Think!

The guy on the street is talking about incubators and accelerators.

You bump into Dave McClure at your favorite coffee shop.

The car sales man sold software for 20 years.

People are talking about open compute in the public bathrooms.

You see the same faces at Starbucks at the same spots everyday, using it like an office space.

The Google Product manager you spoke to a couple of days ago lives right next door.

You meet more Founders than Employees. 

Your car breaks down with a software bug.

You drive into an “Infinite Loop”. 

Dear Partner Manager at Google / DoubleClick - We’ve been trying to get to you for the last one month. This is to get our platform Reduce Data Certified. In our first response I got a form within an email without a submit button. 

In my second email a week later, I was asked to type down the responses line by line which I did and have not got any response.

I have tried to reach out to folks whom I know at Google but looks like it will take some time before I can get that channel to work for me.

In the contrary, last year before the Holidays, I got in touch with Facebook and immediately got a response from many senior folks in FB Partner Management. They were eager to sign on new advertising partners and even got into calls the following week and replied  even through the weekend with real names at the end of the emails and contact info in case I wanted to get in touch with them.

This really brings to me one key question: Is Google too big to respond to partners, customers and the marketplace in general. I wish it were not true. But it seems to me like it has already become too big to be nimble.

In any case, a request: I am still waiting for response from an unnamed person in the Google Doubleclick certification team. We are  losing valuable business because my company, Reduce Data is not yet certified.

Almost everyday, I bump into Entrepreneurs from India in various parts of Silicon Valley. Mountain View, Sunnyvale, Palo Alto, San Francisco, Menlo Park …they’re everywhere.

Surprisingly, many of them are ‘fresh off the boat’. That includes me as well, who came in late July 2012 and got a long term work visa a couple of months later.

In my discussions with Mukund Mohan (EIR, Microsoft Accelerator), it seems that a little over 300 companies have moved just in the last couple of months alone.

Many are gaining good getting traction, getting funded and hiring a lot of employees (both in India and in the US).

There are a couple of things that is driving this

a) Cities like Bangalore have become very expensive.

b) It is probably easier to hire and retain in Silicon Valley than in Bangalore (and sometimes in Chennai).

c) Most software product companies are built for global markets; and there is no reason to build it out of Bangalore or Chennai when you can do it in Silicon Valley

d) Silicon Valley is probably the best place to be in the world to build a technology company (You have to be here to believe it); and the lure of Silicon Valley culture is very strong - It is probably the only place in the world that has a high concentration of technologists, entrepreneurs and venture capitalists. All of it being in one place helps! 

e) VP Level Jobs are at $100k in Bengeluru and Mumbai. 

f) The infrastructure issues that you face (traffic, commute times, power cuts etc) are a serious impediment to building any kind of business. 

Having said that these companies including mine aren’t really shutting down India offices and moving everything to the valley. We  are maintaining an engineering and sometimes additionally an operations team back in India.

But the centre of gravity of these companies is definitely moving west to Silicon Valley and while this is really good for the companies, it is definitely not good news for India in general. 

I logged in to Godaddy.com to change a domain and guess what?. All the domains were missing.

I call customer service, they say I deleted the domains but wouldn’t confirm the IP from which it was deleted

“Godaddy is a large business and we track millions of customers across the world, so we don’t possibly store the IP or region or computer name from which the domain names were deleted “

Another one says,”I didn’t say we don’t save this data, it is just that I don’t have access to it”.

Okay - I keep saying, I didn’t delete it. I deleted the debit cards that were on file so its possible that Godaddy possibly deleted all my domains along with my debit cards but no - the agent on phone doesn’t accept that’s what actually happened. 

“We will restore all of your domains except one domain which was about to expire - we have already released that and it costs $80 to restore it”. 

Me - “So you’re holding my property Illegally and asking me to pay money to release it”

Agent - “Sir that’s the charge the registrar charges …we can come halfway and discount $40”

Me - “Why did you release a domain that belonged to me..the registration was still active. And two days before the domain expired, I renewed the .co domain at $30 for a year”.

The agent, “Since the domain was close to expiry so we released it”.

Me - “So you released my private property back to the registrar without reconfirming even though to me at that at that moment”.

Agent - “Yes - You cancelled it”

Me - “But the other domains are still intact”

Agent - “Yes but the last domain was nearing expiry, so we released it”

Me- “But I just renewed that like 2 days ago, which means it should not have been released”

Agent - “Doesn’t matter. You cancelled it again”.

