Never design a product without explicit metrics and tracking tools.

Imagine driving cross-country without a dashboard. How do you know your school zone speed? Low gas? Without a dashboard, you can't monitor your car. You can't improve what you don't measure, as Peter Drucker said. Product managers must constantly enhance their understanding of their users, how they use their product, and how to improve it for optimum value. Customers will only pay if they consistently acquire value from your product.

I’m Solomon Ayanlakin. I’m a product manager at CredPal, a financial business that offers credit cards and Buy Now Pay Later services. Before falling into product management (like most PMs lol), I self-trained as a data analyst, using Alex the Analyst's YouTube playlists and DannyMas' virtual data internship. This article aims to help product managers, owners, and CXOs understand product metrics, give a methodology for creating them, and execute product experiments to enhance them.

☝🏽Introduction

Product metrics assist companies track product performance from the user's perspective. Metrics help firms decide what to construct (feature priority), how to build it, and the outcome's success or failure. To give the best value to new and existing users, track product metrics.

Why should a product manager monitor metrics?

• to assist your users in having a "aha" moment

• To inform you of which features are frequently used by users and which are not

• To assess the effectiveness of a product feature

• To aid in enhancing client onboarding and retention

• To assist you in identifying areas throughout the user journey where customers are satisfied or dissatisfied

• to determine the percentage of returning users and determine the reasons for their return

📈 What Metrics Ought a Product Manager to Monitor?

What indicators should a product manager watch to monitor product health? The metrics to follow change based on the industry, business stage (early, growth, late), consumer needs, and company goals. A startup should focus more on conversion, activation, and active user engagement than revenue growth and retention. The company hasn't found product-market fit or discovered what features drive customer value.

Depending on your use case, company goals, or business stage, here are some important product metric buckets:

All measurements shouldn't be used simultaneously. It depends on your business goals and what value means for your users, then selecting what metrics to track to see if they get it.

Some KPIs are more beneficial to track, independent of industry or customer type. To prevent recording vanity metrics, product managers must clearly specify the types of metrics they should track. Here's how to segment metrics:

1. The North Star Metric, also known as the Focus Metric, is the indicator and aid in keeping track of the top value you provide to users.

2. Primary/Level 1 Metrics: These metrics should either add to the north star metric or be used to determine whether it is moving in the appropriate direction. They are metrics that support the north star metric.

3. These measures serve as leading indications for your north star and Level 2 metrics. You ought to have been aware of certain problems with your L2 measurements prior to the North star metric modifications.

North Star Metric

This is the key metric. A good north star metric measures customer value. It emphasizes your product's longevity. Many organizations fail to grow because they confuse north star measures with other indicators. A good focus metric should touch all company teams and be tracked forever. If a company gives its customers outstanding value, growth and success are inevitable. How do we measure this value?

A north star metric has these benefits:

• Customer Obsession: It promotes a culture of customer value throughout the entire organization.

• Consensus: Everyone can quickly understand where the business is at and can promptly make improvements, according to consensus.

• Growth: It provides a tool to measure the company's long-term success. Do you think your company will last for a long time?

How can I pick a reliable North Star Metric?

Some fear a single metric. Ensure product leaders can objectively determine a north star metric. Your company's focus metric should meet certain conditions. Here are a few:

1. A good focus metric should reflect value and, as such, should be closely related to the point at which customers obtain the desired value from your product. For instance, the quick delivery to your home is a value proposition of UberEats. The value received from a delivery would be a suitable focal metric to use. While counting orders is alluring, the quantity of successfully completed positive review orders would make a superior north star statistic. This is due to the fact that a client who placed an order but received a defective or erratic delivery is not benefiting from Uber Eats. By tracking core value gain, which is the number of purchases that resulted in satisfied customers, we are able to track not only the total number of orders placed during a specific time period but also the core value proposition.
   

2. Focus metrics need to be quantifiable; they shouldn't only be feelings or states; they need to be actionable. A smart place to start is by counting how many times an activity has been completed.
   

