As of today, we just completed our first event/activity, out giveaway. We had a generous follower that suggested and sponsored this. So we just had to run it. We gave away 2 books and 2 vouchers from a local bookstore. It went well, we managed to gain new followers from doing this and it seems to be making our followers happy.

Coming up next would be the book club. We’ll have a journalist host the first version and it will be done virtually. It will cover 2 books and there’s even a theme to it.

Apart from running events, we have our dev team working on upgrading the borrowing process and our website experience. And we are also in the midst of trying different ways to get our name out.

Alhamdulillah, all is going well so far. Looking forward to see how all of this turns out.

The post Maktaba Books – First major event done appeared first on Azrin Hamdan.

There’s one thing I realised that I’ve been very wrong about – it is to use the term analytics & statistics interchangeably. It’s two different things that look similar but serve very different purposes. Here’s how.

What is analytics

To find the answer for this I went to the ever-reliable Professor Google to look it up and I found 3 definitions that are nice.

Defined by Merriam Webster

Defined by towardsdatascience.com (my favourite)

If you squint your eyes and put the 3 definitions together, you can see that they all mean to look something up about something that already exists so that you know the answer to your questions. Sounds easy right, why do you need someone’s job dedicated to doing just that? When I read more into the article by Towards data science, it explains that answering a specific question with data is much easier than generating inspiration about which questions are worth asking in the first place.

To put it simply I find analytics to be like doing the research review for the reports I did in university and to come up with questions that has not been answered in any of the past research done.

What is statistics

Definition by Oxford Dictionary

Definition by Urban Dictionary

Definition by Towards data science

When combining all the definitions, it seems that statistics is used to test a guess to see if it would be true when you scale it up.

Analytics vs Statistics

I found Towards data science’s explanation on analytics vs statistics to be the best and clearest at differentiating the two and how they are complementary to each other.

It states that analytics helps you find and improve the quality of the questions to ask. While statistics help to find and improve the quality of the answer. They come hand in hand to solve data problems. I start with analytics, move on to statistics, and go back to analytics to repeat the cycle.

Why I want to know and apply both Analytics and statistics to marketing

After learning about the definition and purpose of Analytics and statistics, I’m pretty convinced that I need to know and apply both as a marketer in order to truly be a data-driven marketer.

As mentioned earlier, analytics form & improves the questions to ask and statistics find and improve the quality of the answers. If I only knew analytics, I would only know how to raise questions but I would know how to answer them or find a quality answer. While only knowing statistics, I would have quality answers to pointless questions.

I think knowing one is okay but to be really good at marketing, having knowledge of both analytics and statistics would be necessary.

How I’m applying analytics to marketing

I’m probably doing the same as the average marketer out there:

I use website analytics tools for reports on website traffic performance. I’ve used a bunch of them before, from Google Analytics; Smartlook; Hotjar;  Yandex Metrica; and more. Each with their own strengths.

What I would usually do is to look at the website traffic performance for the overall numbers like website visits, bounce rate, time spent on the website, and which sources & mediums are they coming from. Should I see anything interesting, I would dive deeper into the data to find out why it is happening.

For example, in a situation where I noticed that there’s a relatively high Facebook ad CTR but the form conversion rate of the landing page is relatively low. I then investigate further by looking at the page experience from a technical aspect and copywriting aspect. For this, I usually do check for technical issues first, by checking the loading times of the landing page and the forms. If that is not the problem, I would look at sample of sessions of such users to find out what’s happening and it could be that for some reason, the form button colour blends in with the form background.

But to use it for more than just identifying problems, I could also use it to identify things that are working really well. Like for example, I noticed that leads that are coming from Linkedin may be less than those that are coming from Twitter but those leads are of high quality that matches our intended customer profile (ICP) or have high levels of intention. From that finding, I would probably want to test if we double up on Linkedin marketing activities, it would incrementally improve our marketing results. This goes to my next part on how I’m applying statistics.

My plan for applying statistics to marketing

To be honest, I have not applied any form of statistics in my work. I’m doing a bunch of experiments for marketing but not in a very scientific manner that would require some form of statistics application. Which I realised is a bad practice after reading Sean Ellis’s Hacking Growth book because I need to test for statistical significance to ensure that the result achieved was due to the new variable I introduced during the experiment.

