Ruling is hard. This was maybe my answer to Tolkien, whom, as much as I admire him, I do quibble with. Lord of the Rings had a very medieval philosophy: that if the king was a good man, the land would prosper. We look at real history and it’s not that simple. Tolkien can say that Aragorn became king and reigned for a hundred years, and he was wise and good. But Tolkien doesn’t ask the question: What was Aragorn’s tax policy? Did he maintain a standing army? What did he do in times of flood and famine? <…>

Real-life kings had real-life problems to deal with. Just being a good guy was not the answer. You had to make hard, hard decisions. Sometimes what seemed to be a good decision turned around and bit you in the ass; it was the law of unintended consequences. I’ve tried to get at some of these in my books. My people who are trying to rule don’t have an easy time of it. Just having good intentions doesn’t make you a wise king.

I immensely admire both Martin and Tolkien. If you are also a fan, you can learn a lot about management just by following this imaginary debate. It’s one thing to study from textbooks. But when you recognize patterns in emotionally charged arcs, it’s much closer to gaining your own experience. Let’s go!

Where and why Robb Stark failed as a manager

Robb Stark is the clearest representation of “Ruling is hard. Just being a good guy is not the answer.“ Martin was very thorough and made the Young Wolf face almost every type of CEO problems. I am sure any boss can relate:

• Unclear task setting for not-so-bright subordinates. Robb told Edmure to “hold Riverrun“ and then was furious that Edmure didn’t understend his whole strategy and accidentally destroyed it1. Rookie mistake. Any manager learns pretty quickly that if a task can be interpreted in the wrong way, it will be interpreted in the wrong way.

• Unreliable team members being given important tasks. Appointing top executives and tracking how well they are suited to their jobs is a core part of a CEO’s workload. Roose Bolton is a prominent commander in Robb’s army despite the horrendous reputation of his house. Bolton was a good choice when Robb needed to distract lord Tywin, but holding Harrenhal is a much more straightforward task. It’s pretty clear what needs to be done. Recall Bolton to Riverrun, benefit from his cunning counsel and keep an eye on him.

• Ignoring structural risks even when they are apparent. It took a very long time before Robb realized (only with Catelyn’s help2) that he should defend the North from the Ironborn to preserve his power base instead of valiantly fighting Lannisters. Just like most CEOs love ambitious projects and hate housekeeping.

• Not dealing with a disgruntled team member. Rickard Karstark was unhappy and annoyed for a long time before he actually caused problems. Robb did nothing about it besides having a talk with him. He could have sent Karstark to retake the North, solving both problems.

• Also: personal life affecting business (the marriage with Jeyne Westerling), rigidity in decision making (the Karstark execution), lack of team alignment around strategic goals (allowing the Riverlords to go defend their own lands separately)…

Many of Robb’s mistakes stem from being a good guy: honorable, trusting, forgiving and consistent with his values. It seems Martin’s thesis is proven as much as it can be by means of fiction. But does it mean that another good guy — Aragorn — would also face the same devastating issues and fail? Not at all!

Why Aragorn’s success as a ruler is almost guaranteed

Whether you are a good guy or not is not the only variable that affects your results as a manager. We tend to overestimate how much success is tied to personal traits and underestimate the specifics of a situation: resources, opportunities, threats.

Robb started to rule in insanely harsh conditions. Strong enemies. Zero people he had worked with before. Long borders to defend. No allies. Theoretically royal power, but in practice he is dismissed as a “boy” from time to time. You have to be a grandmaster of Machiavellian games to win.

Aragorn started to rule after he won his war. See how many structural advantages he has:

• A happy and secure personal life..

• Battle-tested talent pool of literal superhumans with the highest moral standards. Can you imagine someone from the Grey Company scheming like the Boltons?

• Wise advisors. Imagine how good Elrond with his thousands of years of experience would be in the role of a management consultant.

• Immense fame and undisputed authority. Aragorn just won the war, performed miracles and married an Eldar princess.

Go to the list of Robb’s challenges and see how many problems this eliminates. And note that all of these factors have nothing to do with Aragorn being a skilled statesman. He can be just okay-ish and the Reunited Kingdom will prosper anyway. Suboptimal tax policy? Who cares if you have vast Eriador ripe for peaceful colonization.

To clarify: Aragorn is very likely to be an excellent statesman. He was in the trenches for decades, knows the country like no one else and got the best possible education in Rivendell. My point is, he had such a massive margin of error that we don’t need to discuss his abilities at all.

Aragorn basically ruled on easy mode. He dealt the hard stuff before his coronation.

This easy mode3, however, has its caveats.

The final management challenge to rule them all

Let’s forget for a moment about kings. What does a business do when it has plenty of structural advantages and enjoys playing on easy mode? More often than not the company loses its focus on creating shareholder value:

• Bureaucracy grows, innovations decline.

• Protecting the status quo becomes more important than development.

• Management gets carried away with empire-building and hype trends.

Recent example: Meta’s $80B spending on the Metaverse. It was not necessarily a bad bet (if it worked everyone would praise Zukerberg’s genius). What’s important is that such risky projects can be funded only when the business is doing exceptionally well. Structural advantages increase chances of inventing problems instead of dealing with disciplined capital allocation.

Aragorn’s final test was to be exactly a wise and good guy, not to screw things up because of ego, greed or fear. The film trilogy highlights how Aragorn passed the test of the Ring, much unlike Isildur. But there is another character in the Legendarium to whom we are bound to compare Aragorn.