3. A great focus metric is one that can be measured within predetermined time limits; otherwise, you are not measuring at all. The company can improve that measure more quickly by having time-bound focus metrics. Measuring and accounting for progress over set time periods is the only method to determine whether or not you are moving in the right path. You can then evaluate your metrics for today and yesterday. It's generally not a good idea to use a year as a time frame. Ideally, depending on the nature of your organization and the measure you are focusing on, you want to take into account on a daily, weekly, or monthly basis.
   

4. Everyone in the firm has the potential to affect it: A short glance at the well-known AAARRR funnel, also known as the Pirate Metrics, reveals that various teams inside the organization have an impact on the funnel. Ideally, the NSM should be impacted if changes are made to one portion of the funnel. Consider how the growth team in your firm is enhancing customer retention. This would have a good effect on the north star indicator because at this stage, a repeat client is probably being satisfied on a regular basis. Additionally, if the opposite were true and a client churned, it would have a negative effect on the focus metric.
   

5. It ought to be connected to the business's long-term success: The direction of sustainability would be indicated by a good north star metric. A company's lifeblood is product demand and revenue, so it's critical that your NSM points in the direction of sustainability. If UberEats can effectively increase the monthly total of happy client orders, it will remain in operation indefinitely.

Many product teams make the mistake of focusing on revenue. When the bottom line is emphasized, a company's goal moves from giving value to extracting money from customers. A happy consumer will stay and pay for your service. Customer lifetime value always exceeds initial daily, monthly, or weekly revenue.

Great North Star Metrics Examples

🥇 Basic/L1 Metrics:

The NSM is broad and focuses on providing value for users, while the primary metric is product/feature focused and utilized to drive the focus metric or signal its health. The primary statistic is team-specific, whereas the north star metric is company-wide. For UberEats' NSM, the marketing team may measure the amount of quality food vendors who sign up using email marketing. With quality vendors, more orders will be satisfied. Shorter feedback loops and unambiguous team assignments make L1 metrics more actionable and significant in the immediate term.

🥈 Supporting L2 metrics:

These are supporting metrics to the L1 and focus metrics. Location, demographics, or features are examples of L1 metrics. UberEats' supporting metrics might be the number of sales emails sent to food vendors, the number of opens, and the click-through rate. Secondary metrics are low-level and evident, and they relate into primary and north star measurements. UberEats needs a high email open rate to attract high-quality food vendors. L2 is a leading sign for L1.

Where can I find product metrics?

How can I measure in-app usage and activity now that I know what metrics to track? Enter product analytics. Product analytics tools evaluate and improve product management parameters that indicate a product's health from a user's perspective.

Various analytics tools on the market supply product insight. From page views and user flows through A/B testing, in-app walkthroughs, and surveys. Depending on your use case and necessity, you may combine tools to see how users engage with your product. Gainsight, MixPanel, Amplitude, Google Analytics, FullStory, Heap, and Pendo are product tools.

This article isn't sponsored and doesn't market product analytics tools. When choosing an analytics tool, consider the following:

• Tools for tracking your Focus, L1, and L2 measurements

• Pricing

• Adaptations to include external data sources and other products

• Usability and the interface

• Scalability

• Security

An investment in the appropriate tool pays off. To choose the correct metrics to track, you must first understand your business need and what value means to your users. Metrics and analytics are crucial for any tech product's growth. It shows how your business is doing and how to best serve users.

10-sketch explanation

Choosing Tasks

Bringing News

carrying out 1:1s

providing critique

Managing Turbulence

Product management wizardry.

Early adopters buy a product even if it hasn't hit the market or has flaws.

Who are the early adopters?

Early adopters try a new technology or product first. Early adopters are interested in trying or buying new technologies and products before others. They're risk-tolerant and can provide initial cash flow and product reviews. They help a company's new product or technology gain social proof.

Early adopters are most common in the technology industry, but they're in every industry. They don't follow the crowd. They seek innovation and report product flaws before mass production. If the product works well, the first users become loyal customers, and colleagues value their opinion.

What to do with early adopters?

They can be used to collect feedback and initial product promotion, first sales, and product value validation.

How to find early followers?

Start with your immediate environment and target audience. Communicate with them to see if they're interested in your value proposition.