I’m still learning my way around applying statistics in marketing but I have a rough idea of how it should be done. Essentially, for any experiment that I do (which should be based on analytics), I need to have a “control” in order to prove that the new idea is worthwhile to implement & scale. The Facebook representatives at the She Loves Data workshop call it to test for incrementality.

Incrementality is a way to measure an event that wouldn’t have occurred without a specific interaction, such as an ad view, and that resulted in the desired outcome, such as a conversion. In the context of ads, it can be referred to as the measurement of ad effectiveness.

Measuring Impact: The Beginner’s Guide to Incrementality

So here’s how I would do it:

Step 1: Have a structure to my experiments

If I had paid better attention in my market research module in school, I would have learned, applied, and memorised this by heart. So, essentially all my experiments should be made up of a hypothesis, a control, a dependent variable, an independent variable, and an experiment timeline or goal.

I also needed to determine my sample size which I based on this article: https://blog.hubspot.com/marketing/email-a-b-test-sample-size-testing-time

Step 2: Run the experiment

Set the experiment as mentioned earlier, let it run for a while, and wait for the results.

Step 3: Test marketing experiments for incrementality

As mentioned earlier, I need to find out if the results of the test were actually caused by the introduction of a new variable and not by something else. To do so, I would take the results of the experiment and apply one of the many tests for statistical significance. After finding out if the results were significant, then I would make a conclusion on this experiment.

Good reads on statistical significance:

• How to Understand & Calculate Statistical Significance [Example]
• A Data-Driven Marketers’ Guide to Calculating Statistical Significance

Applying statistics to a Facebook Ads experiment

If I read and learn all about this but don’t apply, what’s the point right? So I applied this on a Facebook Ad that I was running.

The context is that I had an ad that was performing “okay” but during a call with a Facebook Ad Expert, I was recommended to change the ad optimisation method from doing it based on Landing Page Views to Link Clicks. I was curious to see if this was true by testing it out.

I started off by having a hypothesis “changing the Facebook Ads optimisation based on Link Clicks will not improve the CTR of the ad when compared to optimising is based on landing page views”. From the hypothesis, the independent variable is the method of optimisation and the dependent variable is the CTR.

To ensure the test is done in a random but controlled manner, I did it using Facebook’s A/B testing tool. And I decided to run it for 14 days to get enough of a sample size.

The expected results

The results after 14 days

The statistical significance test results

After getting the results, I checked for statistical significance to find out if I repeated the same test, would I still get similar results and that this result did not happen by chance. 

For this situation, I used the chi-square test (see the formula in the image below) to calculate the significance of the results. The value I received was 19.40. 

Conclusion of this test

In order to determine if the above-mentioned value is good or bad, I needed to take it and compare it to the chi-square table for the confidence level of 0.95 and the level of freedom of 1. When I did so, I found that the value was larger than what was indicated in the table. Hence the results of this experiment are statistically significant and therefore using Link Click Optimisation would be a better option in generating traffic for the website than using Landing Page View Optimisation.

Recommended content around analytics & statistics

If you’re curious about analytics & statistics and how you can use them in marketing, you can try reading the stuff I read to get the understanding to write this essay. Here’s the list:

• Hacking Growth – To convince you why it is important to be data-driven for marketing
• Towards data science’s article on Analytics & statistics – to get a clear understanding of what the two are, how they work together, and why you need it for marketing
• Technical Marketing Handbook podcast – Useful technical knowledge around marketing, mostly on how to gather the best form of analytics and the challenges of it
• Nacho analytics’s blog post titled A data-driven marketer’s guide to calculating statistical significance – How to calculate statistical significance using free tools

Ping me if you have any recommended content for me to add or read that is super useful to understand analytics and statistics better.

That’s all I can say about analytics & statistics for marketing at the moment, I’m still learning about it.

In the meantime, selamat (peace out in Malay)

P.S. Next piece will be a deep dive on analytics, I actually wanted to go into statistics first but it would be dumb of me to do so after finding out the relation between the two.

The post Analytics vs statistics deep dive: Awesome combo for better marketing appeared first on Azrin Hamdan.