After nearly 30 mins of argument, the agent doesn’t relent. He offers to give a discount but will effectively charge me $30 + $11 for the domain including registration. 

I remember that I requested them to “unlock” that domain so that I could transfer it to Namecheap.com. I never got to transferring it but Godaddy either has a serious issue with their system or have found a way to make a quick buck out of customers abandoning ship. 

To think of it, they might have done more harm by shutting down my key domains by simply releasing (my private property) back to the registrar. 

I cannot believe that this day and age, someone could be so cheap with their own long term customers. I’ve learned a valuable business today and that is to never trust a company like Godaddy ever again. 

How many times has this happened to you - You search and buy a product. Then you’re constantly hounded again and again for the next several days with ads of the same product appearing all over. Chances are that it has happened more than once. 

Would you buy the same vacuum cleaner twice? Would you want to buy that same book from Amazon or those amazing pair of shoes from Zappos again? I don’t think so, but today many of these ads tend to appear again even after a purchase. 

Re-targeting is the method of displaying ads again to users who have seen something of interest but did not complete the transaction. 

Re-targeting is useful, but displaying the same ad even after I (or any user) have already bought the product is a waste of advertiser spends and a great way to annoy any user. 

Why does this happen? Many re-targeting solutions don’t really exchange necessary data that tells the Ad Networks if the user has purchased the item or not.

And without this, what happens is that you get a constant barrage of ads of the same item displayed on every other site. 

This has been happening for a while now [http://www.adexchanger.com/data-driven-thinking/personalized-retargeting-overkill/] and therefore not really a new issue. 

To give you an example, I did an actual transaction yesterday evening (1/22/2013) by buying a Vonage line. 

• I saw the service yesterday morning, clicked through but dropped out at the payment step.
• In the evening, I saw Vonage targeting me with Google ads. I clicked on a re-targeted ad which had an offer on it.
• Next, I bought the product in a few minutes and received confirmation email [see slide 1 at the bottom of this article]
• Even 10 minutes after the purchase, I saw the ad again on the same website.  [see slide 2]

I really don’t want two phone connections today. Please forgive me for even buying the first one :-).

This is a classic case of an over use of re-targeting and probably just an inefficient tool  wasting advertiser spends. What’s troubling is that this has happened on a prominent ad network [Google] and not some small random player.   

So how does one fix this?

The solution to this problem is a little difficult but doable. Advertisers can exchange data with ad networks (using cross browser cookies or other mechanisms) to avoid such re-targeting issues. 

A simpler way would be to use a tool / network that has solutions to such problems. 

Another option (which also happens to be a shameless plug ;-))

It is always also recommended that an advertiser use an independent audit system such as Reduce Data to verify ad spends. 

Beyond that, Reduce Data can help identify steps in the campaign funnel causing media waste and give specific recommendations that can help optimize ad spends.

If you would like more information about Reduce Data, please head to our website at http://www.reducedata.com or our blog at http://blog.reducedata.com. 

The majority of Internet services are supported by advertising. I think it is is OK if a few customers decide to take steps to block ads on their browsers. But when an ISP does that, it creates a ridiculous situation (Fast Company:  http://www.fastcompany.com/3004452/french-isp-free-blocks-all-web-advertising).  

I think Google and other large players should limit access to free services (Gmail, Search etc) by putting up a SOPA like blackout for at least for 1 day. This blockade should not be seen as a use of force or coercion but rather a gentle reminder to users saying that ad revenue is important to Internet based free services.  And that they should request their ISP to not to unilaterally block advertising. I also suggest that the services be limited only briefly and that the user can go past and continue to use the product after seeing the message. 

I believe that an action like this will raise awareness of ad supported services and hopefully French consumers will force French ISP to reconsider its decision. 

This blog entry is written in response to a blog by Suhail Doshi of Mixpanel titled Bullshit Metrics (http://sufficientlyadvanced.net/bullshit-metrics)

Suhail Doshi of Mixpanel calls user signups, page views and other similar metrics as Bullshit metrics saying that these metrics don’t really correlate to the success of the startups.  

Maybe Suhail forgot that only in May, Mixpanel itself had touted that it measured 7 billion actions (http://gigaom.com/2012/05/10/mixpanel-raises-10m-in-bid-to-dominate-data-geekery/). What does that mean anyway? Isn’t that bullshit metric as well?

Metrics such as user signups and page views are important. Getting enough users to sign up does matter. Retention, active users and other metrics that can only be measured after the signups have occurred in the first place.  

