ENS Marketing Research Pilot Episode: ENS Econometrics

The full version of the study is available at the link: Medium

Hey there, DAO!

Despite the direction of long-term development of ENS, it is necessary to detect, understand and solve marketing problems. The question is not only about money and investment attractiveness, but also about mass adoption of both the industry in general and the Ethereum infrastructure in particular.

Approximately half a year ago, I observed a significant diminution in consumer engagement with ENS’s principal offering, namely web3 names. Intrigued by the attributes and underlying causative factors of this decline, I embarked on an econometric investigation, the comprehensive details of which are accessible via the aforementioned hyperlink.

Herein, I present a concise synopsis of the study.


Initially, it was imperative to ascertain whether the observed decline conformed to the mathematical definition of a ‘downtrend.’ To achieve this, I employed autoregressive analytical techniques and statistical hypothesis testing, ultimately confirming the existence of a localized downtrend spanning nearly a year (48 months).

The salient features of this downward trajectory are as follows:

  • It is monotonically decreasing, signifying that the decline exhibits a consistent structure and, barring infrequent anomalies, persisted throughout the period under scrutiny.

  • The trend is characterized by a logarithmic configuration, implying that its impact was most pronounced during the initial phases and subsequently exhibited diminishing marginal effects. This is advantageous for us, as within the confines of this model, it is unwarranted to anticipate increasingly severe declines. Any uptick in sales would either attenuate the monotonicity of the trend or signify its dissolution, heralding the onset of a new localized upward trajectory.

  • The localized downtrend exerts a moderating influence on the overarching positive trend. While it is premature to entirely dismiss the global hypothesis positing a long-term increase in sales, the nature of the global trend has undeniably shifted. Prior to the recession, it could be described as monotonically increasing, robust, and positive; post-recession, it has transitioned to a non-monotonic, attenuated, albeit still positive, state.

In the visualization you can see the segmentation that I gave as an example of what the long-term local trends were. Green marks the period when we were certainly growing monotonously. Gray - when we moved without a trend. Yellow – when the trend was not monotonous, but we were still moving up. Red – the current, pronounced, monotonous downtrend.

Within the context of the delineated challenges, a pivotal query necessitated exploration: Is the observed phenomenon an unprecedented occurrence, or could analogous patterns have manifested previously? Given that the subject under investigation pertains to the performance metrics of an economic entity, the potential for seasonal fluctuations in user engagement could not be discounted.


To scrutinize the element of seasonality, autocorrelation analysis was employed. This analytical modality facilitates the examination of temporal correlations within distinct segments of a time series.

In the course of evaluating the autocorrelation function, a hypothesis was posited: Should the data exhibit recurrent patterns at intervals of approximately 48 weeks, the current sales contraction could be attributed entirely to seasonal variations. If, however, the cyclical nature of the data is of a lesser magnitude, a partial attribution to seasonality could be considered.

The empirical analysis divulged that sales of names do exhibit periodicity, characterized by seasonal durations of 18, 23, and 29 weeks. Nonetheless, the salient inquiry persists: Does this identified seasonality specifically account for the current trends?

To address this, temporal segments corresponding to the problem period and spanning 18, 23, and 29 weeks were isolated and juxtaposed against analogous segments from preceding periods. This comparison was executed via two methodologies—seasonal differentiation and correlation coefficients.

Subsequent to this analytical endeavor, it was ascertained that while seasonality exerts a general influence on ENS sales, it fails to elucidate the extant issue.

For the purpose of cross-validation, the Seasonal AutoRegressive Integrated Moving Average (SARIMA) machine learning model was deployed. A comparative analysis between the model’s forecasted values and actual data revealed a significant divergence. Had the current period been genuinely influenced by seasonal factors, the trajectory would have either maintained growth or exhibited an initial contraction followed by resumption of growth.

Consequently, the inference was drawn that the current decline is not attributable to autoregressive variables, at least based on the historical data presently available. This suggests the presence of additional, yet unidentified, contributory factors.

Influential variables can be broadly categorized into three distinct types: autoregressive factors, which we examined in the initial chapter, external factors, which encapsulate market-wide behaviors, and internal factors, which are reflective of the organizational activities of the company. The next segment of the research is dedicated to the examination of external determinants.

Influence of External Factors

Three metrics were selected as proxies for external quality indicators:

  • Total Market Capitalization: Despite its speculative attributes, this metric serves as a robust indicator of mass cryptocurrency adoption and generally mirrors the actual market landscape.