It was really nice to see how we grew from a one-person team to three and now we are a team of six. Alhamdulillah. Everyone seemed to be getting along with each other and were very engaging.

The discussion went well, we mapped out the structure and processes of Maktaba Books. It looks really exciting. 2022 is going to be an interesting year for us.

I look forward to sharing more updates about this.

The post Maktaba Books – First big team meeting appeared first on Azrin Hamdan.

As mentioned in the previous update, we went past 500 followers, so we did an achievement post (see image below). We discovered that we have had 41 books borrowed in total. And that was quite shocking to find out that the uptake has been that good.

Apart from those achievements, we also realised that the dismal conversion rate was not because our website was horrible but it was due to not telling users to borrow and what to borrow. We just needed more intentional traffic. So to do this, we started publishing book recommendations based on the categories we have. This lets our followers know what the categories mean and what books they should read, killing two stones with one bird . To make it more actionable, we reshared those posts as a story and included a link sticker to the book page to borrow it. It has been super effective, the new Instagram update has been super useful.

We also made another interesting discovery – our followers like to engage with content around questions or quizzes. This happened when we decided to share a random Malay vocabulary test by the Bulan Bahasa account asking what is a Violin called in Malay. It somehow blew up, we received 20 over messages answering it trying to find out if their answer was right. It’s still a preliminary finding, we will need to test this theory out more before we can confidently say that they love that kind of content.

We also have some internal updates – our partner, Quiet, have officially opened their co-working space. Their space holds our donated books and users can choose to do a book pick up there instead of a meet up of mail delivery.

Another news is that we now have a new Developer joining the volunteer team. We may also get another 2 volunteers joining our team – a UX Designer and another marketing team member. That would make us a team of 7 . Alhamdulillah, truly amazed and grateful to get to work with such generous people that want to contribute.

The post Maktaba Books – it’s alive and growing appeared first on Azrin Hamdan.

Also, one thing to note: when I talk about automation here, it is mainly marketing automation and/or task automation (ala Zapier).

Why learn automation for digital marketing?

Before I start anything, I ask myself why I should learn about automation and how it is useful for me as a full-stack marketer.

Firstly, it saves time. This is very important when I work in small teams or when I’m working on my side projects. I don’t want to burn most of my time on just doing routine work. I would want to do as little repetitive work as possible and focus more on tasks that add value or require a higher level of thinking. It’ll make something that needs to be done by two to three people into something that can be done by a single person.

Next is that I am not a coder or a programmer but I want and like to build stuff. So this is a way to work around it. With tasks automation tools, I can make my own web app or website’s backend run without writing any code.

By using automation, I’ll free up more time and I can build things online.

How can automation be used in Digital Marketing

It can be used for anything that is easy to replicate and can be broken down into simple steps. Anything that does not require a high level of thinking.

I use it mainly for doing repetitive tasks that don’t add much value to my work but are essential. For example, giving daily reports on the performance of a digital marketing campaign. I’m just sending a message or an email to my boss or team about the numbers that I looked up or calculated. And that takes up 10 to 15 minutes of my time every day – that’s 1h15min in a week, 5 hours a month, 60 hours a year. Sure, there is the option of paying someone to do that or do it myself manually, but when I’m tight on budget & time, I can’t do that. And usually, to do simple automation, I can use the free plans. Even if I upgraded the plans, it’s just US$5 to US$25 per month which is much cheaper than hiring a person to do it manually.

Examples of automation that I use for my work:

• Reporting ad performance to my team and boss
• Getting Telegram notifications for form submissions
• Having form submissions update a database

For more examples:

• Hubspot’s blogpost for 37 different kinds of marketing automation
• Integromat’s apps & services page describes all possible actions for different tools that it integrates with

Lesson 1: do it manually to understand the process and pain points before automating it

This is something I overlooked at the start because I thought that I knew how it works in my head and I just needed to set up the automation. But boy I was wrong. I mean, I could set it all up but the first attempt was bad, there were a lot of parts that didn’t work properly and I missed out on some steps. I had to test it a lot of times, find something that I missed, fix it, and test it again. I repeated this cycle again and again.

