How Aadhaar can usher the biggest political reform in India !!

Aadhaar is arguably one of the best government schemes that have been implemented in India in recent times. It is poised to bring revolution in a lot of sectors in India. One such area, which has unfortunately not been discussed at all, is reforming the Indian electoral process. In this blog, I will try to highlight how easily Aadhaar can be used to usher the biggest political reform in India. Hope someone important in Election Commission reads this post.

One big difference between Indian constitution and any other constitution in the world is that the Indian constitution mentions in great detail about the election process and the powers of Election Commission (EC). Rather than leaving the election implementation details to laws made by the government of the day, the founding fathers decided to explicitly keep them in the constitution. This was one of the major reasons why India remained a democracy even though almost all other countries, that gained independence after World War-2, quickly succumbed to dictatorship or civil wars. India did not have to depend on political incumbents to implement electoral reforms, which adversely impact political incumbents the most. Key electoral reforms in the past, such as photo-based voter id cards, electronic voting machines etc., were spearheaded by EC, despite strong opposition by political parties. Any realistic plan for electoral reforms in India should be implementable by EC, with minimal or no support from the government.

There is no doubt that India is a thriving democracy and we have relatively free-and-fair elections, as compared to other democracies in the world. Due to the efforts of EC over multiple decades, the era of ballot stuffing, booth captures and fully rigged elections (such as 1987 Kashmir elections) are now over. However, there is still much to be desired in our current election process. Some of the major problems include - high cost of conducting elections, missing names in voter lists, multiple entries of the same person in different (or same) booths, fake names in voter lists, names of deceased people in voter lists, difficulty in finding the booth in which a person has to go for voting, enabling voting for people who cannot go to the booth for some genuine reason (E.g. soldiers, migrant workers etc.), low voter turnout especially among well-educated class etc.

Before I start on why integrating Aadhaar with voter id cards & electoral rolls is such a great idea, let me quickly go through the basic features of Aadhaar. Aadhaar is the only database which guarantees almost zero duplication and has 99% coverage of all Indian adults (1.08 crore people). Unlike a bank card or a chip-based driving license card or PAN card, the Aadhaar card can be easily printed at any Aadhaar center. There is no logistics required to transfer the card from a central printing location into the hands of the citizen. It allows instant identification of any Indian, through biometric identifiers such as fingerprints or iris scan. All other fields in Aadhaar database (such as name, address, mobile number, email id, photo etc) can be incorrect, but the biometric data to Aadhaar number mapping is 100% accurate. Moreover, the Aadhaar based India Stack APIs have proven their reliability, scalability, performance and efficiency for both authentication and fund transfer purposes.

In 1993, EC introduced photo-based voter id cards called EPIC (Electors Photo Identity Cards) and it took more than 11 long years for it to become compulsory for electoral voting (there are still some very limited exceptions allowed). EPIC was a major step in reducing electoral frauds. It is now time to link EPIC cards with citizen's biometric information to make it even more secure and error proof. The good news is that because of existing Aadhaar's biometric database, it would take less than 2-3 years and only 1 general election to completely shift to bio-metric based voter id card and electoral rolls. I will call the new voter id card as Biometric Electors Photo Identity Card or BEPIC going forward.

Challenges in linking Aadhaar with EPIC

1. Aadhaar is legally not compulsory for Indians, except for getting certain monetary benefits from the government. No one can be denied the new biometric voter id cards because
• He/she does not have Aadhaar card or 
• He/she does not want to link his Aadhaar with EPIC. 
2. Aadhaar is not an official proof of being an Indian citizen and so a citizen cannot demand BEPIC based on just Aadhaar (it requires some address proof). Many illegal immigrants, especially from Bangladesh, have an Aadhaar card but no EPIC or presence in the electoral rolls. 
3. The minimum age for voting is 18 years. However, Aadhaar card has no age restriction and there is no sanity of the age (date of birth) in the Aadhaar database.
4. Secret Ballot is fundamental to democracy and no IT system can be allowed to store the vote(s) cast by any individual. The current EVMs only store the total votes cast for each candidate and not individual votes. 
5. The time required to cast a single vote in a polling booth cannot be increased, else it would lead to complete chaos of elections which need to be conducted in a single day only.

Proposed Solution

Given the privacy concerns that are increasingly going to be associated with Aadhaar in future, it is important that new bio-metric based voter-id database of EC is not an add-on to the Aadhaar database (or any direct integration), but rather a completely separate database. This database must maintain its own copy of the biometric information for each Indian citizen, along with Aadhaar number (optional) and all other fields that the current voter-id database maintains.