  • Average Daily Transactions on the Ethereum Network: This serves as a measure of specific interest in the network where ENS operates and demonstrated superior modeling performance compared to the number of active users.

  • Fees (Network Commissions + Gas): This variable directly impacts consumer purchasing decisions. Although name prices could have been incorporated into the equation, their generally stationary nature renders them less explanatory.

Regression analysis was employed as the analytical instrument to discern the extent to which sales are contingent upon these external variables. The analysis aims to elucidate the following attributes of dependency:

Temporal: External forces not always instantaneously impact marketing metrics; often, there is a lag that necessitates investigation. Understanding this temporal aspect will pragmatically inform the time required to prepare for shifts in sales levels.

Functional: The relationship between marketing metrics and external variables may not always be linear. For instance, a 10-point increase in a market indicator could correspond to a 100-point surge in company metrics, indicating a power-law relationship, or vice versa, suggesting a root relationship. Identifying the functional form of this dependency will provide insights into the expected degree of responsiveness to external stimuli.

Dynamic: The influence of external factors is unlikely to be uniform over time; it may fluctuate. The objective here is to ascertain, primarily through visual means, whether such dynamics are homogeneous.

Proportional: Employing a positivist approach in interpreting the results, I acknowledge that while my models may not be flawless, any hypothetically superior models would offer a more accurate depiction of the influence of external factors on marketing metrics. Consequently, the influence of controllable internal and autoregressive factors would be comparatively less significant. Utilizing model error descriptions, the proportionality of external influences on marketing metrics can thus be determined, along with the corresponding influence of internal and autoregressive factors.

To delineate these characteristics, a structured algorithm was employed, exemplified below in the context of sales analysis.

  1. Initial Model Construction: We build the initial model based on linear multiple regression using the least squares method. In this way, we evaluate the descriptive ability of the model, the significance of the regressors-variables and the problems of the model. For sales, this stage revealed inadequacies in the model’s explanatory power, insignificant variables, and violations of model assumptions such as non-normal distribution, heteroskedasticity, first-order autocorrelation, multicollinearity, and the presence of outliers.

  2. Paired Models: We build paired OLS with each of their independent variables and try to determine whether there is such a variable that pulls down the indicators and alone gives rise to any problems, and if such a variable exists, we remove it. In the case of sales, at this stage, there were no variables that stood out and we left them all.

  3. Segmentation: We perform segmentation to determine whether there may be a time period that had much better explanatory power and would still be relevant today. For example, if totalmarketcap did not explain sales for the first 50 weeks of the chart, but explained the last 175 weeks perfectly, we would not consider the first 50 weeks in the model. But if the situation were reversed, for example, the first 175 weeks did not explain anything, and the last 50 did, we would not remove anything, since this would be a loss of too much historical data. Or if the first 175 weeks were well explained, but the last 50 were not – nope, because in that case we did not describe the current situation. In the case of sales, it turned out that at some points in time external factors actually described them better, but such areas were either at the beginning or in the middle. And in the case of commissions, the model is indeed relevant, but only in the last half of the chart, so it was decided to leave it unchanged.

    e.g. total market cap segmentation

  4. Lags: We conduct a lag analysis to determine the temporality of the dependence. We sort of shift the independent variables into the past and look at the delay with which sales react to them. Then we analyze visually and conclude whether this increase in dependence is random or systematic. For sales, it turned out that they respond to the totalmarketcap with a delay of 25-46 weeks, with an optimal delay of 37 weeks. For the number of transactions – 69 weeks. On commission - instantly, which is logical.

  5. Segmentation on Shifted Data: We repeat step 3, but on shifted data, using the same principle. And here we find out that the first 28-33 weeks of the total market cap indicator (shifted by 37 weeks) only worsens the model.

    e.g. total market cap segmentation (shifted, 37 weeks)

  6. Revised Model with Segmented and Shifted Variables: We repeat step 1, but with segmented and shifted variables. Thus, we take the sales segment Y(69)…Y(225). The “totalmarketcap” indicator is shifted by 37 weeks without the first 31 weeks, so we take X1(32)…X1(188). The transaction indicator shifted by 69 weeks, that is, X2(1)…X2(157). And the section of the commission indicator X3(69)…X3(225) simultaneous with sales. We build a model and test it for increasing explanatory power, the significance of variables and problems. In the case of sales, we actually improved the descriptiveness, but did not solve the problems and insignificance of the coefficients. We carried out the following steps with the insignificant X3 (commission) used and concluded that it only degrades the quality of the model, so we removed it at this step.