1) Innovators (2.5% of the population) are risk-takers seeking novelty. These people are the first to buy new and trendy items and drive social innovation. However, these people are usually elite;

Early adopters (13.5%) are inclined to accept innovations but are more cautious than innovators; they start using novelties when innovators or famous people do;

3) The early majority (34%) is conservative; they start using new products when many people have mastered them. When the early majority accepted the innovation, it became ingrained in people's minds.

4) Attracting 34% of the population later means the novelty has become a mass-market product. Innovators are using newer products;

5) Laggards (16%) are the most conservative, usually elderly people who use the same products.

Stages of new information acceptance

1. The information is strange and rejected by most. Accepted only by innovators;

2. When early adopters join, more people believe it's not so bad; when a critical mass is reached, the novelty becomes fashionable and most people use it.

3. Fascination with a novelty peaks, then declines; the majority and laggards start using it later; novelty becomes obsolete; innovators master something new.

Problems with early implementation

Early adopter sales have disadvantages.

Higher risk of defects

Selling to first-time users increases the risk of defects. Early adopters are often influential, so this can affect the brand's and its products' long-term perception.

Not what was expected

First-time buyers may be disappointed by the product. Marketing messages can mislead consumers, and if the first users believe the company misrepresented the product, this will affect future sales.

Compatibility issues

Some technological advances cause compatibility issues. Consumers may be disappointed if new technology is incompatible with their electronics.

Method 5 WHY

Let's talk about 5 why, a good tool for finding project problems' root causes. This method is also known as the five why rule, method, or questions.

The 5 why technique came from Toyota's lean manufacturing and helps quickly determine a problem's root cause.

On one, two, and three, you simply do this:

1. We identify and frame the issue for which a solution is sought.

2. We frequently ponder this question. The first 2-3 responses are frequently very dull, making you want to give up on this pointless exercise. However, after that, things get interesting. And occasionally it's so fascinating that you question whether you really needed to know.

3. We consider the final response, ponder it, and choose a course of action.

Always do the 5 whys with the customer or team to have a reasonable discussion and better understand what's happening.

And the “five whys” is a wonderful and simplest tool for introspection. With the accumulated practice, it is used almost automatically in any situation like “I can’t force myself to work, the mood is bad in the morning” or “why did I decide that I have no life without this food processor for 20,000 rubles, which will take half of my rather big kitchen.”

An illustration of the five whys

A simple, but real example from my work practice that I think is very indicative, given the participants' low IT skills.  Anonymized, of course.

Users spend too long looking for tender documents.

Why? Because they must search through many company tender documents.

Why? Because the system can't filter department-specific bids.

Why? Because our contract management system requirements didn't include a department-tender link. That's it, right? We'll add a filter and be happy. but still…

why? Because we based the system's requirements on regulations for working with paper tender documents (when they still had envelopes and autopsies), not electronic ones, and there was no search mechanism.

Why? We didn't consider how our work would change when switching from paper to electronic tenders when drafting the requirements.

Now I know what to do in the future. We add a filter, enter department data, and teach users to use it. This is tactical, but strategically we review the same forgotten requirements to make all the necessary changes in a package, plus we include it in the checklist for the acceptance of final requirements for the future.

Errors when using 5 why

Five whys seems simple, but it can be misused.

Popular ones:

1. The accusation of everyone and everything is then introduced. After all, the 5 why method focuses on identifying the underlying causes rather than criticizing others. As a result, at the third step, it is not a good idea to conclude that the system is ineffective because users are stupid and that we can therefore do nothing about it.

2. to fight with all my might so that the outcome would be exactly 5 reasons, neither more nor less. 5 questions is a typical number (it sounds nice, yes), but there could be 3 or 7 in actuality.

3. Do not capture in-between responses. It is difficult to overestimate the power of the written or printed word, so the result is so-so when the focus is lost. That's it, I suppose. Simple, quick, and brilliant, like other project management tools.

Conclusion

Today we analyzed important study elements:

Early adopters and 5 WHY We've analyzed cases and live examples of how these methods help with product research and growth point identification. Next, consider the HADI cycle.

Deep learning-based generative art is being researched. As usual, self-learning is better. Some models, like OpenAI's DALL-E 2, require registration and can only be used online, but others can be used locally, which is usually more enjoyable for curious users. I'll demonstrate the Stable Diffusion model's operation on a standard PC.

Let’s get started.