If the Aadhaar number is voluntarily provided by the citizen, then the biometric information saved on both the databases must be the same (i.e. or easily matched by a standard algorithm). Currently, Aadhaar does not provide any mechanism for transfer of biometric information to any other database and this need to be custom built specially for the EC. This is where the biggest legal and implementation challenge lies, as this cannot be done by EC alone. This requires government approval and active participation.

If the Aadhaar number is not provided by the citizen, then Election commission need to take the bio-metric details of the resident themselves and classify citizens into the following two types -

1. Indian having Aadhaar but new biometric voter id card is not linked with Aadhaar: 
   Biometric information of the citizen matches exactly with a particular resident in Aadhaar database. The same biometric data must be stored in both databases which can be matched by a standard algorithm.
2. Indian not having Aadhaar but having new biometric voter id cards: 
   Biometric information of the citizen does not match any of the existing residents in the Aadhaar database. The biometric information must be saved in such a format that in future if his/her Aadhaar number is issued, the biometric information between the two databases can be matched.

Similar to the Aadhaar card, the EC centers should be able to easily print and give out BEPIC in real time without any logistics required to track the transfer of card from a central printing location into the hands of every citizen. Similar to Aadhaar, a citizen can have any many copies of his/her BEPIC card as he/she wants.

The link between Aadhaar database and BEPIC database can also be severed in future after enough elector coverage has been reached, if required for privacy or control concerns. There are pros and cons for both cases - having a permanent link between Aadhaar and BEPIC or having a temporary link in the initial phase only. I will not go into that much details in this blog.

EC will also have to replicate the relevant portion of the Aadhaar system which enables instant citizen verification anytime using his/her biometric information (fingerprint or iris scan). Of course, since this system will only be used by EC and no other third party/organization, it can be at a much lower scale/capacity than Aadhaar system. This instant biometric verification system needs to be enabled at each polling booth. This will reduce the manual dependency to verify address and streamline multiple EC processes.

A single system of electoral rolls based on BEPIC needs to be developed for all types of elections - central, state, municipal or panchayat elections. I will call the new electoral rolls as BEPIC electoral rolls subsequently. The main aim of BEPIC electoral rolls will be to ensure that the same BEPIC does not have different addresses in different electoral rolls (including municipal and panchayat electoral rolls). Change of address, at any EC center, by a citizen before any type of election, must be reflected in all electoral rolls for all subsequent elections as well.

Operations & Change Management

• Online linking of Aadhaar number to EPIC number
  Similar to the various forms that exist on National Voters Services Portal such as Form 6, 7, 8, 8A etc, a new form should be created which allows residents to voluntarily link their Aadhaar number with their EPIC number after uploading relevant documents and manual back-end verification. Once this is done, the citizens must be asked to collect their new BEPIC from designated centers. The website http://electoralsearch.in/ used to have a link for linking Aadhaar number with EPIC number but that was discontinued later due to unknown reasons.
• Issue BEPIC at Polling booth stations before elections
  Every year the election commission asks citizens to come to polling booth centers to verify that their names are in electoral rolls and to invite claims and objections for inclusion, deletion, modification & transposition. The Election Commission should enable these EC centers to necessarily have biometric devices similar to Aadhaar centers. All the above processes need to be changed to incorporate the instant biometric verification functionality that is now available with EC.
  Citizens can link their Aadhaar number with EPIC using standard e-sign functionality and instantly get their BEPIC card (without the need for a second visit). If citizens don't want to link their Aadhaar or do not have Aadhaar, then they can give their biometric data at these centers. BEPIC cards can be collected from these centers after 15-20 days or can be simply printed online
• Preparing new BEPIC electoral rolls before elections
  Currently, electoral rolls for each polling booth is divided into various Part numbers with a serial number assigned to each EPIC (each voter). Electoral rolls are refined before each election (state or central election) to remove wrong entries, update address changes of residents, remove deceased residents etc.  Each entry in electoral roll must be classified into the following categories
• EPIC only
• Aadhaar link request pending
• Biometric taken - BEPIC request pending
• BEPIC approved 
• BEPIC issued
• Giving BEPIC card during elections
  Each polling booth should pre-print BEPIC for all citizens for whom BEPIC has been approved but cards were not issued. On election day they must be given their BEPIC before casting their votes. Even voters who were issued BEPIC card but did not bring their BEPIC card must be given a new copy before casting their vote.
• Change in the voting process during elections
  BEPIC cannot be made compulsory in the first election after its launch. However, biometric details (or Aadhaar e-sign) can be taken for all non-BEPIC voters, after they have cast their votes, so that in the next election BEPIC can be made compulsory. The important thing is that taking biometric details for creating BEPIC can be outsourced to agencies, similar to how Aadhaar details are currently being taken. All voters are asked to carry some ID proof when they come to cast their votes. If Aadhaar card as an ID proof is presented as the most preferable option by EC messaging, the queues in polling booth can be made much shorter.