  7. Addressing Model Assumptions: We solve model assumption problems by manipulating variables. First, we define a functional dependence, and transform the graph of the dependent variable using the inverse function of this dependence. In the case of sales, they have an exponential relationship, and accordingly we transform the sales graph using the natural logarithm. Then we solve the problem of multicollinearity - the dependence of the regressors on each other, using the method of principal components, combining X1 and X2. We then get rid of outliers—critical, out-of-system extreme sales values—by identifying them with Fisher’s z-test and imputation using k-nearest neighbors. After all this, we build a model and understand that all the prerequisites of the OLS method are met, except for the autocorrelation of first-order residuals, therefore we cannot use this method to interpret the relationship.

  8. Final Model Construction: We build final models that solve or take into account existing problems. With autocorrelation of residuals, we have two options. This is either to build models that take it into account, or models that solve it. But the problem is that all models that would solve this problem are based on adding the factor of influence of the independent variable on itself. Thus, 4 models were built: 1 without taking into account the previous values, 3 – taking into account. The first model is a generalized least squares method with a covariance matrix compiled by the Cholesky decomposition of the matrix of autocorrelation values. The following models - taking into account autoregressive factors, are presented in different variations, but for example we will take ARIMAX - the same machine learning model that we used above, but with the addition of exogenous factors.

    GLS Predictions

    ARIMAX Predictions

  9. Error Analysis: Thus, we have the “exogenous factors – sales” and “exogenous and autoregressive factors – sales” models. Now we can take the absolute percentage error of the second model, but precisely when comparing it with the original data without imputation emissions, and assert that this is the share of the influence of marketing on sales. And taking the first model, its absolute percentage error minus the absolute percentage error of the second model we can consider the influence of autoregressive factors. And 100% minus the percentage error of the first model will be the influence of external general market factors. Thus, we can create a component diagram using the average values for all historical observations.

    Marketing <=> Autoregressive + Exogenous Factors proportion (ARIMAX)

  10. Marketing Effectiveness Evaluation: Now we can evaluate the effectiveness of marketing within this proportion by week. To do this, we use linear interpolation, where the starting point will be the moment of the maximum percentage overestimation of the predicted values by the second model. This will be 0% effectiveness of marketing influence, because in this case the mathematical expectation was maximally higher than the values observed as a result. And as the end of the countdown, that is, 100%, we will consider the moment when the model maximally underestimated the value of sales. Thus, we obtain such a graph of marketing effectiveness within its 23.5% average share in sales dynamics.

    Weekly Marketing Performance (ARIMAX)

  11. Finally, we can get a historical graph that reflects how many sales individual groups of factors brought us.

Based on the comprehensive analysis conducted, several key insights have been gleaned:

  • Temporal Shift in Dependency: There exists a temporal lag in the relationship between name sales and external factors. This is advantageous as it provides us with the opportunity to preemptively address associated challenges.

  • Function of Dependency: The dependence is exponential. From the point of view of the absolute number of sales, this can have a twofold interpretation: on the one hand, market growth gives us multiple growth, but even with a fall in market indicators, we are heading down at the first cosmic speed.

  • Dynamics of Dependency: The dynamics of sales are largely congruent with market trends, indicating that our sales performance is closely tied to market fluctuations.

  • Proportion of Factors: Perhaps the most disconcerting revelation is that our influence over sales is relatively limited, accounting for only approximately 23.2-23.5% of total sales. The remaining influence is exerted by external market factors and past performance metrics.

These findings underscore the critical importance of external market conditions in shaping our sales trajectory and highlight the limited scope of our internal control. This necessitates a strategic reevaluation to enhance our resilience against market volatilities and to capitalize on favorable market conditions.

What should an ideal picture look like?

There is no point in taking the proportions of classic business as a standard, since we are in a new and rather chaotic area. All we can say is that the public perception of the ENS product is dominated by a speculative image.

But is this really a problem?

Let’s imagine three hypothetical situations.

The market is in a bullish phase, everyone is interested in web3. After a certain lag (as we found out earlier), ENS sales go up sharply. We can really either relax or use marketing as a success multiplier and get more people in this bullish phase. And then fix sales at a consistently high level, if we turn the outliers into systematic levels of sales (autoregression factors will already play a role here).

The market is in a bearish phase, everyone is screaming that crypto is a scam. Here we can relax and wait for the next bullish phase. Well, or use marketing influence as a shock absorber so as not to end up in a situation where it is impossible to implement the entire development route map due to lack of funds.