The next time I did it, I tried doing it manually first until I was sure that I knew all the steps that I needed for the specific process. Only after that, did I start automating it. This saved a lot of painful hours of testing and fixing.

So yeah, lesson 1: first do it manually and then automate it.

Also see: Zapier’s blog post on when to automate

Lesson 2: Picking the right automation tool that can achieve your automation goal

After knowing and understanding the actual processes that I want to automate, the next key lesson that I learned was choosing the right automation tool. There were so many automation tools out there and the main difference would be the tools it could integrate with and the price (which usually gets more expensive when the frequency of automation cycles are higher and the number of processes done).

The most famous automation tool would be Zapier. They are easy to learn and use but are very expensive to scale up. I prefer to use Integromat, the paid plans have more bang for the buck and they are the only automation tool that can integrate with Telegram easily. It even helps you make Telegram bots easily without any coding needed.

If you want to check out Integromat, use my referral code here: https://www.integromat.com/?pc=azrintegromat.

Here’s a comparison list of automation tools: https://www.remote.tools/zapier/alternatives

Lesson 3: test, monitor & adjust

This is the most painful part, testing; monitoring; and adjusting my automation. It is painful to do but very important to keep doing it consistently, especially for very long and complicated processes like the automation I did for Maktaba Books. There’s always some form of error happening.

After doing a bunch of automation, I found that I needed to do three kinds of tests – individual tests of each automated process, system test of all the automated processes together, and real-world test (getting a small group of people to test it in real conditions, not in a controlled environment).

For monitoring the live automation, I did it by spending 30 – 60 mins every 2 days to see if there’s anything that needs to be adjusted. Automation tools usually have a notification system that will let you know if something stopped working. You can rely on that but I just wanted to be sure so I checked manually at regular intervals

So always be testing, monitoring, and adjusting.

Lesson 4: get better by automating complicated processes

At this point, I’m already comfortable and familiar with the steps of automating simple processes. The next step I took was to automate a process that was complicated with many moving parts. So I decided to automate one of the processes of my passion project – Maktaba Books’ borrowing process. This involved:

• Collecting information from a book borrower together with the book that they want to borrow
• Store the information above in our database, update the book database to change the availability of it
• Based on the information shared, a message would be sent to the book owner to notify them that someone wants to borrow their book
• Being the messenger that sends messages between the borrower and the book owner
• Reminding borrowers of due dates
• Sending messages to offer book extensions
• Updating the book database to show the book’s availability when the book is returned

This was very challenging and I took about one to two months to set this up, test it and have it live and running. But it was very satisfying to complete.

Once you’re comfortable, take on big projects.

Automation examples or ideas for you to try and figure out

Description of automation	Difficulty
Sending Google Sheet data updates as an email	*
Using a Telegram bot to notify updates about Google Sheet changes
Triggering an email & a message via an app (Telegram, Slack, etc.) whenever a new web form is submitted
Automate the whole backend of a website, like how I did it for Maktaba Books. Message me on Linkedin to find out how

Automation can be a bit tough to learn (much easier than learning how to code) but fun once you know how to and can make your marketing life much much easier. It is very useful knowledge to have as a full-stack marketer. What have you tried to automate? Tell me by messaging me on Linkedin, would love to get more inspirations

Till next time, selamat (peace out in Malay),
Azrin Hamdan

P.S. Next topic, I’ll be focusing on analytics because I feel that I’ve been half-assing it

The post Automation deep dive: learning how to grow more limbs appeared first on Azrin Hamdan.

Our website is getting a decent amount of traffic at about 10 – 15 user visits per day. But many of them or not doing much there – they land on the page, scroll through the list of books, and that’s all. 96% of them stop there without seeing more of the books or clicking any buttons. That’s a huge drop off. My hypothesis is that it is hard for them to find a book that they want or they don’t know what they want and there’s nothing to help them decide.

To improve on our 0.56% conversion rate (borrowing books), we will be spending more time to find out why users are not borrowing books and how we can improve that. So, we split it into two parts – asking users that have not borrowed “what’s stopping them?” and asking our power users “why they borrowed from us, what is the problem we solve for them?”. While that is ongoing, we will test my hypothesis that users need help to decide on a book by posting content around book recommendations and describing our book categories.

