The Situation with Cryptocurrency Mining

For those who are new to the blog, I am the principal developer of Sia, a cloud storage platform based on the blockchain. Obelisk is a cryptocurrency ASIC manufacturing company that was started by myself and some of the Sia team members about a year ago. Our journey with Obelisk has given us a lot of insight into the world of cryptocurrency mining. Our first ASICs will ship in about eight weeks.

We started Obelisk because we thought that coin developers had a poor understanding of the mining industry as a whole and that the best way to learn about it would be to get our hands dirty and create a miner.

We’ve learned a lot about mining since we started Obelisk, including about GPUs, ASICs, FPGAs, ASIC resistance, mining farms, electricity, and a slew of other topics coin developers should be more aware of. We can’t share everything we know, but we have compiled information on a few important topics that, in my opinion, will be useful to cryptocurrency developers and other community members.

ASIC Resistance

We have been pessimistic regarding ASIC resistance for a considerable amount of time, and our journey into the hardware industry firmly established our position. Hardware is very adaptable. In order to be useful for general computation, general-purpose computing devices like CPUs, GPUs, and even DRAM make significant compromises to their true capabilities. By removing all of that generality and concentrating on a single aspect, most algorithms can see significant optimization for basic hardware development.

The vast majority of algorithms that are resistant to ASICs were created by software engineers based on presumptions regarding the limitations of specialized hardware. Most of the time, these assumptions are wrong.

Equihash is probably the easiest target because a lot of people believed in the equihash algorithm and we have claimed for close to a year that we know how to make equihash ASICs that are very effective.

The key is to learn how to sort. It appears that many algorithm designers are unaware that an ASIC can combine the chip’s computational and storage components. A GPU must travel to off-chip memory, bring data to the computational cores, manipulate the data, and then return the altered data to off-chip memory when performing equihash computations.

Equihash allows you to perform the majority of your data manipulations in situ, significantly reducing the amount of energy required to move data back and forth and also significantly reducing the amount of time between adjustments to the data. Because the data manipulations that need to be performed are sufficiently straightforward, you can simply merge the memory and computation. Efficiency and speed are greatly enhanced as a result.

Naturally, when Bitmain released powerful ASICs for equihash, we were not at all surprised. Our own internal research indicated that the Bitmain ASICs would be 5 to 10 times more efficient than they actually are. This could be due to a variety of factors, but in general, we believe it is reasonable to anticipate the release of more powerful equihash ASICs in the coming months.

We also had loose plans for ethash, the algorithm for Ethereum. Although ethash was not as compatible with ASICs as equihash, products on the market today demonstrate that you can still work well enough with outdated GPUs. The majority of the other algorithms we’ve looked at have shortcuts that are even more significant than the ones you can take with equihash. Ethash is by far the most resistant to ASICs.

In the end, custom hardware that performs better than general-purpose hardware will always be possible. Everyone I’ve talked to in favor of ASIC resistance has consistently and significantly underestimated the flexibility that hardware engineers have to design around specific problems, even with a limited budget, I can’t emphasize this enough. There will always be a way for custom hardware engineers to outperform general-purpose hardware for any algorithm. General-purpose hardware is fundamentally constrained by this.

Hardfork Resistance

A lot of people think that there are three types of computing: ASIC, GPU, and CPU While these are the categories that most people are familiar with, there is only one type of chip in the chip industry: a ASIC Nvidia, Intel, and other businesses internally refer to their products as ASICs. The categories, as they are known to the general public, really show how adaptable the ASIC is.

I would like to measure flexibility using a scale from 1 to 10. We will install an Intel CPU at a “1” on one side. A bitcoin ASIC will be installed on the opposite side, marked with a “10.” Chips can be created by designers in any range on this scale. You lose a lot of flexibility when you go from a “1” to a “10,” but you gain a lot of performance. While sacrificing flexibility, you also reduce the amount of effort required for design and development. A GPU is rated a “2” on this scale.

Products that fall somewhere in the middle of a GPU and a fully inflexible ASIC rarely see development because, by the time you’ve given up enough flexibility to move away from a GPU, you usually only have a very specific application in mind and are willing to give up every last bit of flexibility to get the best performance. Another reason there aren’t many things in the middle is that designing fully inflexible ASICs is much cheaper.