The market is in stagnation, no one knows where the development will go next, as is happening now. Here we can relax and wait for the outcome of the situation together with everyone. Or we can take this as the most powerful opportunity to finally disengage ourselves from speculative perception. If the whole market is hibernating and we are increasing our influence, then our product is not just a money-making tool or a beautiful trinket for people who have nowhere to spend their money, but a really important part of the web3 infrastructure.

The marketing effectiveness rate averages 60–65% and very rarely falls below 50%. This is certainly good, but it should be remembered that these are only values ​​that lie within the 23.5% range within which we influence sales through marketing. In order to positively differentiate ourselves from the market and autocorrelation factors, we must systematically go beyond 100% of this range.

But how can you get beyond 100% above the range if we take into account the influence of “yesterday’s success”?

From the judgments about the breaking of the trend, I believe that autoregression factors do not obey the will of the actors. Our task is to stabilize sales at a relatively high level by turning statistical outliers into a system. In this case, we can say that the influence of autocorrelation factors is our merit, and not a derivative of industry phenomena. That is, roughly speaking, we need to positively violate the expectation of the models until we manage to gain a foothold on the values that are currently extreme.

Extrapolation of Regression Data to the Conversion Funnel

Using the algorithm described above, I analyzed not only sales, but also:

Search queries, as this metric reflects user interest at the awareness and interest stages of the conversion funnel.

Unique visits as they reflect stages of consideration, evaluation and intent.

You can see the analysis of these variables in the full version of the study.

This is what extrapolation of the data received from the algorithm to the conversion funnel looks like in a logical diagram:

Based on the diagrammatic representation, the following conclusions can be elucidated across different stages of the customer journey:

Pre-Awareness to Awareness

  1. Minimal Marketing Influence: The impact of marketing at this stage is relatively insignificant, which is consistent with ENS’s marketing strategy that predominantly targets users with a foundational understanding of the product.
  2. Dominance of Auto-Factors: A substantial proportion of the influence at this stage is attributed to sustainable development factors or auto-factors, likely driven by network effects.
  3. High External Influence: A considerable percentage of the influence emanates from external factors, potentially attributable to speculative interest in the ENS token.

Awareness to Consideration, Evaluation, and Intent

  1. Significant Marketing Impact: Marketing exerts a considerable influence at this stage, corroborated by the efficacy of ENS’s marketing campaigns, such as targeted search engine advertising.
  2. Sustained Influence of Auto-Factors: The proportion of sustainable development factors remains elevated, likely propelled by word-of-mouth and network effects.
  3. Diminished External Influence: The influence of external factors is at a nadir, suggesting that well-informed users are less susceptible to speculative forces.

Consideration, Evaluation, and Intent to Purchase

  1. Reduced Marketing Influence: The impact of marketing wanes at this stage, possibly due to factors such as suboptimal website design and SEO, which impede marketability. This suggests that ENS’s marketing strategies may not sufficiently engage users at the evaluation stage to catalyze conversions.
  2. Persistent Auto-Factor Influence: The influence of sustainable development factors remains robust, likely underpinned by brand loyalty, word-of-mouth, and network effects.
  3. Resurgence of External Factors: The influence of external factors rebounds, indicating that users, when evaluating the product, may be more inclined towards speculative rather than practical considerations, viewing the names as a revenue-generating asset.

From this we can draw practically significant conclusions that will help us identify the problems of the approach to marketing. This is exactly what the third chapter of the study is devoted to.

Final Part: The Impact of Marketing on ENS Success

The subject matter under scrutiny in this chapter is considerably more intricate and expansive than that of its antecedent chapters. The primary objective herein is to undertake a comprehensive analysis of an organization’s overarching marketing strategy or discrete campaigns. The realization of this aim necessitates the utilization of an array of internal metrics and specialized analytical instruments. Therefore, the study of the object in this episode of research is only a demonstration that the problems do not lie on the surface.

To substantiate this point, I conducted a cursory analysis of the Website-to-Purchase conversion rate, with a specific emphasis on the dimension of user experience. The global user interface underwent modifications on two distinct occasions, thereby enabling the segmentation of historical data into three discrete phases:

The first stage is "the manager” era, the second is "the application” era, the third is "the v3 application” era.

From the segmented analysis, I drew the following key conclusions:

  1. In regard to the prevailing circumstances, the UX enhancement, strategically synchronized with the launch of the new NameWrapper, has elicited a markedly positive impact on conversion metrics, thereby ameliorating the extant crisis. Contrary to the transient downturn experienced during the prior comprehensive overhaul of the user interface, the current update manifests no such decline and can be deemed a good exemplar of ENS’s marketing efficacy.