Some good news is that we’ve passed a milestone of having 500 followers. 99% of which were gained organically. Alhamdulillah, glad people find what we do to be useful.

The post Maktaba Books – Good views but not much action appeared first on Azrin Hamdan.

Since I’ve decided to focus on polishing the product, over the next few months my main role would be focused on operating the platform – making sure everything runs smoothly and fix any automations that are not working. I expect this to last till the end of the year until version 3, which is getting build by an actual developer, is ready to run. This is a very painful process for me (maybe for everyone too?) because I’m not very good at doing routine housekeeping of any kind. But it’s something I have to do to help keep this going, so I just have to persevere.

While I focus my energy and time on operating the tech side of Maktaba Books, the rest of the team will be focused on growing the usage of our platform by our existing users. I’ve shared this mission with them and they were more than happy to take it on. I’m looking forward to what they come up with.

The post Makataba Books – Maintaining, fixing errors and increasing engagement appeared first on Azrin Hamdan.

The new setup

When we started this, we were running it on Instagram – we had a Google Sheet filled with 400+ books and people had to DM us to borrow the books, it was very manual. Now we have a website for all the books (with photos & summaries) and the borrowing process is all automated using a combination of Telegram bot, Google Forms, Google Sheets, and Integromat.

Promoting the launch

We promoted it organically on Instagram and using Facebook Ads (we had free credits).

Organically, we posted a couple of images on Instagram feed and Instagram stories. To maximise the reach we created a close friends list of all our followers and shared the post directly to them via Instagram messages.

For the paid ads, we were experimenting with placement, creative, copy, and audience targeting. This was mainly to reach new people.

I’m really happy and thankful with where we are, the help we’re getting and the team we have. I’m really glad to have Nurhadi Tahan, Izza Haziqah, and Abdul Aziz with me on the team. Alhamdulillah. Even if this go sideways, it is still a win.

The post Maktaba Books – Launch & Tech Dev Team Member appeared first on Azrin Hamdan.

I was soo wrong about it. Only the borrowing automation was ready. I still needed to have a proper book catalogue with pictures and summaries, guide on how the new process works (for both borrowers and book owners), onboard the book owners, and plan out how to do the actual testing.

For the book catalogue, it took me about a week to finish. With the help of my girlfriend, I managed to update the images and summaries at a rate of 50 books a day. It now looks like proper digital library, you can check it out here if you want.

Creating a guide for the book owners and borrowers was tough. Explaining what they need to do step by step in a way that’s is very easy to comprehend took us a few revisions to get it right because we knew that we should spend more time here to avoid bigger headaches down the road.

You can have a look the book owner guide and the borrower guide.

After setting up the idiot-proof guide, we then proceeded to work on getting book owners onboard our Telegram process. This required a lot of cross platform communication. We had to ask their Telegram contact details on Instagram messaging and once we had the Telegram details and for some reason they don’t reply to us, we would message them on Instagram.

Last part was to actually plan out what we wanted to do for testing. After some discussion, we agreed to have it tested by ourselves (the team), our friends, and the existing Maktaba Books power users. That would total up to 10 users.

For the testing we wanted to get feedback on the borrowing process, and after returning the book. So we know exactly annoys/makes it difficult for users. To collect feedback, we created a Google Form to capture it, where it’s a mix of qualitative and quantitative questions. And in any situation we are interested to hear more about their feedback, we would actually have a call with them to dig.

So yeah, now we’ve been officially testing since 13 September. The next step is to go through the feedback, apply it and launch it fully.

The post Maktaba Books – Testing update appeared first on Azrin Hamdan.

I am excited but kind of worried of the problems that might appear during the testing phase. I’m not worried about the number of problems that we are going to find but I’m worried about finding problems that we cannot solve.

I think the best way to look at it is to find all the problems as early as possible so that we have more time to fix it. Because at the end of the day, we want to have a system that works well and is easy to use.

As of today, 3 people have tested it and we’ve found quite a number of problems. It’s nothing too major that we cannot fix.

I’m looking forward to launch this September with the team.

The post Maktaba Books – Testing appeared first on Azrin Hamdan.