The Baikal miners and the Google TPU are two examples of products that fall somewhere in the middle of an ASIC and a GPU. These are chips that can perform significantly better than a GPU in a wide variety of applications. The Baikal case in particular is interesting because it uses the same fundamental chip to replace GPUs for a large number of coins. It would appear that these chips are adaptable enough to follow hardforks as well.

Because chip designers do have the ability to make flexible chips, ranging from slightly flexible to highly flexible, with each piece of flexibility costing only a little bit of performance, the strategy of hardforking ASICs off of a network will lose its potency as more of it occurs. We believe that a Monero miner could withstand hard forks with less than a 5x performance loss because the Monero developers have pledged to maintain the same general structure for the PoW algorithm.

An algorithm with three parameters is Equihash. Any naive hardfork from Zcash to eliminate ASICs would likely require altering one or more of these parameters because Zcash mining only uses a single option for these parameters. We were able to develop a fundamental architecture for equiahsh ASICs that could successfully perform a hardfork using any parameter set. This indicates that in order to disrupt our chip, a more fundamental change would be required as opposed to a simple hardfork adjustment to the algorithm parameters. Despite this adaptability, we believe our ASIC could significantly outperform GPUs in terms of speed and efficiency. Our designs ended up on the shelf as a result of our failure to secure funding for the equihash ASICs.

The capabilities of ASICs are once more at the heart of this conclusion. Many people, I believe, were under the impression that small hardforks on a regular basis would be sufficient to disrupt any ASICs on the network and were unaware that flexible ASICs were an option. ASICs can attempt to be hardfork resistant, particularly when the changes are less significant, just as algorithms can attempt to be ASIC resistant. It may be sufficient at times.

Monero’s Secret ASICs

It was made public a few months ago that ASICs had been secretly developed to mine Monero. According to my sources, they had been mining using these secret ASICs since the beginning of 2017 and had completed almost a full year of mining before they were discovered. The group had more than enough money to try again with other ASIC-resistant coins because the secret ASICs had a huge return on investment.

Prior to being discovered, it is estimated that Monero’s secret ASICs comprised more than fifty percent of the hashrate, and nobody was aware of this. A 51 percent attack could have been carried out at any time because a significant portion of the Monero supply was concentrated in the hands of a small group at the time.

It appears that the ASICs were shaken by Monero’s hardfork. Although Monero has announced a twice-yearly PoW change, we may see another round of secret ASICs with greater flexibility. I do not believe that the ASIC designers attempted to incorporate flexibility into their ASICs. Even if you think you only have a 30% chance of your ASIC surviving the PoW hardfork, the block reward for Monero is high enough to make it worth your while to look into hardfork resistant ASICs.

My strong suspicion is that Monero will construct a second batch of secret ASICs. These ASICs will likely be more conservative and adaptable, attempting to adhere to the hard forks that Monero releases every six months.

Other Secret ASICs

There have been numerous other rumors about secret ASICs. People who own secret ASICs rarely talk about them, but as of March 2018, we had heard rumors about secret ASICs for both Equihash and Ethash, as well as for numerous smaller coins that do not yet have ASICs. Prior to the Bitmain Z9 announcement, we believe three distinct groups were actively mining Zcash with different ASICs.

We are aware of mining operations that are prepared to pay millions of dollars for exclusive access to cryptocurrency-specific design templates. Someone who has exclusive access to secret ASICs can profit millions from even low-ranking cryptocurrencies. Secret mining has spawned a nebulous underground industry as a result. Because of the high level of secrecy involved, it is largely based on rumors and previous relationships. However, it is still a very profitable industry, and even when things go wrong, like the Vertcoin hardfork, the benefits to secret miners far outweigh the setbacks.

At this point, I believe it is reasonable to assume that every Proof-of-Work coin with a block reward of more than $20 million in the previous year either currently has at least one group of secret ASICs mining on it or will soon have secret ASICs mining on it. GPU returns are the most straightforward method for identifying this, but as ASICs continue to permeate every coin on the market, this metric will become unreliable as there will be no GPU-only coin to use as a baseline, at least not one large enough to support all of the massive GPU farms that exist.