  2. It is pertinent to note that the oscillations in conversion rates observed during the secondary period are not attributable to this facet of the Customer Journey, as no significant alterations were implemented during this interval. Furthermore, it is noteworthy that peaks in conversion rates are generally congruent with zeniths in sales. Specifically, the moment of greatest conversion was created by a situation in which users intended to buy a name even before visiting the site, and in most cases they made a purchase. Consequently, the nature of the average website visit has changed from intent to consideration/evaluation as conversion rates began to trend downwards.

In this regard, I make the following


3.2 Segmented Marketing ROI Analysis


The initial episode of the investigation elucidated the efficacy of marketing strategies through indirect indicators, specifically via an inverse relationship with external variables. To attain a comprehensive understanding, however, it is imperative to assess the return on marketing investments. The outcomes of this subsequent analysis will provide critical insights into whether additional financial resources should be allocated to promotional activities, or if a reevaluation of the existing advertising paradigm is warranted.

Required Resources:

Data segmented on either a weekly or monthly basis, delineated by specific expenditure categories, is essential for a nuanced analysis. For instance, for the month of January 2022: expenditures on contextual advertising amounted to $3,000 USD; investment in YouTube integration was $1,000 USD; and allocation for Twitter integration stood at $500 USD, among other line items. This granular financial breakdown will facilitate a more targeted evaluation of the efficacy of each marketing channel.


In general, the delineation of dependencies will be ascertained through the refinement of regression models and the application of machine learning techniques. However, the objective diverges from that of the current study. Our primary concern is not necessarily the performance metrics in relation to overarching market factors, but rather to quantify the impact that each facet of our marketing endeavors exerts on sales revenue.

  1. Formulation of a multifaceted model to identify distinct segments, temporal lags, underlying issues, functional dependencies, and ultimately, the computation of cost-effectiveness;
  2. Development of paired models to ascertain the significance of the multiplicative effects of each individual variable.

3.3 Qualitative Analysis of Marketing Aspects


With insights into the cost-effectiveness of our marketing efforts at distinct time intervals, we can assess the success or failure of individual campaigns. This data equips us to systematically manage our marketing budget, enabling us to identify which particular segments warrant investment and which require immediate optimization.

Required Resources:

Settings parameters for contextual advertising services;

Terms of reference for specific marketing campaigns;

The implementation of these campaigns (description of what was done), although I can do this manually, simply by knowing which posts belonged to which campaign and determining this by the time that the campaign lasted.


  1. In light of the findings from Section 3.2, pinpoint the areas where marketing expenditures failed to yield effective results through regression analysis.
  2. Conduct a comprehensive review of each campaign that was notably effective as well as those that were markedly ineffective, detailing their specific attributes.
  3. Formulate actionable recommendations outlining which strategies should be pursued and which should be avoided for future campaigns.

3.4 Target audience analysis


Demand for the ENS product, as unusual as it may seem due to the digital nature of the product, is limited. No matter how hard we try, we cannot go beyond the notional total number of cryptocurrency users at the moment. However, we can understand how the number of our consumers is currently classified. But let’s say that our average buyer is now an American 18–25 years old. This means that in some comparisons with other categories of persons, we have approached the “ceiling” with this category. This just means that we need to try to expand into other categories.

Required Resources:

Professional tools for analyzing website traffic, with segmentation by country, age, gender.


  1. Examination of the predominant categories constituting our target demographic;
  2. Ascertainment of underrepresented segments that, given the product’s intrinsic utility, could potentially exhibit interest;
  3. Implementation of experimental protocols and pilot studies geared towards the engagement of novel consumer groups.

All of the above can be supported by audience surveys, which require the consent of the community and an official email or social media account.

And a bunch of other things, from examination demand in the secondary market to analyzing reputational capital.

Post-script: I admit that I may not have fully summarized the content of the entire study, so I recommend following the link and reading the full version.
I thank @matoken.eth for providing some metrics.
I will be sincerely grateful to you for criticism and pointing out logical errors, if any.

Let’s make the ENS product more marketable together!


Wow, this is quite a lot of work. I think this will be appreciated.

This provides exceptional evaluation and examination of key data points and offers insight about the correlation of marketing effectiveness and the influential factors that are within.

Your analysis appears pointed toward traditional marketing analytics and campaign optimization rather than directly addressing the core values of Web3. Decentralization, trustlessness, censorship resistance, should be incorporated. It’s important to measure and evaluate metrics that reflect the adoption and success of decentralized tech.