Many years before Isildur, the Númenórean king Ar-Pharazôn also ruled in the easy mode. His army was so powerful that he captured Sauron without much fight. Some time after he started to worship Morgoth, became obsessed with immortality, and declared war on the Valar4.

Aragorn didn’t do any such things, passing the final challenge. So the land’s prosperity was a foregone conclusion.

Let’s summarize

• Martin is totally right: ruling is hard as hell.

• But sometimes a leader has plenty of structural advantages and rules in the easy mode.

• The problem with easy mode is that people tend to create problems themselves. Being a good guy helps prevent that.

And the final conclusion:

There are times when an organization needs a crisis manager with exceptional problem-solving skills. And there are times when we need a reliable leader who who lets the business come into its own.

If you are not in the easy mode and need help with management challenges, you can schedule a call with me and I’ll help to sort them out.

4

Which quickly led to his death and the destruction of Númenor.

• 75% of purchases will be done via an AI concierge.

• AI will make most of the decisions, customers will give final approval.

• To be competitive, retailers should ensure the quality and AI-readability of their product data and also outsource most decisions to AI, optimizing marketing, promotions and pricing together.

• Marketing plays a relatively small role because of highly objective AI assistants, which are incentivized to maintain users’ trust and give them the best choice.

My take

I think it is a solid prediction; we certainly have prerequisites for such an ecosystem, so this vision is likely to become an everyday reality. But I don’t think it is the whole picture.

It reminds me of sci-fi novels written 50 years ago. How did their authors imagine working with information in the future? Many of them described a global knowledge database, a convenient digital library which helps to find any information on Earth.

And they were right. Now we have Wikipedia, Google and ChatGPT. But none of them predicted1:

• Social media & scammers with $99 courses.

• Search Engine Optimization & link bombing.

• Ads personalization to the point of paranoia.

The future of ecommerce is also much messier. There are at least three reasons.

First, many users will prefer a slightly biased AI assistant with ads to an absolutely objective and ad-free one, but with $20 monthly subscription. Think about OpenAI’s recent move introducing ads on the free plan.

Second, in many cases there is no one objectively best answer. If a user wants “a t-shirt with a funny print”, how many options can satisfy them completely? Thousands. So an AI agent can suggest a product from the highest bidder without compromising user’s utility.

Third, AGI is still present only in inspired talks of AI lab founders, who need another multi-billion funding round. AI systems are trickable. They rely on the data available. Influence this data → influence AI decision. The more users trust AI, the more tempting it is for businesses to exploit this trust. So they will try and will succeed2.

What businesses should do, then?

1. Prepare for AI-driven purchases, of course, as Felix suggests (btw, here is his newsletter).

2. But at the same time invest in branding, SEO and GEO. They are here to stay until human nature changes.

The businesses that will win in 2049 are those who learn to exploit the AI ecosystem, not just scrambling for a tiny profit margin at the bottom of the food chain.

Here are some points from a November study:

• Google is still by far the biggest source of traffic.

• AI traffic on sites is still tiny (only around 1%).

• ChatGPT sends more traffic than other AI bots combined.

• Perplexity is in second place.

Other sources generally confirm these conclusions.

Before we use this data as the basis for a marketing strategy, we need to understand the limitations of such studies. There are two main problems:

1. At least part of the traffic from every chatbot is technically impossible to attribute. When we use a web analytics platform like GA4, part of referral traffic is lost to direct traffic.

2. The share of untrackable traffic is different for every chatbot.

What makes visits from chatbots trackable

There are two ways a web analytics system can tie a session to a specific traffic source.

The first is the Referer request header. It is sent from a client (browser) to a server and contains the address of the resource from which the request (click) was sent.

The second is appending utm_source to outbound links.

If neither of these methods is used, the click will be attributed to direct traffic. We can guess about the source, but we have no evidence.

How popular AI chatbots use Referer header and UTM

I conducted a mini-study to find out how different bots handle different types of links.

I asked each chatbot about real estate agencies, their reviews and current market prices to ensure that the answer contains different types of links. Then I clicked on each type of link on the desktop web and then in the mobile app.

For each case I observed whether the link has UTM and whether a proper Referer was received. The results:

Here is the full log of the study with the details for the each type of links: AI tracking

As you can see, traffic from apps is mostly untrackable. Only ChatGPT and Perplexity use UTM.

Interesting details:

• Gemini proxies some links with Google Search, the link with “domain.com” anchor sometimes has “https://www.google.com/search?q=https://www.domain.com” address.

• Perplexity is the only chatbot which makes all its referral traffic attributable (Referer on web, UTM on mobile app). It looks like a deliberate policy.

• Deepseek does the opposite: absolutely no UTM, referrers are explicitly prohibited by rel=”noreferrer” inside <a> tags.

• ChatGPT mostly behaves like Perplexity1, but in some edge cases (for example, clicks on unverified links from a warning popup) it does not use UTM nor Referer.

Conclusions

1. The real volume of AI traffic is significantly higher than 1%. A big part of mobile AI traffic is being misattributed as direct due to technical limitations and privacy policies of apps.

2. It is very possible that Gemini, not Perplexity, actually holds a solid second place in referral traffic volume. Gemini has much bigger audience, but its traffic is much less attributable.