What It Does

Stable Diffusion uses numerous components:

• A generative model trained to produce images is called a diffusion model. The model is incrementally improving the starting data, which is only random noise. The model has an image, and while it is being trained, the reversed process is being used to add noise to the image. Being able to reverse this procedure and create images from noise is where the true magic is (more details and samples can be found in the paper).

• An internal compressed representation of a latent diffusion model, which may be altered to produce the desired images, is used (more details can be found in the paper). The capacity to fine-tune the generation process is essential because producing pictures at random is not very attractive (as we can see, for instance, in Generative Adversarial Networks).

• A neural network model called CLIP (Contrastive Language-Image Pre-training) is used to translate natural language prompts into vector representations. This model, which was trained on 400,000,000 image-text pairs, enables the transformation of a text prompt into a latent space for the diffusion model in the scenario of stable diffusion (more details in that paper).

This figure shows all data flow:

The weights file size for Stable Diffusion model v1 is 4 GB and v2 is 5 GB, making the model quite huge. The v1 model was trained on 256x256 and 512x512 LAION-5B pictures on a 4,000 GPU cluster using over 150.000 NVIDIA A100 GPU hours. The open-source pre-trained model is helpful for us. And we will.

Install

Before utilizing the Python sources for Stable Diffusion v1 on GitHub, we must install Miniconda (assuming Git and Python are already installed):

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-Linux-x86_64.sh
chmod +x Miniconda3-py39_4.12.0-Linux-x86_64.sh
./Miniconda3-py39_4.12.0-Linux-x86_64.sh
conda update -n base -c defaults conda

Install the source and prepare the environment:

git clone https://github.com/CompVis/stable-diffusion
cd stable-diffusion
conda env create -f environment.yaml
conda activate ldm
pip3 install transformers --upgrade

Download the pre-trained model weights next. HiggingFace has the newest checkpoint sd-v14.ckpt (a download is free but registration is required). Put the file in the project folder and have fun:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Almost. The installation is complete for happy users of current GPUs with 12 GB or more VRAM. RuntimeError: CUDA out of memory will occur otherwise. Two solutions exist.

Running the optimized version

Try optimizing first. After cloning the repository and enabling the environment (as previously), we can run the command:

python3 optimizedSD/optimized_txt2img.py --prompt "hello world" --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Stable Diffusion worked on my visual card with 8 GB RAM (alas, I did not behave well enough to get NVIDIA A100 for Christmas, so 8 GB GPU is the maximum I have;).

Running Stable Diffusion without GPU

If the GPU does not have enough RAM or is not CUDA-compatible, running the code on a CPU will be 20x slower but better than nothing. This unauthorized CPU-only branch from GitHub is easiest to obtain. We may easily edit the source code to use the latest version. It's strange that a pull request for that was made six months ago and still hasn't been approved, as the changes are simple. Readers can finish in 5 minutes:

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available at line 20 of ldm/models/diffusion/ddim.py ().

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available in line 20 of ldm/models/diffusion/plms.py ().

• Replace device=cuda in lines 38, 55, 83, and 142 of ldm/modules/encoders/modules.py with device=cuda if torch.cuda.is available(), otherwise cpu.

• Replace model.cuda() in scripts/txt2img.py line 28 and scripts/img2img.py line 43 with if torch.cuda.is available(): model.cuda ().

Run the script again.

Testing

Test the model. Text-to-image is the first choice. Test the command line example again:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

The slow generation takes 10 seconds on a GPU and 10 minutes on a CPU. Final image:

Hello world is dull and abstract. Try a brush-wielding hamster. Why? Because we can, and it's not as insane as Napoleon's cat. Another image:

Generating an image from a text prompt and another image is interesting. I made this picture in two minutes using the image editor (sorry, drawing wasn't my strong suit):

I can create an image from this drawing:

python3 scripts/img2img.py --prompt "A bird is sitting on a tree branch" --ckpt sd-v1-4.ckpt --init-img bird.png --strength 0.8

It was far better than my initial drawing:

I hope readers understand and experiment.