Benefits of linking Aadhaar with EPIC

1. Aadhaar has already done the heavy lifting of checking each person's biometric identifier with everyone else to determine that it is the unique one and reject duplicates. The costly and time-consuming step will not have to be done by EC for all those cases where citizens themselves volunteer to link their Aadhaar number.
2. BEPIC will lead to near zero election fraud (missing/duplicate/wrong entries in electoral rolls) in Indian elections. By leveraging Aadhaar, Indian elections can have lower election frauds than any developed country in the world.
3. Providing a valid address proof is very difficult for certain sections of the society such as homeless people, migrant workers, flood/earthquake victims etc. Biometric based instant verification devices at each EC center, connected to a central database, can be easily used to remove the need for any other address proof.
4. Currently transferring the address mapped on your voter-id (and electoral rolls) from one state/city to another takes a very long time as it requires paperwork at both ends. This leads to effective disenfranchisement of a large section of the society. The proposed solution will enable single visit, at-the-counter instant transfer of address in BEPIC and change in the designated polling booth for any citizen.
5. Suppose a person goes to the wrong polling booth during an election, he can be easily redirected to the correct polling booth. It is technically possible to even create solutions where a person can go to any polling booth to cast his/her vote. Similarly, it is also possible to create a system for early voting (similar to USA) or online voting or removing the need for indelible ink. However, these are not advisable in Indian context at present.
6. Due to a focused effort by EC, the electoral rolls for central and state elections have few errors. However, electoral rolls for local municipal & panchayat elections are extremely flawed. Local elections in most places in India cannot be called free and fair. The BEPIC electoral rolls system will clean up the electoral rolls of municipal/panchayat elections as well. 
7. Conducting elections is very expensive for a developing country with a huge population such as India. A very large percentage of the cost is because of manual efforts to maintain multiple electoral rolls, continuously conduct data correction exercises before each election etc. BEPIC will result in a one-time high capital expenditure but will reduce the operating expenditure substantially.
8. Mature democracies around the world are moving towards participatory democracy from representative democracy. Going forward there will be far more referendums and other forms of direct public participation in shaping policy. BEPIC will make India future ready for participatory democracy.

I have long believed that the solution to most of the Indian problems is technology. Technology will help us leapfrog several challenges that other developed countries took decades to solve. This step alone will not usher India into an era of clean politics but will go a long way in institutionalizing free and fair elections and deepening the roots of democracy in this country. Given the pace at which new applications are being built on top of Aadhaar, I am very confident that some variation of what I have described above, will be implemented by EC in the next 5-7 years. It will be very interesting to find out how accurately I was able to predict the future !!

Make right trade-offs in designing your mobile app

Smartphones & mobile internet are one of the defining technologies of our age. The promise of very high stock market valuation, industry disruption potential, unlimited customer reach, easy incremental improvement, low capex cost, high customer engagement and very low entry barrier has led almost every technology company and even non-tech-savvy companies to come up with their own mobile app(s). However, most companies are now facing the following harsh realities of the extremely competitive app business -

1. Usage/Reach: An average mobile user just has 10-30 mobile apps on his phone at any given point of time. Hence out of the millions of mobile apps, very few of them have more than a few thousand installs.
2. Customer Acquisition Cost: The customer acquisition cost is extremely high for a new mobile app, ranging from Rs 500/- to Rs. 1,500/- (This varies widely based on the industry). It is generally far cheaper to acquire a new customer through traditional means rather than through a mobile app for any business.
3. Retention: Customer loyalty/stickiness is difficult for 95% of the companies who are not aggregators and whose product/services do not require repeated/periodic customer engagements.
4. Customer Lifetime Value: Trying to monetize a mobile app has a direct negative impact on the overall reach & customer retention objectives. Even meeting non-monetary business objectives (indirect benefits) from a mobile app is tricky. For example, if a mobile app gives too many notifications, the customer will soon uninstall the app. Hence the overall Customer Lifetime Value (both direct & indirect benefits) frequently falls below the initial estimation.
5. Operational cost: While the capex required to create a new mobile app is low, the opex to maintain the mobile app is significant. This is especially important for non-tech-savvy companies. 