Web 3 Definitive Metrics

  • number of users holding self-sovereign identities
  • growth of decentralized application
  • level of user control over personal data
  • number of ENS holders actively using their names

Community-Related Data Points

  • community engagement and governance correlations
  • level of community participation, feedback, and contributions,
  • most importantly; how decentralized decision-making influences product adoption via socail media like twitter.


  • number of users who have full control over their accounts and assets
  • number of centralized exchanges that have incorporated ENS functionality as part of their user experience
  • CEX to DEX conversion rates.
  • self-soveriegn personal treasury projects

Your analysis is valuable for traditional marketing purposes and corporate privacy invasion and data collecting strategies, but does not comprehensively address the core values of Web3 and decentralized identity. In my opinion, ENS should grow organically within. We have to remember that decentralized identity is a choice and not a product to be pampered through targeted advertising campaigns. It’s very important the those who are choosing decentralized identity by using Ethereum Name Service Domains understand why they are choosing to use this protocol.

For example, Ask people why they chose to own their cellular device. For the most part, you will receive responses that are vague or trivial and don’t necessarily align with the values of which Apple holds. ( Not a canvas response of my personal opinion, just observation). Now if you ask most individuals why they have an ENS name, they will most likely include at least one value that is aligned with web3 and ENS.

You have a great analysis and I am sure that you spent a lot of time working on this. Personally, I think that if you incorporated factors aformentioned; it could be very valuable to ENS as a whole, both the DAO and ENS Labs.

1 Like

I’m glad you appreciate my work.

I concur that Web3 innovations ought to be disseminated solely in alignment with our core value system. These values resonate deeply with me, and I envision a scenario where blurring the lines between traditional and decentralized commerce could jeopardize our unique identity. Throughout my analysis, I’ve endeavored to articulate that our primary objective in marketing isn’t merely to boost revenue, but to differentiate ourselves from speculative interest.

It is important to accept the existence of one socio-economic feature of the demand for innovative products. There is always a significant group of people who disguise their speculative interest to themselves by believing in the value of the product or in the values it carries. And here by speculative interest I mean not only that a person perceives it as an economic asset and plans to earn money from it in the future, but also as, for example, a social asset - when a person thinks that purchasing a product will give him some status and will allow himself to be classified as a member of any community.

So, for us, following a sharp increase, a decline in sales is only a reduction in profits, to which we can turn a blind eye, based on our values. And for such a person, this is a loss of emotional or monetary investment, which he is not able to compensate for with values due to their absence. Having lost faith in the value of the product, such people not only begin to perceive it as a tool of speculation, but they also transmit this position to the environment, creating a toxic image of the product, and this is the main danger.

This is true, but only if we don’t take into account:
Survivorship bias, in which people who fall into the above category are either not identifiable because they no longer associate our product with them, or do not intend to communicate with us.
Social Desirability Bias, which is expressed in the desire of consumers to conform to the widely spread image of a supporter of our values, even if they do not actually adhere to them.

In general, what I mean is that we must recognize the existence of the problem of speculative interest, otherwise at some point it will destroy our plans to build a sustainable global web3 community.

In general, in my research I emphasize that the simplest way to avoid correlation with speculation is to systematically violate the mathematical expectation of the sales level associated with this correlation. By breaking away from external market factors, we protect ourselves from tidal forces that infect our product in a toxic, speculative manner that does not correspond to our values. Then we get this ideal picture:

and which we could observe, for example, until November 2021. Until then, unfortunately, we will experience difficulties similar to the current ones.

I greatly appreciate the list of metrics you provided.
Analyzing them will help build a system of recommendations for the formation of marketing strategies that would take into account the specifics of Web3 and ENS as a whole.

But it is important to note that due to the fact that since they are both qualitative and quantitative, they have different required levels of access for parsing, as well as different objects of influence, it may take months to collect them, develop the algorithm, build a model and draw practical conclusions.

However, I am prepared to embark on this journey should there be a interest in the expertise I bring to the table.

I’m not entirely sure what this means, could you provide an example? As a decentralized service and protocol, there really is not a provable solution to hinder speculative domainers. There isn’t much ENS will implement to combat this as it would start to detract from ownership infringement.

I would recommend to replace ‘sale(s)’ with ‘registration(s)’ as the cost to acquire the right to ownership and possession of an ENS name is simply a fee that is forwarded to funding ENS development and supporting the DAO, it’s contributors as well as greater native Ethereum ecosystem initiatives.

I want to re-iterate that decentralized identity is a choice. Yes, there should be an awareness to what it offers. Quite frankly, I don’t think there is a better marketing convention that markets ENS names than putting .eth in your twitter. As far as I am aware, that was completely organic and has never been a requirement of anything or has any persons persistently directed others to do so in any official capacity.