Stable Diffusion UI

Developers love the command line, but regular users may struggle. Stable Diffusion UI projects simplify image generation and installation. Simple usage:

• Unpack the ZIP after downloading it from https://github.com/cmdr2/stable-diffusion-ui/releases. Linux and Windows are compatible with Stable Diffusion UI (sorry for Mac users, but those machines are not well-suitable for heavy machine learning tasks anyway;).

• Start the script.

Done. The web browser UI makes configuring various Stable Diffusion features (upscaling, filtering, etc.) easy:

V2.1 of Stable Diffusion

I noticed the notification about releasing version 2.1 while writing this essay, and it was intriguing to test it. First, compare version 2 to version 1:

• alternative text encoding. The Contrastive LanguageImage Pre-training (CLIP) deep learning model, which was trained on a significant number of text-image pairs, is used in Stable Diffusion 1. The open-source CLIP implementation used in Stable Diffusion 2 is called OpenCLIP. It is difficult to determine whether there have been any technical advancements or if legal concerns were the main focus. However, because the training datasets for the two text encoders were different, the output results from V1 and V2 will differ for the identical text prompts.

• a new depth model that may be used to the output of image-to-image generation.

• a revolutionary upscaling technique that can quadruple the resolution of an image.

• Generally higher resolution Stable Diffusion 2 has the ability to produce both 512x512 and 768x768 pictures.

The Hugging Face website offers a free online demo of Stable Diffusion 2.1 for code testing. The process is the same as for version 1.4. Download a fresh version and activate the environment:

conda deactivate  
conda env remove -n ldm  # Use this if version 1 was previously installed
git clone https://github.com/Stability-AI/stablediffusion
cd stablediffusion
conda env create -f environment.yaml
conda activate ldm

Hugging Face offers a new weights ckpt file.

The Out of memory error prevented me from running this version on my 8 GB GPU. Version 2.1 fails on CPUs with the slow conv2d cpu not implemented for Half error (according to this GitHub issue, the CPU support for this algorithm and data type will not be added). The model can be modified from half to full precision (float16 instead of float32), however it doesn't make sense since v1 runs up to 10 minutes on the CPU and v2.1 should be much slower. The online demo results are visible. The same hamster painting with a brush prompt yielded this result:

It looks different from v1, but it functions and has a higher resolution.

The superresolution.py script can run the 4x Stable Diffusion upscaler locally (the x4-upscaler-ema.ckpt weights file should be in the same folder):

python3 scripts/gradio/superresolution.py configs/stable-diffusion/x4-upscaling.yaml x4-upscaler-ema.ckpt

This code allows the web browser UI to select the image to upscale:

The copy-paste strategy may explain why the upscaler needs a text prompt (and the Hugging Face code snippet does not have any text input as well). I got a GPU out of memory error again, although CUDA can be disabled like v1. However, processing an image for more than two hours is unlikely:

Stable Diffusion Limitations

When we use the model, it's fun to see what it can and can't do. Generative models produce abstract visuals but not photorealistic ones. This fundamentally limits The generative neural network was trained on text and image pairs, but humans have a lot of background knowledge about the world. The neural network model knows nothing. If someone asks me to draw a Chinese text, I can draw something that looks like Chinese but is actually gibberish because I never learnt it. Generative AI does too! Humans can learn new languages, but the Stable Diffusion AI model includes only language and image decoder brain components. For instance, the Stable Diffusion model will pull NO WAR banner-bearers like this:

V1:

V2.1:

The shot shows text, although the model never learned to read or write. The model's string tokenizer automatically converts letters to lowercase before generating the image, so typing NO WAR banner or no war banner is the same.

I can also ask the model to draw a gorgeous woman:

V1:

V2.1:

The first image is gorgeous but physically incorrect. A second one is better, although it has an Uncanny valley feel. BTW, v2 has a lifehack to add a negative prompt and define what we don't want on the image. Readers might try adding horrible anatomy to the gorgeous woman request.

If we ask for a cartoon attractive woman, the results are nice, but accuracy doesn't matter:

V1:

V2.1:

Another example: I ordered a model to sketch a mouse, which looks beautiful but has too many legs, ears, and fingers:

V1:

V2.1: improved but not perfect.

V1 produces a fun cartoon flying mouse if I want something more abstract:

I tried multiple times with V2.1 but only received this:

The image is OK, but the first version is closer to the request.