Because there is so much money at stake, the ASIC game has developed to such an advanced level. A high-risk production run can be justified by the fact that even small coins can be worth tens of millions of dollars.

Manufacturer Exposure

Bitmain and other companies that sell ASICs to the general public tend to be less vulnerable than consumers to issues like ASIC hardforks. We estimate that Bitmain spent less than $10 million to market the A3 using Sia as an example. After announcing the A3, Bitmain sold more than $20 million worth of hardware for which they spent $10 million designing and producing. Bitmain had recouped their entire initial investment and more before any of the miners had produced any returns for their customers.

Bitmain is unaffected by a hardfork in this instance. The fact that Bitmain made money from Sia is beyond the developers’ control. In addition, it appears that was the situation with the Monero miners that Bitmain announced. Although Bitmain didn’t even get a chance to talk about the miners until after Monero talked about their hardfork, it seems like they sold enough old hardware to cover their costs and make a good profit.

The manufacturers are heavily favored in the mining game. They have greater knowledge of the industry’s state than anyone else, and they have control over the supply and production of hardware. A miner’s profitability is largely determined by variables that the manufacturer controls without sharing with the public.

We witnessed Halong’s Decred miner “sell out” of an unidentified batch of 10,000 miners. After that, it was discovered that more than half of the mining rewards were being transferred to a single address that was known to be associated with Halong. This indicated that they did indeed keep the majority of the profits and hashrate for themselves. Our investigation of the mining equipment strongly suggests that the equipment’s total manufacturing cost is less than $1,000. This means that anyone who paid $10,000 for it received a significant profit premium from the manufacturer, allowing them to produce nine more units for themselves. Beyond this, the purchaser has no idea how many were sold or where the shipping difficulties would be. The buyer does not, but the manufacturer does know if the buyer will be able to return the item. The customer has complete faith in the manufacturer.

If a cryptocurrency like Sia has a $10 million monthly block reward and a batch of miners is expected to last $120 million, a company shouldn’t expect to make more than $120 million from building miners. However, manufacturers can actually produce significantly more.

In the case of Bitmain’s A3, a limited number of miners were made available to the general public, and shipping took less than ten days. Shortly thereafter, YouTube videos of individuals who had purchased the miners and were, in fact, earning $800 per day from them began to circulate. This set Bitmain up for a very successful batch 2 by creating a great deal of mania around the A3.

Although we are unsure of the exact number of A3 units that were sold, we have a strong suspicion that the profit margins on their batch 2 sales were greater than the potential block reward from mining with A3 units. That is to say, Bitmain sold mining rigs worth more than $100 million despite the fact that even assuming free electricity, the block reward was insufficient to allow customers to recoup their investment. They have done something similar before with the Dash miners, and this is not the first time. We call it flooding, and it is yet another illustration of the perilous imbalance that exists between customers and manufacturers.

At the end of the day, manufacturers of cryptocurrency miners sell money printing machines. A business that only wants to make as much money as possible will sell a money printing machine for more money than they think they can print themselves. The buyer must comprehend the manufacturer’s decision to sell the units rather than keep them for themselves.

There are a few good reasons why a manufacturer would be better off selling a money printer than keeping it. The first is capital because manufacturing is a time-consuming, expensive process. If the manufacturer does not have sufficient funds to construct their own units, it makes sense to sell them and use the proceeds for production. It comes down to the manufacturer trading in future profits for present profits, which is a common financial transaction.

The price of electricity to run money printing machines is yet another reason why the manufacturer might decide to sell them instead of operating them. If the manufacturer is only able to get a certain electricity deal, there might be someone else who has cheaper electricity or better datacenters and would be willing to buy the units at a higher price than the manufacturer values them at. However, unless you have access to favorable electricity deals or are otherwise operating a cutting-edge professional business, you are unlikely to perform better than the manufacturer.

Lastly, the manufacturer might have a different reason for wanting money quickly rather than investing in hardware over a longer period of time. However, this is probably not the case in cryptocurrency mining because miners typically have a shelf life of less than two years, which, for a business, is not a very long time to wait for healthy returns.