3. Traffic from Claude and Deepseek is also likely underestimated in the studies.

Published AI traffic figures are both low and incomparable across platforms. The same is true for individual reports. The top AI referrer in your GA4 may simply track more reliably than competitors, not actually send more visitors.

P.S. I recently created a tool to make sense of messy GA4 reports for sorting referrers that aren’t hidden: https://io.trudov.com/tools/ga4-attribution-cleaner/

First, modern conversion paths are complex. Before ordering something, people consult ChatGPT, search for reviews and alternatives, then postpone the purchase till Black Friday, only to finally be swayed by a discount offer on Instagram. Increasingly stricter privacy laws don’t make the tracking and analytics any easier. As a result, it is difficult to link a specific marketing effort with sales.

Second, the most popular web statistic service — Google Analytics 4 — has an interface clearly designed for aliens from Alpha Centauri, not humans. If you are not a seasoned analyst, but a business owner with the usual myriad of responsibilities, you risk getting lost in the reports.

The solution: Key events report with Data Driven Attribution (DDA)

Google Analytics 4 has Data Driven Attribution (DDA) which uses machine learning to determine how much credit each channel should get for acquiring a lead. It assigns partial credit to each touch, but most of it goes to the channel that does the heavy lifting.

We still won’t get 100% certainty about which marketing effort brought us the particular client; it’s just not possible. But DDA solves the complexity of attribution as well as a modern web analytics system can, so we can stick with it instead of arguing whether SEO should get credit for the first or the last click. All models are wrong, but some are useful.

Setting up the report

You can track leads in GA4 with Key events1, which are the direct descendants of goals in good old Universal Analytics. Here is the simplest path2:

1. Go to the Explore tab (on the same level as Reports)

2. Create a Blank exploration

3. Add Source/medium (From Attribution) and Event name (from Event) to Dimensions, Key events to Metrics, drag Source/medium to Row, Key events to Values and Event name to Filters.

4. Set the filter “Event name” to “exactly matches” your Key event name (like “Inquiry submitted”).

Here is how it should look:

You can save this “Exploration” and quickly access it in the future.

Now we have the data. Key events values should have decimals — a sign that our attribution model is really DDA. To get the final number of leads associated with organic traffic, we need to export the report:

In this spreadsheet we should combine all key events from all the sources which we try to affect with SEO & GEO. It is not only “google / organic”. And not even just all “organic”. Google can classify traffic from chatbots or foreign search engines, like Yandex, as referral.

You need to sum all sources which are related to search or AI answers because efforts to boost visibility in LLM also impact traditional SEO and vice versa.

To speed up the process, I created a free online tool that does this summarization with a couple of clicks. Just upload the CSV export file from GA4. The tool will combine Source/medium by category automatically:

All you need to do is combine the categories Search Engines and AI Chatbots.

Understanding the results

What is the real-world meaning of the number we get by combining key events for each source in a messy export file?

It is the number of conversions we got from search engines and chatbots according to Google Data Driven Attribution.

If our final goal (as it should be) is to assess SEO&GEO efforts and link them to financial results, we need to ask two important questions:

1. How much do we trust Data Driven Attribution? We can compare the number from DDA with first-click and last-click results3. For simple user paths (for example, on affiliate websites) the difference between attribution models is usually less than 10%. In that case DDA is quite reliable. The greater the distinction, the more uncertainty in DDA we face. If the difference is around 30%, we are forced to think in probabilities and assess scenarios where the final number of leads is higher or lower.

2. What part of search-driven conversions is due to SEO effort, not intrinsic search traffic? In my previous article I argued that SEO work and budgets should be divided between basic optimization and active search marketing. So in terms of ROI we have to measure only active part (“hunting“). I will write a separate article on how to do that.

With these two clarifications you will have an unbiased assessment of how many leads you are actually getting from SEO&GEO. Every marketing specialist or agency is heavily incentivized to attribute a lion’s share of leads to the channel they work with and thus grab the credit for sales success. Data complexity allows everyone to find a way to justify this. That is why it is so important for founders and managers to be able to make their own judgements.

This report gives you the foundation. If you want to go deeper by isolating brand traffic, assessing CAC or finding highest-ROI SEO moves for your project, schedule a call with me and we'll find a way to get things done.

2

The most intuitive way to get a report of organic conversions is of course to go to Events tab and add the filters by the specific event and marketing channel:

But when we try to assign a filter, we notice that we cannot filter by channel with DDA. Only by:

• Traffic source (works on the session level → forces us to the last-click attribution).

• First user source (works on the user level → forces us to the first-click attribution).

No machine learning magic which we need to make an estimation of the real value of SEO.

See the “aliens from Alpha Centauri” logic here? GA4 makes a cool feature, which solves a real problem and then just hides it so you need a tutorial like this to make sense of the reports.

Quirks like this are one of the reasons why BigQuery, not the GA interface is the de-facto standard when we talk about any serious analytics, but if we want just to get the number of SEO leads, it is overkill, so we use Explorations.

Businesses rightfully hate spending blindly. SEOs rightfully hate being accountable for things they do not control, like a client’s developers carelessly busting a delicate redirect map.

It is no wonder that we still have no universal transparent framework for measuring SEO. Should we mourn this imperfect world and continue to treat SEO budget as casino money? No, because you can quickly solve half of the problem in two steps:

1. Realize that SEO is not a traditional advertisement channel with a single interface. It is a messy bunch of practices that serve two different goals.