Stable Diffusion struggles to draw letters, fingers, etc. However, abstract images yield interesting outcomes. A rural landscape with a modern metropolis in the background turned out well:

V1:

V2.1:

Generative models help make paintings too (at least, abstract ones). I searched Google Image Search for modern art painting to see works by real artists, and this was the first image:

I typed "abstract oil painting of people dancing" and got this:

V1:

V2.1:

It's a different style, but I don't think the AI-generated graphics are worse than the human-drawn ones.

The AI model cannot think like humans. It thinks nothing. A stable diffusion model is a billion-parameter matrix trained on millions of text-image pairs. I input "robot is creating a picture with a pen" to create an image for this post. Humans understand requests immediately. I tried Stable Diffusion multiple times and got this:

This great artwork has a pen, robot, and sketch, however it was not asked. Maybe it was because the tokenizer deleted is and a words from a statement, but I tried other requests such robot painting picture with pen without success. It's harder to prompt a model than a person.

I hope Stable Diffusion's general effects are evident. Despite its limitations, it can produce beautiful photographs in some settings. Readers who want to use Stable Diffusion results should be warned. Source code examination demonstrates that Stable Diffusion images feature a concealed watermark (text StableDiffusionV1 and SDV2) encoded using the invisible-watermark Python package. It's not a secret, because the official Stable Diffusion repository's test watermark.py file contains a decoding snippet. The put watermark line in the txt2img.py source code can be removed if desired. I didn't discover this watermark on photographs made by the online Hugging Face demo. Maybe I did something incorrectly (but maybe they are just not using the txt2img script on their backend at all).

Conclusion

The Stable Diffusion model was fascinating. As I mentioned before, trying something yourself is always better than taking someone else's word, so I encourage readers to do the same (including this article as well;).

Is Generative AI a game-changer? My humble experience tells me:

• I think that place has a lot of potential. For designers and artists, generative AI can be a truly useful and innovative tool. Unfortunately, it can also pose a threat to some of them since if users can enter a text field to obtain a picture or a website logo in a matter of clicks, why would they pay more to a different party? Is it possible right now? unquestionably not yet. Images still have a very poor quality and are erroneous in minute details. And after viewing the image of the stunning woman above, models and fashion photographers may also unwind because it is highly unlikely that AI will replace them in the upcoming years.

• Today, generative AI is still in its infancy. Even 768x768 images are considered to be of a high resolution when using neural networks, which are computationally highly expensive. There isn't an AI model that can generate high-resolution photographs natively without upscaling or other methods, at least not as of the time this article was written, but it will happen eventually.

• It is still a challenge to accurately represent knowledge in neural networks (information like how many legs a cat has or the year Napoleon was born). Consequently, AI models struggle to create photorealistic photos, at least where little details are important (on the other side, when I searched Google for modern art paintings, the results are often even worse;).

• When compared to the carefully chosen images from official web pages or YouTube reviews, the average output quality of a Stable Diffusion generation process is actually less attractive because to its high degree of randomness. When using the same technique on their own, consumers will theoretically only view those images as 1% of the results.

Anyway, it's exciting to witness this area's advancement, especially because the project is open source. Google's Imagen and DALL-E 2 can also produce remarkable findings. It will be interesting to see how they progress.

This report from Mark Gurman has stuck with me:

In the News and Stocks apps, the display ads are no different than what you might get on an ad-supported website. In the App Store, the ads are for actual apps, which are probably more useful for Apple users than mortgage rates. Some people may resent Apple putting ads in the News and Stocks apps. After all, the iPhone is supposed to be a premium device. Let’s say you shelled out $1,000 or more to buy one, do you want to feel like Apple is squeezing more money out of you just to use its standard features? Now, a portion of ad revenue from the News app’s Today tab goes to publishers, but it’s not clear how much. Apple also lets publishers advertise within their stories and keep the vast majority of that money. Surprisingly, Today ads also appear if you subscribe to News+ for $10 per month (though it’s a smaller number).

I use Apple News often. It's a good general news catch-up tool, like Twitter without the BS. Customized notifications are helpful. Fast and lovely. Except for advertisements. I have Apple One, which includes News+, and while I understand why the magazines still have brand ads, it's ridiculous to me that Apple enables web publishers to introduce awful ads into this experience. Apple's junky commercials are ridiculous.