Page views cannot be wished away when most Internet businesses are still dependent on advertising as a source of revenue:

The blog talks about Tumblr’s 20B impressions each month as one example of a bullshit metric. Now if we all agree that revenue is an important mechanism to identify a key metric then I guess we all agree that advertising revenue which is driven by ad impressions is a key metric. Ad impressions are  directly proportional to page views. This means that page views is a metric that cannot be wished away.  While for many, this may not be the single most important metric but it is an important metric nevertheless.

New metrics of measuring engagement especially in media such as engagement in the form of likes, or re-tweets are also important but again, similarly they aren’t necessarily the only metric that can be correlated to success.  

Many Internet businesses depend on advertising revenue (including social media giants like Facebook and Twitter) and what matters to them are Page views which leads to ad impressions, clicks, conversions, cost per conversion, brand lift, etc.

The survey results below clearly highlight the industry standard: 

Metrics used by Brand Marketers in North America to determine effectiveness of online ads:

Active users are an important metric for Facebook. But Page Views are important in order to justify ad impressions: 
Page views are also measured even through photo flips because  this helps Facebook justify the various number of impressions of ads it serves up to each user.  As the pages are flipped, the ads change because FB follows the standard practice that was set by various publishers for a long time – display different ads on different pages and earn more.

Facebook’s charges are impressions based or sometimes click based. (Cost Per Milli - CPM - which represents 1000 ad impressions or on Cost Per Click - CPC basis).   Page views are  inherently, a very important metric.

Facebook ad screen showing how the ads are sold:

What is changing is that advertisers have been saying that things like Click Through Rate (CTR) or eCPM don’t matter but none of that has been leading to any kind of One Key Metric. It all is leading to several different metrics depending upon the mind of campaign run but again they are all linked to Page Views.

Engagement is an important metric like for example videos played. I do not disagree. But may not the only key metric. The survey below clearly shows that no single metric is important to the ad agencies / advertisers who are the primary source of revenue to someone like (Google) Youtube.

Conclusion:

I don’t think everyone is going to start dropping existing metrics and rush to find that single key metric that they need to focus on. There are a lot of people who do not agree with this and are already labeling it a fad. I don’t have an opinion on One Key Metric.

But calling key metrics that actually are directly related to the business’s success as “bullshit metrics” is being stupid or just being ignorant. 

There are thousands of blogs and books on performance optimization. Yet, I thought it might be appropriate to put in some of my learning into a blog. So here it goes, a starter guide to performance tuning of web apps.

Simple Page Specific Optimizations
 

There are a lot of simple optimizations that can be done at the page level. The simple optimizations are

• Load fewer resources (such as images, css, js) or put all of them into a single js or cs file.
• Optimize image sizes, if possible use sprites.
• Minify JS..
• Load only the JS or CSS necessary
• Use tools like YSLOW to test page performance and optimize
• Cache all static resources (more on this later).

Caching And Serving Files using the File System:

The easiest thing to do in situations where the content is the same for all users is to cache the entire page and deliver it via the file system. 

Caching And Serving Files using the Nginx and Memcached:

You can also use the above pattern to store data into memcached and force nginx to directly connect to memcached and serve the files. The result will be much faster than fetching files from a file system. Remember, the ‘HTML Generator’ is a component that you will need to write. This component will fetch results and store it into Memcached with the file name as a key. We can then modify the  nginx configuration to look into memcached first if a request comes in. If the file is not available into memcached, then the request could be passed to the file system or the underlying application server. 

Caching Data into Memory

Assuming that there are scenarios where the entire page cannot be cached for all users. In such scenarios, it is better to apply caching to cache a specific data set, a specific part of the page, a recomputed result and so forth. This cache could be specific for each user or a common piece that could be used for all users.

A cache is typically an in-memory system such as memcached.

Example: Loading some dataset into memory using nodeJS code:
 

var client = new memcache.Client(11211, ‘127.0.0.1’);

client.connect();

connection.query(“SELECT business_name,id from business”, function (error, rows, fields) {

  client.set(rows[i].id,rows[i].business_name ) function(error, result){

}

——

A partial segment could be a table that is repeatedly shown to the users. This could be generated and stored into memcached for use anytime later..

Example: Creating a list of links and storing it into memcached using Javascript / NodeJS:

var news;

for (i=0;i<=1;i++)

{

news=news+items[i]+”</br>”;

}

memcachedvariable.set(“news_links”,news);  //set news links into memcached.