2. Assign separate budgets and KPIs for each of these goals.

The two sides of SEO

1. SEO-housekeeping

Many people think SEO is a constant fierce battle for the TOP-1 spot. In reality, before we can even start dreaming about TOP-1, we have to do plenty of mundane stuff, where we need to fight only with own sloppiness1.

When we are done with implementing a fresh SEO checklist and Google Search Essentials, the site will reach the level of basic SEO hygiene. If the competition in your niche is not off the charts, you should reliably occupy the TOP-1 position for searches with your brand name and appear at least in TOP-30 for the queries that match your offers.

You also should from time to time appear in TOP-5 for long and specific keywords, if your page is objectively the best match.

In short, basic level of optimization gives you:

1. Basic visibility: a trickle of potential clients from Google and (possibly) from chatbots.

2. Brand protection: when the people you influenced via other channels google your brand, they see your site on the TOP-1 position, not your competitor.

2. SEO-hunting

After making sure we are done with housekeeping, we can start being more aggressive and target specific keyword clusters to reach the top of the search results. It is time to do link building, advanced text optimization, finding non-obvious queries and creating content for them, optimizing for click factors. Whatever it takes to get more leads.

Budgeting and measurement, finally!

When an SEO does the basics they work like a lawyer who ensures compliance. Their work does not directly bring clients. And it is perfectly normal. Problems begin when we try to tie this work to sales, mixing hunting and housekeeping. But when we separate them, everything become clear:

• Housekeeping: focus on processes. Assess the amount of required work → allocate a budget for this work → control execution and quality2.

• Hunting: focus on the results. Assess additional traffic from your target keywords → allocate budget → set Google Analytics to assess how many leads and revenue you got from this traffic → control CAC and ROI based on this budget.

Splitting these two streams of work helps you to make more grounded decisions and manage SEO contractors better (which is the real goal behind the measuring!):

• A separate budget for SEO-hunting makes it comparable with other marketing channels, so you can better decide how much you want to spend and set educated expectations for CAC. Everything becomes more familiar and intuitive.

• Basic optimization is more important at the launch, but to keep the site alive and well you need constant attention. Assigning separate (even if small) budget helps not to forget about it.

• Housekeeping is relatively straightforward; hunting requires higher skill. You can hire different people/agencies for both parts, making sure expensive hunters don’t bill you at their elite hourly rates for pressing a couple of buttons.

Of course it is not that simple. There are tons of details and nuances (where exactly does the line between housekeeping and hunting lie? what attribution model is better? how to filter brand traffic? how to deal with the inevitable lag in results? what to do with the overlap between SEO and GEO efforts?). I will publish separate articles about these nuances.

You can also schedule a call with me and I’ll help you make sense of these nuances in your particular case.

— What about burning developer hours, distracting product focus from core utility and noisy user data? Our investors are not going to like it.

— Wait, gamification is not a cheat code? And investors are actually smart?

— Correct.

To avoid the deadweight, you need to recognize the two scenarios where gamification actually works and be suspicious of everything else.

Products with a natural fit for gamification

Let’s take a look at the essence of any game. Which is:

Challenge → Achievement → Reward

So the best possible foundation for gamification is the activity where users are already involved in challenges that matter to them. All we need to do is to help them to celebrate achievements. Closing rings on the Apple Watch is a perfect example:

• The challenge: a high level of physical activity.

• The achievement: 10000 steps.

• The reward: a notification about closing the ring.

Adding a virtual reward for a real-life challenge makes the everyday grind more tolerable and adds value to the product. It is a clear win.

If your product has no such natural fit for gamification, you might be tempted to stimulate users offering valuable rewards, like free subscriptions or vouchers. It can work too, but forced gamification comes with side effects. Some customers will be annoyed with distractions, others will try to exploit the system and milk rewards. Sure, the activity metrics can jump, but is it really the activity you want?

Personal anecdote: I always find Xbox reward points irritating. Reward for playing every day? Seriously? There is no challenge and no achievement to celebrate. It is just another bothering notification.

I checked whether people are trying to take advantage of the program and of course they are. There is a whole subreddit with 45k users who try to maximize point-earning with approaches like this:

Given the scale of Microsoft’s user base, their reward system can still make sense. If you have a huge audience, even a small increase in users activity can give you a positive ROI despite large upfront costs. That’s economies of scale at work.

What if your product has no inherent challenge and for now you’re only dreaming about the economy of scale? Well, you have one more case when gamification makes sense.

Gamification as a targeted booster of sales

Here is a savvy attempt to implement gamification not for abstract user engagement, but for boosting revenue and margins from Airbaltic:

Note, they do not try to make a direct upsell like “Pay additional 100 euro for business class“. They invite you to place an offer, challenging you to guess other people’s offers and promising a relatively cheap upgrade as a reward in case you are right. And don’t forget the sweet feeling of you outsmarting the system if you won with the minimal bet!

I have no stats on how much this mini-game boosted sales, so I cannot declare this case a success. But it has all the traits of good tactical gamification:

• low-cost implementation (just one landing page upgrade + a couple of new email templates);

• used in a tough part of customer journey where traditional methods likely hit their ceiling;

• clear and intrinsically important success metrics (we either make more money or not).

Conclusions

There are two cases when gamification makes sense:

1. Your users are already playing, just do not know about it: they face challenges and crave recognition. Products related to learning, good habits, competition are a natural fit for gamification.