We know publishers want and probably requested this. Let's keep Apple News ad-free for the much smaller percentage of paid users, and here's your portion. (Same with Stocks, which is more sillier.)

Paid app placement in the App Store is a wonderful approach for developers to find new users (though far too many of those ads are trying to trick users, in my opinion).

Apple is also planning to increase ads in its Maps app. This sounds like Google Maps, and I don't like it. I never find these relevant, and they clutter up the user experience. Apple Maps now has a UI advantage (though not a data/search one, which matters more).

Apple is nickel-and-diming its customers. We spend thousands for their products and premium services like Apple One. We all know why: income must rise, and new firms are needed to scale. This will eventually backfire.

Every week I meet many people interested in VC. Many of them ask me what it's like to be a junior analyst in VC or what I've learned so far.

Looking back, I've learned many things as a junior VC, having gone through an almost-euphoric peak bull market, failed tech IPOs of 2019 including WeWorks' catastrophic fall, and the beginnings of a bearish market.

1. Network, network, network!

VCs spend 80% of their time networking. Junior VCs source deals or manage portfolios. You spend your time bringing startups to your fund or helping existing portfolio companies grow. Knowing stakeholders (corporations, star talent, investors) in your particular areas of investment helps you develop your portfolio.

Networking was one of my strengths. When I first started in the industry, I'd go to startup events and meet 50 people a month. Over time, I realized these relationships were shallow and I was only getting business cards. So I stopped seeing networking as a transaction. VC is a long-term game, so you should work with people you like. Now I know who I click with and can build deeper relationships with them. My network is smaller but more valuable than before.

2. The Most Important Metric Is Founder

People often ask how we pick investments. Why some companies can raise money and others can't is a mystery. The founder is the most important metric for VCs. When a company is young, the product, environment, and team all change, but the founder remains constant. VCs bet on the founder, not the company.

How do we decide which founders are best after 2-3 calls? When looking at a founder's profile, ask why this person can solve this problem. The founders' track record will tell. If the founder is a serial entrepreneur, you know he/she possesses the entrepreneur DNA and will likely succeed again. If it's his/her first startup, focus on industry knowledge to deliver the best solution.

3. A company's fate can be determined by macrotrends.

Macro trends are crucial. A company can have the perfect product, founder, and team, but if it's solving the wrong problem, it won't succeed. I've also seen average companies ride the wave to success. When you're on the right side of a trend, there's so much demand that more companies can get a piece of the pie.

In COVID-19, macro trends made or broke a company. Ed-tech and health-tech companies gained unicorn status and raised funding at inflated valuations due to sudden demand. With the easing of pandemic restrictions and the start of a bear market, many of these companies' valuations are in question.

4. Look for methods to ACTUALLY add value.

You only need to go on VC twitter (read: @vcstartterkit and @vcbrags) for 5 minutes or look at fin-meme accounts on Instagram to see how much VCs claim to add value but how little they actually do. VC is a long-term game, though. Long-term, founders won't work with you if you don't add value.

How can we add value when we're young and have no network? Leaning on my strengths helped me. Instead of viewing my age and limited experience as a disadvantage, I realized that I brought a unique perspective to the table.

As a VC, you invest in companies that will be big in 5-7 years, and millennials and Gen Z will have the most purchasing power. Because you can relate to that market, you can offer insights that most Partners at 40 can't. I added value by helping with hiring because I had direct access to university talent pools and by finding university students for product beta testing.

5. Develop your personal brand.

Generalists or specialists run most funds. This means that funds either invest across industries or have a specific mandate. Most funds are becoming specialists, I've noticed. Top-tier founders don't lack capital, so funds must find other ways to attract them. Why would a founder work with a generalist fund when a specialist can offer better industry connections and partnership opportunities?

Same for fund members. Founders want quality investors. Become a thought leader in your industry to meet founders. Create content and share your thoughts on industry-related social media. When I first started building my brand, I found it helpful to interview industry veterans to create better content than I could on my own. Over time, my content attracted quality founders so I didn't have to look for them.

These are my biggest VC lessons. This list isn't exhaustive, but it's my industry survival guide.