Manufacturer Agility

In the conventional world of chip development, it takes approximately two years to launch a development effort and release a chip. In the case of the Sia and Decred miners that we built, it looks like it will take approximately 13 months to complete the project and deliver the product. I think we could complete the task in about nine months if we had to repeat it.

The chip’s custom routing consumes a significant portion of the time. Place-and-route, a much faster method of chip development, reduces development time by approximately three months but results in chips that are two to five times slower than those produced by a full-custom team. We believe that our product delivery timeline could be reduced by close to six months if we utilized a place-and-route design approach.

We estimate that Bitmain spent approximately five months developing the A3 miner, and Halong spent approximately nine months developing the B52 miner. Given the relatively poor performance of each, we suspect that place-and-route methods were utilized to complete both of these.

Those are the development timelines for a new chip. The timelines are significantly shorter if the objective is to chase a hardfork. There are a lot of shortcuts you can take to cut down on the amount of time it takes to get your product to market if you know in advance that you will need to redesign your chip. A good team with a well-planned base architecture can probably finish designs in about two weeks if they need to change a design to accommodate a change. You can acquire a new set of chips in approximately 40 days from that point on with assistance from a hot lot. The next step is to package these, which will take about a week, and send them to the manufacturer for assembly. The final step is to transport the units to a datacenter and begin mining.

We believe that you could, at least theoretically, upgrade a chip to adapt to a hardfork and have miners mining on the new hashing algorithm within about 70 days if you had all the wafers, parts, and everything reserved in advance. In practice, Bitmain would probably need three to four months to adapt an existing chip to a hard fork, and they would need four to five months if they hadn’t reserved wafers in advance. These numbers can probably be extended by another 30–60 days by any company that is not Bitmain.

Economies of Scale

Some people are already well-versed in the circumstances surrounding economies of scale. Each dollar works harder the more money you spend. I have been able to observe this effect at every level of scale, from billions to tens of billions of dollars, and it persists at all of these scales.

Volume orders are the simplest manifestation of this. You can get a single price if you order one hundred thousand heatsinks. You can save money if you order one million heatsinks. The cost goes down as you keep scaling up. This effect occurs when manufacturers reach the point where they are able to purchase and dedicate equipment to your order, as well as maintain that equipment at 100 percent efficiency. This effect is present for nearly all hardware components. In addition to cost savings and increased customization and specialization, as you scale up, your products also become more cost-effective.

It only makes sense to buy out all of the manufacturer’s capacity at some point. Equipment costs make up a big part of the cost of making something. The effective cost per component of equipment that is in use 100% of the time will be two times higher than that of equipment that is idle 50% of the time. You can start getting fully dedicated equipment that runs continuously for you as your order volumes and lead times grow, bringing the price even lower.

In a similar vein, that equipment must be manufactured by someone. The manufacturer can start to dedicate pipelines to you and keep their own equipment operating 100% of the time if you scale up to the point where you are constantly ordering specific equipment. As a result, the equipment you use for manufacturing is becoming cheaper now that it is always in use.

This is only the beginning. Each provider, manufacturer, etc., at each step. is going to charge a profit, which is usually about 30%, depending on how common your orders are. If you have sufficient funds, you can begin vertical integration, which involves either purchasing your manufacturers outright or establishing your own entity without margins.

There are numerous steps involved in hardware. There is the acquisition of raw materials like oil and iron, their refinement, and the production of base parts that can be sold for more general goods. These base parts typically have lead times of more than six months, so suppliers typically keep a lot of them in stock so they can deliver parts to customers more quickly. Because each step focuses on general-purpose parts rather than a specific product, it typically introduces inefficiency and a middleman. Costs will be reduced, lead times will be reduced, product quality and performance will be improved, and you will be ahead of anyone with that kind of scale if you have a specific product with enough volume and scale to justify dedicated supply chain elements.

To give you a rough estimate, I believe you can save approximately 30% per component for every ten times the amount you spend. That is to say, you might be able to acquire mining units for $500 each if you spend $100 million on them. In addition, if you have more money to spend, you can reduce the cost of mining units from $1 billion to $350 per miner. Your per-miner cost may then fall to $245 if you reach $10 billion. However, not only are your mining machines becoming more customizable and more efficient, but they are also becoming less expensive. A quality moat is built in addition to a dollar moat built with scale.