2. You have a specific weak point of customer journey and your gamification idea can be quickly and objectively tested.

In other cases gamification means risk of improving metrics for the sake of metrics, not for users and money.

Practical implications

• For founders: if your product has no inherent challenge, gamification is roadmap deadweight. Delay its implementation until you have a very good reason to do so (like the economy of scale in case of Microsoft).

• For product managers: consider gamification as a local tool to do heavy lifting in parts of product which now have bad metrics. Make sure you can easily measure the effect.

• For investors: if early-stage startup leans towards gamification despite its product having no intrinsic challenges and achievements, it is a red flag. Moreover, assessing the startup’s unit economics, you should separate users who pay for core utility and those involved in games. They usually have drastically different LTV.

Not sure your product roadmap is free from forced gamification or other negative ROI features? Schedule a call with me and I will help you weed them out.

This post is based mostly on my experience with ChatGPT, Gemini and Claude, but I think these problems are universal for all LLMs.

Analyzing data-heavy documents: hallucinations are inevitable

LLMs can catch general vibes and extract a high-level insights from a verbose PDF. But when you need actual insights from subtle details or numerical comparison, results are absolutely unreliable.

I asked several chatbots to compare a company’s annual and quarterly reports to find what exactly is changed. The responses looked good… until I started verifying.

Even Notebook LM, specifically designed to work with provided sources (and usually good at it), makes obvious mistakes:

Brainstorming with rich context: regurgitation loop

There are two types of brainstorming:

1. You describe a problem and ask for the solution.

2. You paste a brain dump and ask it to find underdeveloped but valuable ideas.

LLMs are very good with the first type and terrible with the second. Of course, it is hard to judge if you only work with brain dumps (garbage in — garbage out). But I have tried several times with much higher-quality material, i.e. articles good enough to motivate me to leave a comment.

I often feel that my comment only scratches the surface of an interesting topic. So I feed the article and the comment into a chatbot and ask it to suggest new directions. In 90% of times it fails spectacularly. Usually there are absolutely no fresh ideas, just a “word salad” echo of what I already wrote.

Custom instructions: sharp decline in response relevance

Custom instructions are very useful. But once you stop describing tone of voice and general approach, you risk annihilating the quality of the responses.

I tried to set detailed custom instructions, giving plenty of my personal context. I deleted almost everything very quickly, because the chatbot funneled every conversation into advice about my goals and questions. In theory it is useful.

In practice: you request to translate a menu in a restaurant, the chatbot complies but also begins obsessively warning you that some crab species are poisonous. That’s the moment you realize that “be proactive” and “my goal is health and longevity” do not always combine well in custom instructions.

I found that the quality of answers deteriorates even if you explicitly say when this instruction (or memory) should be used. For example, I asked Gemini to remind me about a comprehensive health check-up when I upload a result of a blood test. But it started to remind me about it in almost every conversation, no matter how distant from the topic.

Conclusion: more context — more information entropy

Some of these issues can certainly be alleviated through fine-tuning, advanced prompt engineering, agentic workflows. But I doubt they can be fixed completely. As you may notice, all three problems are connected to an increased amount of data. The very nature of LLMs makes handling rich context problematic. As context grows, the "Signal-to-Noise" ratio doesn't just drop; the model loses the ability to distinguish the goal from the background.

Further reading: Context Rot: How Increasing Input Tokens Impacts LLM Performance by Chroma Research.

So, naturally, when I started my management career, I thought I just needed to find the answers to those same questions—only this time, for my team.

I believed that being a boss required only ensuring that:

1. my team deeply understood their objectives and how to tackle them, and

2. my team was motivated.

Obviously, this was a naive way of thinking, so very soon I added the third element: assisting the team in execution.

But only much later I understood that this help can be significantly different for different people. And my success as a manager heavily depends on gauging how much and in which ways I should step in.

The third element is the trickiest because of the long feedback loop. If you fail to set the right target or provide motivation, you’ll know pretty quickly.

However, it’s hard to tell whether you’re effectively helping with execution or not. Maybe you control too much, or perhaps you give too much slack. Maybe your team needs more precise standards, or maybe you’re suppressing initiative. How can you tell? Eventually you will find out, but it might be too late.

You have to treat every interaction with your team as a test of a micro-hypothesis to see if it was helpful. Use two ways to test this hypothesis:

• Eagerly seek honest and specific feedback from your team: was your intervention helpful? could they have achieved the same result without it? what lessons did they learn from the situation?

• Pay attention to short-term projects. Assess small wins and failures. They might seem unimportant to the bottom line, but they bring invaluable insights.

You need both. Without feedback, you cannot boost individual performance; without routine post-mortems, you cannot organize efficient teamwork.

I tried everything: public posts, Notion templates, custom GPTs. Comically, the result was always the same. No major insights, and by March, half the goals had become irrelevant.

Last December, sitting on a train with my laptop, I finally figured out a way to make the reflection impactful. Now, a year later, I feel it has helped me shape 2025 for the better. I ended up with highly relevant goals and had much more clarity in day-to-day decision making.

Strategize like a CEO (and actually better)

To make a New Year's reflection productive, you need to plan your year the way companies plan strategy.

“Strategy” sounds serious and intimidating. But any moderately intelligent person is in a far better position to develop their personal strategy than 99% of CEOs in business planning. Just a couple of reasons:

• You have the complete picture of your life, performance, and resources. Top managers rely on reports and statistics that become outdated the moment they are created. You feel all the minute details every day.