In spite of these major challenges, it is very clear to most companies that they will have to keep investing in mobile apps, till the time they get it right. If they don't do this, then they are at a big risk of their business model being disrupted by someone else.

I believe that the key to success is to understand that there are multiple categories of mobile apps and different design principles need to be followed for every mobile app. At each step of creating the mobile app some or the other trade-off has to be made. The key to success is to make the right trade-offs based on the category of the mobile app and its usage pattern. In this post, I will try to define few broad rules/heuristics for mobile app design for different categories of mobile apps.

There are multiple ways (customer need, industry segment etc.) of classifying mobile apps into different categories, but for this post, I want to classify them, based on different design requirements, into the following three broad categories -

1) Mass-market mobile apps

Mass-market mobile apps either do not require user registration or new registrations/onboarding can be easily done online with no offline component. These mobile apps are open to everyone and are not restricted to a segment of the population. They can be further classified into the following four types -

1. Aggregators & Market-Makers (Eg. Uber, Amazon, Bookmyshow, Oyo rooms etc) 
2. Networking (Eg. Facebook, WhatApp, LinkedIn, Tinder, Jeevansathi etc.)
3. Personal Utility Applications (Eg. Gmail, Maps, m-Indicator, Clean Master etc.)
4. News, Entertainment & Games (Eg. Netflix, Youtube, Hotstar, Pokemon etc.)

The mass-market category of mobile apps are most well known and they are generally cited as examples to create the do's and don'ts list for mobile app designing. Each of these four sub-types within the mass-market category, require slightly different design approach. 

Instead of exhaustively listing down all the common well-known design heuristics (such as late/delayed login, frequent controls on bottom and less frequent controls on top etc.), I will only be listing the ones which I believe are less known -

• If you can choose between using either email id or mobile number as the basis for new customer registrations, always prefer mobile number over email-id. I am still surprised by the number of product managers that are not aware of this even after Whatsapp's remarkable success story.
• Integrate voice commands wherever possible and as soon as you can. Voice commands are going to be the next big driver of almost all mobile apps in a multi-lingual world and it would take considerable time & efforts to gain expertise in this. Hence starting early will be a big advantage for most companies in long run.
• Customers worry about battery backup and data usage the most and this is unlikely to change for a long time. These can be one of the biggest hidden advantages of a mobile app, especially for mass-market apps (E.g. UG browser)

2) Target customer segment mobile apps

This category of mobile apps are meant exclusively for a target customer segment such as - (a) Customers using company's high-involvement product/service, (b) Residents of a city/state, (c) Subscribers of a repeat service/product (d) B2B services etc. Some good examples are Maruti Care, Kotak Bank, Bharatgas, e-Aadhaar etc.

The main difference between this category and the mass-market category is that these mobile apps are not meant for new online customer acquisition. At best, these mobile apps can be used for lead generation, but the actual customer acquisition process is much longer. The main purpose of these apps is to improve customer service, get repeat orders, get new customers referrals, increase customer engagement/loyalty etc.

Most non-tech company apps belong to this category, but there are very few success stories in this category. Even large product companies such as GE, Philips etc are struggling with their mobile apps.

Some common design rules for these types of mobile apps are -

• Be extra careful to ensure that there is no way in which duplicate accounts or cross accounts can be created. One of key to success for this category of mobile apps is not to give access outside the target customer group and to have zero duplicate or cross accounts even if the customers change their email id or mobile number or address or even their names.
• The app must be designed for quick installation and un-installation as most customers will install the app for doing particular tasks and then will uninstall the app. The app must be as light as possible on internal memory usage, especially when the app is not in use even if it means trading off some non-core features.
• Give customers a detailed option to customize the notification that they receive from the mobile app. Each unwanted notification has the risk of customer uninstalling the mobile app especially since the stickiness of these apps are very low.
• Companies must be open to integrate with other partners, suppliers, 3rd party agents etc. if required to meet all customer needs through a single app only (E.g. Maruti Care)
• Creating a smart chatbot (chatterbot) to be integrated with top chat applications such as Facebook Messenger, Google Chat, Yahoo! Messenger etc. is a much wiser option for a lot of companies, rather than creating separate mobile app.
• Wherever possible, prioritize giving data to top aggregators proactively, rather than creating a separate mobile app for the company.