Playing Dirty

When we first started Obelisk, a number of different sources contacted us to warn us that Bitmain plays dirty and that we would be stopped if we tried to manufacture in China.

As a result, we discussed the issue with everyone we worked with and proceeded cautiously with a Chinese-owned American manufacturer. Since manufacturing was going to be one of our most significant expenditures, the prices were less than half of what we would have paid to manufacture in the United States.

We hid the manufacturer’s name from our website and any public data, and we were very careful about who we gave the manufacturer’s name to privately. We did everything in our power to keep the entity separate from Obelisk. Wherever possible, we placed parts orders through a separate entity.

Our manufacturer contacted us after hours on a Friday night, after any reasonable time had passed, and informed us that they would not be able to manufacture for us with little notice or justification. Our attempt to manufacture in China had failed miserably, as we had been warned. We are estimated to have lost more than $2 million as a result of this setback.

There is absolutely no proof that Bitmain was in any way involved. Other businesses have contacted us to confirm that they have had similar experiences, but neither of them had any tangible evidence that Bitmain was involved in any way. Since, in contrast to the majority of the other things I’ve been saying, we really only have a bunch of warnings that ended up being correct to inform us, I honestly wasn’t sure whether to include this section in the blog post.

However, it has been suggested to us from all sides that Bitmain has been and will continue making moves within our supply chain to ensure that we cannot succeed and that they do the same with all of their competitors. It is well established in the industry that Bitmain plays dirty.

Mining Farms

Manufacturers and economies of scale perhaps do not predominate in mining farms. Good electricity deals typically come in smaller packages, are distributed globally, are challenging to locate, and each situation is unique. As a result, it has been challenging for larger businesses to develop a system for snatching cheap electricity from around the world. Instead, the smaller parties that hold the datacenters with the least expensive electricity and hashrate tend to be the ones with the most.

The average professional mining farm pays between 4 and 6 cents for electricity, and another 3 to 6 cents for management and maintenance, according to what I can find. For large-scale mining farms, the median monthly cost is probably somewhere around $50 per kilowatt. We anticipate this number to fall closer to $35 per kilowatt per month (including maintenance, land, taxes, etc.) as techniques improve and the industry expands. between 2019 and 2020. Unless the price of cryptocurrencies continues to rise rapidly over the course of the next year, we do not believe that anyone paying more than $80 per kilowatt hour per month will be able to compete.

According to our findings, the top 5% of miners appear to be earning less than $20 per kilowatt per month, and the top 20% of miners appear to be earning less than $35 per kilowatt per month already. If the price of Bitcoin were to drop significantly, these mining operations, in my opinion, would be able to continue operating, while anyone paying $50 or more would be compelled to close their facilities.

Based on everything we have seen, we estimate that Bitmain is around $30 per kilowatt per month. It is extremely difficult to know where Bitmain is at. That is to say, they are performing better than the average mining operation, but they are by no means elite.

The Chip is not Important

The chip itself seems to be the primary focus of the majority of mining startups. Based on what we’ve seen, the chip only tells part of the story. So, while the chip is important (apologies for the title), you won’t be able to compete as a manufacturer if all you have is the best chip in the world.

The ultimate objective for miners is to perform as many hashes as possible while spending as little money as possible. Hashrate can be obtained for less money by purchasing faster chips. Additionally, obtaining hashrate requires less electricity thanks to a chip that uses less energy. However, you aren’t just spending money on electricity and chips. You spend money on PCBs, controllers, ports like ethernet ports, power supplies and power management, enclosures, fans, datacenter shelves, and other things.

The chip is just one part of the equation for successful mining, after all. You will end up with a chip that will cost you money if you don’t consider the whole picture. Butterfly Labs’ design of a high-performance chip that generated hundreds of watts of heat was one of many factors that led to their demise, and this was one of them. Bitmain chips, on the other hand, typically consume about six watts each. Butterfly Labs had to deal with expensive, cutting-edge, unreliable cooling systems, whereas Bitmain was able to throw on a forest of fat heatsinks. As a result, their powerful chip was late to market and too expensive to operate.