• You have much more agency in your moves. A CEO works in a web of regulations, conflicting interests and politics. Although you have your own stakeholders (like a spouse), the final decision is always yours.

This approach is still more of a challenge than just recording wishes. But you’re absolutely up to the task.

Do not let the worn-out corporate vibe of the word “strategy” repel you. Developing a strategy simply means means figuring out a rational way to deal with the concrete situation you find yourself in.

On the most basic level, strategy = (real picture + true goals) - fluff. This already gives you clarity and a tangible edge.

Being a bit more advanced: strategy = high-leverage pivot x concentration of force + risk management.

Five questions to develop a strategy for the next year

Answer them in writing. Be honest and do not show them to anyone: the goal is to bring real changes, not to feel better.

Which external forces and circumstances influence me the most?

When we write New Year’s resolutions, we put all the weight on our supposed discipline and strive to be a better self. When a sane CEO wants to change a team’s behavior, they change the environment and incentives, reducing friction on the path toward desired behavior and increasing it for the undesired.

If you need to ride to your gym for 45 minutes, you will not go there regularly no matter how much willpower you exert. If you put a pair of dumbbells next to your desk, they will never gather dust: you’ll use them just to give yourself a bit of rest from chats and calls.

You need to go deep to understand which forces are affecting you, both directly and subtly. A work culture conflicting with your values? A person who causes discomfort but at the same time stimulates your growth? Or even an ugly broker app interface that hinders investing?

What are the opportunities I have?

What is new on the horizon you can use? This question is not only about job openings. Think about which new things in your environment can become catalysts for change according to the logic of the previous question.

Another way to identify the opportunities is to find your unfair advantage — a combination of your background, personal traits and assets which makes you especially able to compete in a certain area.

What risks do I face?

New Year’s resolutions tend to be overly optimistic. To actually improve your life in the coming year you need to think hard about risks and how to deal with them.

There are two types of risks to consider.

• Fat-tail risks. Low probability, devastating effect. To identify them, imagine yourself in the worst situation next December (sick, broke and alone) and trace how you got there.

• General fragility risks. Much more probable, the effect is slow. Eroding relationships, stagnating skills, stacking unhealthy habits. Think about tendencies you have right now and those likely to develop due to external forces (again, we need answer #1 here).

Think outside the scope of usual self-help books with their obsession of weight management and other overused themes. Your life is unique and so are the risks.

For me, one example of general fragility risks was a bunch of old projects. I had no use for my first blog, my first SAAS and several old SEO experiments for a long time. But I continued to pay hosting bills, update the CMS and so on. That’s a waste of time and a tax on cognitive load. I archived everything and freed up mental space.

What am I too anxious (or angry) to ignore?

Do not ignore, exploit. Use strong persistent emotions as a source of steady motivation. If something continuously bothers you, you'll work twice as hard to fix the situation.

Treat it like the Dark side of the Force. You need to be careful not to spiral into obsession, but channeling this power into a constructive, growth-oriented project is a clear win.

It is almost impossible to give concrete advice here. All the answers are tied to a specific personality. I myself use this heuristic: “I’m not fighting a specific problem; I’m becoming someone who doesn’t face this type of problem anymore“.

What do I forbid myself to do next year?

The last one is the killer. This isn’t the part where you lie to yourself about eating less sugar or scrolling less on your phone. It is about potentially good, but currently irrelevant goals.

Without a clear “I will not spend time and energy on these good, but non-essential tasks,” the strategy is just wishful thinking. Time and energy are limited. You need focus. You need clear priorities. And saying something like “this year I will not even try to learn Spanish” helps you stick to your priorities without overthinking.

That is where most corporate strategies fail. Managers are too afraid of being articulate and avoid cutting through the noise. You too will feel the pressure to avoid answering the question. Fight it. Be strong enough to live with an idea of yourself that is less flattering than a demigod who crushes every goal imaginable.

So, how do you construct your edge?

#1 question → you understand how your environment messes with you.

#2 question → projects and goals with high upside.

#3 question → contingencies and hygiene.

#4 question → a highly relevant project you have an unlimited motivation budget for.

#5 question → making sure you have time and focus for all the above.

Keep your notes, re-read them and make adjustments throughout the year. This will lead you to new ideas, concrete plans and impactful decisions — in your own unique way.

And a reminder about the most important part. It isn't a motivational exercise; it’s accepting reality as it is and allocation of resources. If you’ve done this right, you should feel a strange mix of relief and discomfort. It’s like realizing that gravity exists. It is here, it is inescapable, it keeps you from floating freely like a balloon. But in return, it allows you to walk with confidence.

And why would I be so certain of that? Well, I can speak from personal experience.

I’m currently working on developing a new SAAS platform for SEO audit and monitoring. Buddler.AI is a startup, no doubt, and in my efforts toward making the project take off, I spent a month searching for a person who could turn my ideas and vision into clear concepts and technical tasks.

Dozens of interviews and great candidates later, I was happy to have found a suitable person.

We hit the ground running but…

It finally hit me. Something was off.

If you want to delegate product development to a PM, you have to “infect” them with your idea and make sure that you’re both on the same page at all times.

Sharing a vision and understanding a market niche is far more complicated than creating precise instructions to be implemented by your team. And starting a new project, you’ll already have enough of other thorny issues to be busy with anyway.

