#87: An important PM skill that I only learned recently

And an impressive AI search tool

Image generated by Dall-E 2. Prompt: a traffic police pointing to the left, at the crossroad, color pencil drawing

One of the best things that this job taught me is how to make people do exactly what I need them to do. In other words: how to give instructions.

For your context, when our new app was about to launch, we needed to test and make sure all the most critical bugs were fixed.

We had to prepare a lot of test cases and ask our Quality Assurance and other Product Managers to help to ensure we leave no stone unturned.

Things started to turn sour when some of my testers produced different results and none of them look like the expected outcome. It turned out that the instructions in my test cases were both unclear and easy to misunderstand.

Here are the three rookie mistakes I made:

1. I was not using the same feature names as my testers. For example, while I call a page "Payment page", they call it "Checkout", and "Payment page" refers to something entirely different

2. I skipped the first basic steps. For example, there was one case that I asked people to click on this button, but I never mentioned all the pages they must visit before they can actually click the button.

3. I forgot the preconditions. For example, some test cases can only be conducted with a few settings turned on/off, but I assumed they know nonetheless. Well, my testers never knew that such settings existed.

These mistakes don't apply to just product testing. I realized that to give "good" instructions, we must really know who we're talking to: what vocabulary they are using and what is their level of understanding of the topic.

If you're thinking "that sounds easy", I hope you will watch this video clip in full (it’s really fun):

No disrespect to anyone, but I think when giving instructions, bearing in mind the other party is a layman is always time-saving and less error-prone.

Weekly discovery: Perplexity.ai

This week I discovered a pretty impressive AI-powered search tool called Perplexity.ai.

Although I'm not sure what data this AI is trained on, it seems to give me quite accurate answers compared to Google. The best thing about Perplexity AI is the answer is written in a conversational format, rather than presenting a list of websites, or sometimes content snippets, like Google.

This is quite convenient if you're writing an introduction about something and don't want to spend too much time consolidating all the facts.

Let's give Many One Percents a shot:

Nice sum-up!

Apparently, Perplexity.ai did the homework about me better than ChatGPT, which might indicate a bit about how differently they are crawling the Internet:

How about Perplexity AI vs our good ol' Google? I'm still trying myself, but in the meantime, you can check out this video:

I'm a bit skeptical about the "AI as a search engine" concept, so I will keep a close eye on this trend and update you later!

Around the internet

1. Microsoft is reported to be in the work of integrating ChatGPT into Bing. If this is true, then I would say this is really strategic to compete with Google! | LINK

2. Well, if you look at available films on Disney+, you can partly understand why

Rex Woodbury @rex_woodbury

Surprising stat from Disney’s earnings call: over 50% of Disney+ subscribers are adults without kids

3. How traditional craftmanship in Mexico survives thanks to Internet trends | LINK

4. Trust me, not all of your friends really know the origin of the memes they sent to you. Outsmart them, by learning about their history. Here's a popular one - LINK

A little question

More than two people have asked if I plan to use AI-writing tools to write this newsletter, and to be honest: I don't know.

On one hand, using AI can help spice up my wording and save me time - instead of writing, I can do more research on cool things on the Internet. Maybe I can write more, in less time. You'd never know if it's me who's writing or the AI.

Look at how ChatGPT adds more flavor to my paragraph in the previous section

On the other hand, AI strips off the joy of writing. I like to write on weekends to calm down and reflect. Browsing and researching don’t bring me as much happiness. It's writing a letter to you, by myself, in my free time, then receiving your replies, that matters.

So my burning hypothetical question is:

How would you feel if you know that a newsletter (mine or others) is written using an AI product (note that all the news and links are still collected manually)?

I'd love to receive your reply. But in case you're busy, a simple vote is okay:

---

Thanks for reading this week’s newsletter. If you find it useful, please help share it with your friends :)

By the way, I will take a two-week break to celebrate the Lunar New Year with my family and friends in Vietnam, so there won’t be any newsletters on the 22nd and 29th of January.

Hit me up if you want to catch up, I will be in Hanoi from 16-25 Jan, and Saigon from 26-29 Jan.

Wish you a relaxing weekend and a productive week ahead!

Yours,
Tuan Mon

P/s: Someone cute forwarded this email to you? You can subscribe below to keep up with my weekly newsletter!

Many One Percents newsletter is my weekly curation of Internet gems, mostly about technology, productivity and product management. If you want to support me, you can either subscribe to this newsletter, or donate via these channels: Momo | Paypal | Vietcombank | Patreon