favicon a bit—used realfavicongenerator to generate and test.

OG Tags

I also added OG tags—meta tags that make your links look better in social media previews. At the very least, include og:title, og:description, and og:image (use an absolute path). All our images are served via bunny.net, but I’m still fine-tuning that setup.

application/ld+json

Added some structured data with application/ld+json—no idea if it works, but it seems cool. Check out structured data markup.

 {
        tag: "script",
        attrs: {
            type: "application/ld+json",
        },
        content: JSON.stringify({
            "@context": "https://schema.org",
            "@type": "WebSite",
            url: canonical?.href,
            headline: ogTitle,
            description: page_description_seo,
            image: [imageUrl?.href],

            mainEntity: {
                "@type": "Article",
                headline: page_description_seo,
                url: canonical?.href,
                dateModified: data?.lastUpdated,
                image: [imageUrl?.href],
                author: {
                    "@type": "Organization",
                    name: "ASO.dev",
                    url: "https://aso.dev",
                },
                publisher: {
                    "@type": "Organization",
                    name: "ASO.dev",
                    logo: {
                        "@type": "ImageObject",
                        url: fileWithBase(config.favicon.href),
                    },
                },
            },
        }),
    },

Custom 404 Page

We built a custom 404 page. Defaulting to the index page is considered an error and could hurt SEO.

Setting the noindex meta tag on the 404 page:

// 404
if (canonical?.pathname === "/404") {
    headDefaults.push({
        tag: "meta",
        attrs: {
            name: "robots",
            content: "noindex",
        },
    });
}

Setting up a custom 404 page in Nginx config:

 location / {
            proxy_redirect off;
            absolute_redirect off;

            proxy_set_header Host $http_host;

            try_files $uri $uri/ =404;

            # try_files $uri $uri/ /index.html;
            add_header Cache-Control "no-store, no-cache, must-revalidate, max-age=0";
            add_header Pragma "no-cache";
            add_header Expires "Thu, 01 Jan 1970 00:00:00 GMT";
        }
        # 404 page
        error_page 404 /404.html;
        location = /404.html {
            root   /app;
            internal;
        }

Redirects

We had a bunch of 301 redirects. We made sure all URLs ended with a trailing slash (e.g., https://aso.dev/aso/ instead of https://aso.dev/aso) to avoid duplication. After a thorough review, we eliminated outdated links in the code and added old URL redirects via NGINX.

  # https://aso.dev/app-info/app-info/ https://aso.dev/aso/app-info/
  rewrite ^(/ru|/en)?/app-info/app-info/?$ $1/aso/app-info/ permanent;

hreflang

We added x-default to the hreflang attribute. Didn't know that was a thing until recently.

// Link to language alternates.
if (canonical && config.isMultilingual) {
    for (const locale in config.locales) {
        const localeOpts = config.locales[locale];
        if (!localeOpts) continue;
        const langPostfix = localeOpts.lang === "en" ? "" : localeOpts.lang;
        headDefaults.push({
            tag: "link",
            attrs: {
                rel: "alternate",
                hreflang: localeOpts.lang,
                href: localizedUrl(canonical, langPostfix).href,
            },
        });
    }
    headDefaults.push({
        tag: "link",
        attrs: {
            rel: "alternate",
            hreflang: "x-default",
            href: localizedUrl(canonical, '').href,
        },
    });
}

SEMrush

I’ve been using SEMrush longer—it’s easier to downgrade to a free plan, but 100 checks and their pricing aren’t great.

Backlinks

Getting backlinks is tricky. You need links from reputable sites—spammy backlinks will hurt you. One good link from the New York Times beats hundreds from random blogs. We’re building up our backlink profile by listing our site on startup and indie project platforms. We found an Excel sheet with hundreds of link opportunities and are slowly working through it.

Launching on Product Hunt can help—subscribe for updates. We've delayed the launch a few times, but it’s coming soon.

Also, we try to write genuinely useful articles, not just filler.

I might have missed something or gotten things wrong — feel free to share in the comments!