There are a few other reasons why an early-stage startup doesn’t need a product manager:

1. Close communication with developers is incredibly useful for startup founders. It helps creating actionable guiding lines, enables a deeper understanding of technical challenges and ultimately can create the path for the best decisions to be made;

2. The early stages of any project are all about change. The smaller the team, the easier it will be to change the course of action if needed. Flexibility is key.

And, more importantly, when getting started with a new project, you shouldn’t be aiming to find a person with the all right job titles in their CV. Instead, you should be seeking people with a burning fire in their soul, steadfast determination, and the hunger to make your vision happen.

It’s considerably easier to find such people when you already have a product with a concise positioning as well as, preferably, some early indicators of success.

As a piece of advice I myself keep in mind and am now also sharing with other founders, best discourage the impulse to delegate essential tasks during the early stages and be laser-focused on creating a clear and practical demonstration of why vision is right - and worth working for.

Yulife offers group insurance and additional options for employees. They have access to a pretty mobile application that helps to develop healthy habits. Users also earn bonuses from Amazon and other big brands.

I've decided to invest $10000 in Yulife. There are five reasons why.

1. Yulife uses several modern trends.

It is a cool product for young employees. The app has a 5-star rating on Trustpilot.

2. Yulife is growing very fast.

It's team definitely knows how to sell. The nice app provides healthy retention.

3. The company has a lot of ways to modify and develop its product.

They could add different bonuses and new entertaining features for even better retention.

4. The potential market is huge.

Insuretech companyes may repeat the succes of Fintech companyes.

5. I have confidence in the founder.

Last but not least! I watched Sammy Rubin's presentation and talked with him via Zoom. I think he knows what he does!

My lists quickly became cluttered. Some of the tasks lost their relevance. Then there was a rush, which ruined all plans, but, having returned to my normal rhythm, I sadly discovered dozens of overdue tasks. Then the work process changed; I forgot I should adjust my to-do list to it….

Finally, the “personal effectiveness tool” turned into a demotivating mess of irrelevant tasks mixed with rough ideas. I gave up and accepted that a to-do list, which hurt to look at, is an inevitable evil.

The Painless Way to Use a To-do List

It exists! A couple of weeks ago, I said to myself "That’s enough!" and set up the work with Todoist so that it would be nice. 

I have deduced four main rules for myself and I want to share them.

1. One account both for personal and work tasks 

I am accustomed to having different to-do lists at home and at my office computer. This seemed logical until I had to work from home because of the coronavirus. Having combined the tasks, I just physically reduced the time spent planning and keeping the list up to date.

Of course, the tasks are not mixed, they are sorted out according to projects. 

2. A to-do list must be integrated as much as possible with other tools that you use

The logic is similar: it is easier to deal with one source of tasks. 

I am comfortable about Todoist automatically copying tasks from Google Calendar, everything else integrates with it. Flights and hotels are added to Google Calendar by themselves; you can create a new event from a letter, confirming your registration for a webinar in your calendar, with a couple of clicks, and you can do the same with any e-mail.

Looking at the order of business on Google Calendar, I can plan new tasks with ease.

The second rule is obvious, again; I have been using integration for a long time. 

All the magic happened to me when I began to follow these two rules and combined my work calendar with the personal one. 

Interestingly, Todoist doesn’t have the option to use two different Google Calendars simultaneously as out-of-the-box feature. If you attempt to add the second calendar, it offers you to choose from the calendars of the same Google account.

I just dragged my work calendar into a separate project in the old office Todoist account and after that shared this project to my personal account.

Thus, Todoist really collects everything that is scheduled for a particular day. This is the very moment when the to-do list starts to delight you when it transforms into a "one-stop shop".

You can see your tasks for the day and know for sure that nothing is forgotten, you needn’t wait for any other notifications and reminders.

The control of the situation is yours now.

3. The tasks are clearly tied to the day when you are really going to do them.

You must not set a due date for new tasks at random. It is better to think about what time it will be easier or more convenient to fulfil the task. Every time when there is a necessity to write down a task, something like the following inner dialog takes place: “So, I'm discussing this project with John on Wednesday; by Thursday, I'll know all the details, let’s add it here.”

There is a special section where unsorted tasks are collected by default - Inbox.

Previously, I always had it alarmingly full. Now, I keep there only the tasks for which I have not chosen the optimal time. 

There are only few tasks of that kind. Most tasks are sorted by their category and due time, or they are left without a deadline as some ideas for the future.

4. I clearly limit a set of tasks for each day.

If I get a new task, I schedule it for the next day or later. So I always have a clear plan for my day.

Also, I always have a big strategic task from the “important, but not urgent" group. After fulfilling the tasks attached to the day, I focus on the key task of the week.

You should not break down absolutely all tasks into stages and determine deadlines for them - many projects with a creative component just need to be done until your endeavor comes to fruition.

What is the outcome?

Every morning, I get down to my to-do list and feel that I want to follow it.

Trash doesn’t accumulate any more, nothing is annoing, the inevitable routine does not take too much energy. Todoist has became a really useful tool. 

Elated, I immediately paid for the annual premium subscription. I have no doubt that I will use it at maximum.

The point is not in the certain tool (I haven’t tried any alternatives of Todoist for a long time; it’s likely that there are other good options). The point is in the approach. If you use your to-do list exactly for its intended purpose - as the exhaustive list of the tasks that need to be done today - everything will be OK. Nevertheless, if you are going to try Todoist, it will be nice of you to use my referral link. If you use it, you will take two months of premium subscription for free and so will I.