I am in no way suggesting that it is necessary to in any way restrict people who are subject to speculative attraction from purchasing an ENS product. This is impossible in any case.

As an example, the yellow and red periods on the very first chart. Yellow is an influx of high speculative demand, since nothing more can be described, since the practical value of the product did not change at this time.
Red is disappointment in speculative value, from which we get negativity that people spread to the public.

In order to accurately understand that such an effect exists, it is necessary to conduct an analysis of sentiments, which is what I can do in the future.

Of course, the term sales is used more to diversify the narrative.

Can we say that all the people who share our values (if we really target them exclusively) read Twitter and meet people with the .eth tag there? Isn’t it necessary to unite like-minded people, and not exist in a small, somewhat closed group? Even if we proceed not from commercial, but from ideological intentions?

Let me give you an example from my life. I now live in Yerevan, like hundreds of thousands of people with a Russian passport. Many of them, like me, adhere to libertarian or even anarchist ideas, which arose among them as a countercultural phenomenon to the existing situation in Russia. Therefore, there is a possibility that they adhere to the ideas of decentralization, anonymity and confidentiality. They, like me, use cryptocurrency (mainly they operate stablecoins through centralized exchanges). But at the same time, according to my personal observations, they know nothing about ENS. They mostly don’t use Twitter, probably also because it’s a fairly centralized platform.

And there are hundreds and thousands of such communities around the world. We really don’t want to at least make our presence felt by continuing to rely on an audience from the US and these countries?

Globally, I do not agree that ENS relies solely on organic growth, because advertising is bought from the influencers on Twitter. But since such work is being carried out, why not do it systematically and efficiently?

So what do you suggest?

Most part of this research is just a form of signaling that there are problems and a way of describing the current situation. There are not very many practical insights in the first two chapters. However, they do provide a basis from which to explore the subject of ENS’s marketing success.

This study uses publicly available metrics and free tools.

If I am provided with the resources I need, described in the “proposals” part, I will be able to perform the next episode, which will be completely focused on studying the actions of the company.

answering question, I offer a list of recommendations that can be used to modernize ENS global marketing strategy.

For example, if I was given the right resources to write


I could infographically describe which ENS marketing campaigns were worth repeating and which were worth abandoning.

If I have access to resources for this chapter:

I will be able to analyze the ENS demographics that I wrote about in the last post. With the help of the recommendations in this chapter, we can diversify the community by country of origin, age and gender, for example.

You can also add sentiment analysis to this (for this I need the Twitter API V2), or the analysis of Web3-specific metrics you mentioned above (for this I would first need to at least determine what tools I might need).

Honestly; my opinion and maybe others would agree, I don’t think demographics are important at all. Demographics seem to do more harm in the grand scheme of things. They set up social constructs for discrimination and favoritism by proxy. It’s all just unhealthy to me.

If someone wants to display where they are from then that is a personal decision they make and should not influence ENS marketing strategy at all. Especially the gender demographic. The only area where gender should be measured is through medical insurance and ENS isn’t doing that. Otherwise it’s not my business or your business as to what gender ‘insertgenderhere.eth’ identifies as. Again, personal opinion but I do think that many here would agree.

The best way to gauge sentiment would require an algorithm that measures data points on chain. Anything off chain is intrusive and should be off limits. I mean you are free to do what you want to do. You mention qualitative and quantitative data and Twitter isn’t the right place or quantifiable data because there are too many discrepancies that would result in adjusted data, which is inaccurate. All the data needed is on chain to measure sentiment, that’s the beauty of the EVM ledger. I just don’t think this traditional corporate reporting approach is aligned with web3 decentralizers.

It seems to me that to some extent we do not understand each other. If I tell you that the majority of ENS users are Americans 18-25 years old who associate themselves with the male gender, could this be considered discrimination, or a reflection of real factors?

I am just calling for differentiation from such an average category and trying to expand to non-US citizens who are over 25 years old and have a gender identity other than male.

No, it doesn’t constitute discrimination. However, we should question the purpose behind collecting such data. Why does it matter whether someone identifies as male or female in a system where this information serves no practical function? What’s the significance of knowing a user’s age and gender in a decentralized community striving for independence from institutions that exploit such data for financial gain?

This is where concerns about invading privacy emerge.