Bitmain is Impressive

Bitmain is frequently underestimated. Yes, they dominate due to their economies of scale and have the most money. However, they also win because they have the quickest time to market of any company. They dominate because they employ the best cryptocurrency chip developers. They are the dominant force because they have innovated in dozens of areas to eliminate costs and inefficiencies from areas that the majority of people are unaware of. They pay well and hire the best people. In addition, they put in a lot of effort to guarantee their dominance at each iteration.

There is not much more to say at this point. I think that a lot of people underestimate Bitmain or think that they can’t keep up without playing dirty because they play dirty. But that is incorrect. They play dirty because they know they can get away with it and because it is yet another place where they can improve their business. Additionally, they optimize every other activity. Understanding mining necessitates an appreciation of the impressive, highly skilled, and refined entity that controls the majority of mining today.


The most important takeaway from this is that mining is only for big businesses. There is no simple way to alter the equation, but spending more money gives you an advantage. At the very least, with traditional consensus in the style of Nakamoto, a large entity that produces and controls the majority of the hashrate appears to be the outcome, and at best, there are two or three major players who are all on equal footing. However, I don’t believe we’ll ever see a situation in which many manufacturing companies all produce miners that are competitive. Centralization is just a byproduct of manufacturing, and it occurs under many different conditions.

Even though this is bad news, Bitcoin and other cryptocurrencies based on Proof of Work are not doomed. Although decentralizing hashrate is desirable, there are numerous other mechanisms and incentives in place to constrain monopoly manufacturers. The situation with Bitcoin and Segwit2x is a great illustration of this. Despite the fact that more than 80% of the hashrate was clearly in favor of activating Segwit2x, the motion as a whole was unsuccessful.

To deal with a hostile hashrate base, cryptocurrency developers and communities have access to numerous additional tools, such as hardforks and community splits. The owners of hashrate are aware of this, and as a result, they take care not to act in a way that could provoke a revolt or jeopardize their healthy revenue streams. As developers and inventors, we can continue to work on structures and schemes that are secure even when the hashrate is all pooled into a small number of places now that we know to anticipate a largely centralized hashrate.

New to trading? Try crypto trading bots or copy trading on best crypto exchanges

The Situation with Cryptocurrency Mining was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.

Why Visitors Immediately Exit Your Website

Let’s fix it!

Image from Google

The most important metric that you should be tracking on your website is its bounce rate.

It is a number that tells what percentage of visitors leave your site after browsing just one page or within the first 10 seconds (the exact definition depends on the web analytics tool you use). The higher the bounce rate, the higher the number of potential customers you lose because bounced visitors think your website has nothing to offer them and leave without any further interaction. With ever-reducing attention spans and ever-expanding options online, your visitors have very little motivation to actively explore your website for what you offer and how it benefits their day-to-day lives. So, it becomes the responsibility of the first page a visitor lands on to convince him that spending time here is worth it.

Optimizing (reducing) bounce rate is thus tremendously important to your business. There are multiple reasons why visitors leave immediately after arriving; the most prominent amongst them being:

  1. Expectation mismatch: Your visitors are expecting to see something on your website while your page talks about something else. This happens often, especially when you have paid advertising for a specific offer and you link that advertisement to a generic page such as your homepage. Matching the expectation of your visitor is very important. Your site might be talking about multiple different things, but a visitor arrives on it to learn more about what the source said you offer.
  2. Organic search irrelevance: Search engines are getting incrementally better at finding relevant content for a user search query, but they are still not perfect. I’m sure while reviewing your web analytics reports, you must be regularly surprised by: ‘how come this search query found this page on my website? Your visitors too feel the same. Try comparing your bounce rate for organic visitors (those who came via search engines) versus non-organic visitors (those who came via other sources); you would see that the former metric is higher than the latter.
  3. Your website sucks: Your visitors expect a visually appealing, easy-on-the-eyes website upon arrival. They are already sick of advertisement-loaded, poorly made websites all over the Internet and if yours is no better, they won’t be very happy about it. Give your visitors a pleasant surprise by having a website with the right contrast, the right typography, the right layout, and the right color scheme. Hire a top-notch designer and pay them whatever you can but please make sure your website looks good.
  4. Lack of call to action: Well-designed and drafted Call-To-Action (CTA) banners can make a huge difference in conversion rate. Missing or badly designed CTAs perhaps be the single biggest reason why visitors bounce from your website. Once visitors have gone through the page they landed on, don’t let them struggle with what they should be doing next. Guide them to the actions that you think are optimal for that page. If it is a blog, you want them to subscribe to blog updates. You may want them to go through relevant case studies and whitepapers if it is a corporate site. In a nutshell, don’t let them think too hard about what their expected next action is on this website. Guide them gently using calls to action placed prominently at the right places on the website. Mostly, these right places are the ones where a visitor has just completed their original purpose (for which they landed on the site) and is wondering what to do next.
  5. Too many options: Having too many options for a visitor can also lead to a higher bounce rate. This is partly the reason why you will observe that the bounce rate on your homepage is probably higher than on your inner website pages. Having too many links/calls to action competing for visitor attention can increase anxiety and lead to visitors leaving the website in search of a better alternative. This is where experienced conversion optimization firms (such as VWO) and good web designers help you in creating a proper layout with different calls to action appealing to different kinds of visitors, hence reducing bounce rate.