The Final Insight

Writing the text, I was surprised: everything is so self-evident! I wonder how many areas of life exist where it’s very easy to put things in order - provided that you realize your habitual problems and make a little effort to handle them.

Good luck!

2005: A List of Links and a Few Fields

The first part of the page contains links and a short description of Google’s services. The form is very small and does not collect personal information.

By the end of 2005, this page included even more links.

This approach seems strange today, as the common practice now is to present the main content as high on the page as possible. Why did Google put the form at the bottom of the page?

There are two explanations.

1. The list of services both explains and highlights the value of a Google account. This was important in 2005 because a lot of people were newcomers to the Internet.

2. The list of services helps promote different Google products.

2006: The Classical Google Registration Form

The big block of links has been replaced by a single string. Probably, Google had learned that a block of links was bad for the conversion rate.

The number of fields has increased. On the screen, we can see questions about location and “Personalized Search.” This set of fields existed for a very long time, and I still associate the Google signup form with it.

2009: Small UI Changes

The form has lost its borders and looks more clear and modern. The “Stay signed in” checkbox has been added.

2012: Attractive Form with a Large Amount of Text

The 2012 page is complicated and colorful. The form is distinctly separate from the text and contains several new fields: name, birthday, gender, mobile phone, and other email addresses. All elements are carefully designed.

Let us study some interesting details.

• The form is big enough to attract attention and small enough so that the form does not look as hard to fill.

• The “Month” field is dropdown, but the “Day” and “Year” fields require text input. Why? Because there are only 12 months. Selecting one of them with two clicks is very convenient. However, it is easier to write numbers to select the days instead of scrolling through 31 options.

• The “Mobile phone country code” field is only a compact flag. Usually, Google detects countries very precisely, so people do not need to interact with this field.

2013: Less Text, More Air

2014: Goodbye, Profile Avatars!

The block with the suggestion “Set up your profile and preferences just the way you like” has disappeared, perhaps because it was not so important for users. The information about Google on mobile devices is far more prominent. This may be because A/B tests demonstrate that bright profiles distract users from the form.

2019 – now: Clean, Simple, and Minimalistic

Notes

Interestingly, sometimes Google has an old form and a new form at the same time. I am going to study this occurrence more carefully.

I have created a video with all the significant changes of the form: 15 years in 40 seconds!

1. It's ok to be small! Do not pretend you're a huge corporation!

The service started to earn money right after the launch. That time it had plenty of bugs and a very poor feature set. Its positioning was at an embryonic stage. However, the project was quite successful. Why?

I am sure that the main reason was honesty. I never said that my site is the best service in the world. I always said that it is only my pet project. I told that the service was developed by myself and that I am a non-professional programmer. I used a ridiculous domain. When I wrote about a new feature I also warned about possible bugs.

Thanks to my candor potential customers did not have inflated expectations. More importantly, they saw an alive person through the imperfect site. A person always much more interesting and involving than a lifeless interface.

2. You never know what part of your product will be popular

I spent dozens of hours developing sophisticated and unique tools. But people did not like it. They liked other tools which were much more simple.

For example, I created a tool for quick analysis of a website's sections. It combines data from several sources: search results, sitemap, web analytics system, and search engine's webmaster tools.

It is a really helpful tool. Using it is a simple way to detect a lot of SEO problems and possibilities. But this tool was used only 310 times, including near 70 times when I used it myself. It is a terrible result. Since the tool was released my site has attracted about 240 000 visits. So the tool was used only in 0,1% of the sessions.

3. The best way to attract loyal customers is catching a hot trend

Okay, there I am going to tell about the most popular (about 180 000 page views) and profitable part of service.

In March 2017, Yandex (the biggest Russian search engine) announced a new anti-spam algorithm. A lot of websites lost half organic traffic for several days. The SEO community was very anxious.

I was very interested in researching the new algorithm because I received a huge number of questions from my old and new clients.

So I carefully analyzed a large selection of affected pages and wrote an article about my study. In that article, I described several text metrics that appeared quite useful to detect over-optimization.

The article got many shares and traffic. I also got plenty of comments with request for creating a tool that could measure metrics I invented.

Or course I did it.

This tool is now available on my new project and free to use.

4. You have to create literally great manuals and it is very hard for founders

People do not know how exactly your product works. And most of them do not like to spend time trying to understand its features. You’d better carefully explain to them each single piece of your product.

But when you’ve created something you know it very well. It is hard for you to speak the same language as your customers. Remember my fail from lesson number 2. I created a valuable tool but I couldn't tell about it plainly.

The best way to solve the problem is motivating other people to tell about your product. I have a couple of ideas about it. You can:

• Reward customers who write detailed reviews.

• Ask aspiring bloggers to write about your product in return for free access. Keep in mind: it is not about an advertisement it is about an explanation of features. So you needn’t care about the size of the blogger’s audience.

5. You should think about integrations even at the beginning

People already have familiar ways to solve problems that your product deals with. If your tool could embed itself in this ways instead of replacing them you are a winner!

When I created my project's API, I got new mentions from people who use it in their products. Also, I got several new customers with much work to do and large budgets.

I think I should have created an API much earlier. And also I should have tried to integrate the API in something very popular among my audience. Google Sheets, for example.