Let’s consider an example with the following address: 0xaDe039…

0xaDe039… logs into a decentralized application (dApp) that requests their gender. 0xaDe039… responds with “male.” The dApp employs analytics tools and captures the IP address from the submitted form. Now, 0xaDe039… and IP address are linked together. IP address is tied to a server known as west-01. It’s worth noting that west-01 exclusively serves a specific geographic region. As a result, we now have knowledge that 0xaDe039… is a male user connected to IP address in the region labeled as XYZ. Then we discover a Twitter profile that references the address 0xaDe039… User 0xaDe039… has not disabled their geolocation privacy settings on Twitter, leading to the logging of all their locations.

This example illustrates why I believe probing for such information is unnecessary. Additionally, this type of data can often deviate from reality, leading to inaccuracies.

If this information is not stored on the blockchain and fails to provide consistent and accurate data, it becomes futile and can potentially be exploited for malicious targeting. I have no desire to be a target. I’d rather not be considered a ‘targeted’ consumer of any kind, to be completely honest. Why should someone else profit from my identity attributes in a network that upholds values against the involuntary collection of demographic and protected category information?

In my experience, I’ve never received compensation from anyone who used my data without my explicit and transparent consent.

Once again, this represents my personal viewpoint. I encourage others to share their thoughts, whether in agreement or disagreement.

perfect example. well… almost perfect.
Screenshot from 2023-09-13 23-46-18

1 Like

You gave a very good example and I agree with it. But I suggest a slightly different way.

  1. Collect data not from the blockchain, but from website traffic. The ethnic composition and other demographic data will always be proportionately equal to what we have among those who register a name.
  2. Receive them in anonymized form, in the form of a percentage, so as not to violate the privacy rights of individuals.
  3. Use them not for targeted actions, such as targeted contextual advertising, but to diversify the platforms through which we distribute and the methods by which we conduct social networks.

A little about the so often mentioned contextual advertising. Setting it up for personal data such as region, gender, age, financial situation, and so on is wrong, I agree.

But I think they will agree with me that if a person Googles about web3 domains, he should learn about our existence. For some search keywords, for example ‘web3 domains’, the ENS website will be located in 25-35 place in the search query, depending on the region selected in Google.


Yes, if you enter, for example, ‘eth domains’, then the ENS site will be in 1-3 place. But aren’t we promoting that ENS is not just .eth?

If you do not take into account the delivery of advertising, for example using DuckDuckGo, the situation will not change much. This also means that we generally do not work with search algorithms, unlike, for example, our competitors.

yeah i would love to see results in this order :


i’m not sure how that is controlled perhaps someone else may know.

1 Like

So, @Ecosystem_Stewards, could you guide me on further actions?

Is the above what you are looking for? If so, this is not something the ecosystem is set up to provide.

Is there a request that I may have missed that you’d like guidance on?

I’m sorry, but I still couldn’t figure out who at ENS is involved in marketing in the broad sense. Should I contact Meta-Gov, Public goods or ENS Labs?

Here is a list of what I need:

Description and budget spent on specific marketing campaigns. This is just a metric; no third-party resources need to be used.

Web analytics tools. This is either free Google Analytics, in order to use it I need you to give me access to analytics reports. Or paid tools, for example SimilarWeb ($400/mo.).

SEO analytics tools. Again, it could be free Google Ads and Google Search Console that I need to have access to.

Twitter analytics tools for sentiment analysis. There are no good alternatives to Twitter API v2. I need either a monthly subscription ($100/mo.) or a Bearer token, which the SMM team quite possibly has.

To sum it up, I either need funding or access.

ENS marketing is handled by ENS Labs; the director of marketing is @sadaf.eth.

Some organizations engage in marketing that helps their business like ens.vision. They do not have financial support from the Ecosystem WG.

The Ecosystem WG focuses on supporting projects like eth.limo, 1w3.io, and namesys.xyz.

Aight, thanks.
Hey, @sadaf.eth ! Can we discuss providing access to Google analytics reports (GSC, .Analytics, .Ads)? Could you provide information on advertising expense reports and campaigns conducted? Does your team have an OAuth v2 API on Twitter and can you share the Bearer Token?

At the very least, that would tell us that there are a lot of potential ENS users who aren’t being reached, and we should do something about that. For that reason alone it seems worth measuring.


Maybe, I judt don’t see gender needing to be measured of the goal is to be a universal domain naming system for devices that aren’t gendered. ( for now at least :stuck_out_tongue: )

Whoever figures out how to build marketing campaigns to reach people who will be interested in ENS before they know they are interested in ENS will be knighted.

Perhaps a carefully trained; self-learning dynamic LLM data pipeline analytic initiative.

1 Like

That is the main goal!

1 Like