How to fix the high bounce rate?

Reducing the high bounce rate significantly is possible. As different websites serve different goals and cater to different audiences, there is no sure-shot way of fixing the bounce rate. For instance, LA Tourism successfully reduced its bounce rate by 43%.

Though there are several general methodologies you can try for reducing the bounce rate:

  • Segment bounce rate by landing/entry page: your website’s overall bounce rate conveys absolutely no actionable information; it is vague and imprecise. The best way to get a true picture of your website’s bounce rate is to see the bounce rate for each landing/entry page. Using your web analytics tool, see which are the top 20 landing pages on your website and what their bounce rates are. You will be surprised to know that there is a dramatic difference in bounce rate across different landing pages. Your top priority should be to fix or optimize pages/categories which are most trafficked and have the highest bounce rates.
  • Surveys: there are many tools on the web which allow you to survey visitors who are about to bounce, just before they leave the website. Though I particularly don’t like such methodologies because they frustrate an already unsatisfied visitor, you may find them useful for your website and audience type.
  • Visitor moves and heatmaps: products such as VWO Insights help you to record mouse movements, clicks, scroll activity, and keypresses of your visitors to find out what exactly they do once they arrive on your landing pages. It can be a great way to find out that, for example, most of your visitors don’t notice your ‘SIGNUP NOW’ button in the sidebar (And you thought people aren’t interested in the offering). Similarly, you can see heatmaps to find out where exactly on the page visitors are clicking/engaging and if it is optimal.
  • Testing: The only way to find out what works is to test it. You should set up a split test to try multiple different website designs, layouts, styling, calls to action, etc. Hire a testing agency, if you wish, but make sure you are doing testing on your website continuously to always reduce the bounce rate.

What are your strategies for reducing bounce rates? Do you think you are doing a good job on your website (as far as optimizing bounce rate is concerned)? Are you satisfied with your existing bounce rate?

The answers to these questions can only be found through systematic and continuous testing on your website. Once you have the test results in your hand, make sure you focus your website optimization efforts first on the bounce rate and then on other metrics.

Why Visitors Immediately Exit Your Website was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.

VGG16 and VGG19 CNN Models: How to Extract Features, Display Filters, and Create Feature Maps

Learn how to extract features from VGG16 and VGG19 CNN models, display filters, and create feature maps.

Continue reading on Medium »

A Hint of the Future: How ChatGPT3 Will Revolutionize Marketing and Advertising

When you hear the word “AI,” what comes to mind? Do you recall Terminator? The Matrices? Or how about the aliens in Star Wars, who are…

Continue reading on Medium »

Utilizing Game-Based and Gamified Learning to Integrate Web3 Into Ledger’s Educational Programs

Examine the pedagogical approaches used in Ledger educational programs — Web3 Best Practices in depth

Continue reading on Medium »

Businesses fade away in cryptocurrency, but protocols endure.

The largest cryptocurrency lending platform, trading firm, broker, and second-largest cryptocurrency exchange all failed. Cryptocurrency…

Continue reading on Medium »