3) Captive customer or Business Support mobile apps

These are apps meant for employees & business partners for helping them with their day-to-day job or business functions. There is no marketing budget required to make these apps a success. For example - Uber Driver, Amazon Seller etc.

• The key to success for these apps is not to be light and glossy but to be as productive and comprehensive as possible.The trade-offs for this category of mobile apps is completely different from the previous two categories.
• The initial registration process for these can be as complicated as possible and the mobile app can be heavy on mobile resources. Process workflow optimization must be prioritized over UX optimizations.
• Many organizations choose not to put their internal apps on public app stores. This is a very big mistake as without having them on the app store, there is no efficient way of ensuring that every customer has the latest updated version of the app. Making the app available on public app store does not mean that you have to spend time & money on optimize display on all screen sizes or create a way for online registration etc., which companies may not want to do for internal apps.
• These mobile apps are generally used as means to implement process changes in the organization and hence their launch must be accompanied by careful change management and training.

In this post, I have deliberately tried to avoid discussing the latest technical trends, frameworks & architecture for mobile app creation even though it is a very important decision to take. The focus has been to enable Product Managers to take the right trade-off decisions, regardless of the technical details. Many companies make the mistake to keep upgrading and redesigning the UX design of mobile app but not to change the wrong back-end trade-off decisions that they might have taken earlier. Hopefully this post will help companies avoid making such mistakes.

ABC of making a company website

In this post, I will be talking about how to create corporate websites for non-technology companies only. By non-technology companies, I mean companies who do not provide their products/services over the internet and whose majority of sales are completely through offline channels. Such companies typically do not have in-house IT expertise and outsource most of their IT requirements. Yet for such companies, having a strong presence on the internet is extremely crucial in this digital age. 

The cost & time required for creating a corporate website varies widely, based on the business requirements & the IT vendor(s) chosen. Companies typically spend a huge amount to get their corporate websites created or updated after every 2-3 years. However the end results are generally very disappointing, even if reputed and expensive IT vendors are employed. The average traffic and sales lead from the website is low, the content is out-dated and the website fails to enhance brand loyalty and other business objectives. The reason for these repeated failures is not due to the IT vendor(s) but because of the mistakes made by the company during the implementation process.

Generally the Digital Marketing department in non-tech companies has very little influence. The general belief in senior management is that the tall promises given by IT vendors do not materialize and the money spent gives an extremely low return on investment. This results in a general lack of seriousness and effort in creating (and maintaining) a good corporate website, leading to a self-fulfilling prophecy. The following are the top four mistakes made by organizations when creating a corporate website -

1. The SEO (Search Engine Optimization) Requirement Document is either missing in the initial business requirement document given to the IT vendor or else is very poorly drafted. Considerable time & effort needs to be devoted by the higher management to create an SEO Requirement Document. This task cannot be delegated to any third party. It can only be done by a person having deep knowledge about of the company, its competitors, the sales funnel, customer segmentation, customer online behavior etc.
2. The dynamic content that needs to be regularly updated (by different departments) on the website is not given enough importance in the business requirement document. It is generally very difficult for the relevant non-tech-savvy departments in the company, to update any content on the website, after it has gone live. Also generally there is no custom auto-feedback provided to the person who is updating the content, making the whole process of maintaining the corporate website it seem like a non-urgent futile exercise.
3. Far too much time & money is spent on the main home page of the website and the overall branding of the website. Very less time is spent on designs of secondary pages (current and future), customer communication messages (email & SMS content), customer visible error/popup messages, two-way communication tracking, mobile viewing experience on different screen sizes & browsers and other basic hygiene issues.
4. Only standard analytic tools & add-ons (such as Google Analytics, Adobe Analytics, Mixpanel etc.) are used for getting website usage statistics & customer feedback. These tools give a number of irrelevant metrics that no-one needs to track and they fail to provide the 2-3 metrics that needs to be monitored for each department by their respective heads. The general rule of thumb is that no department can track more than 3 metrics successfully.

I like to classify the users of a corporate website into the following four broad categories - 

1. Potential customers (i.e. end users for a B2C business & business procurement teams for a B2B business) visiting the website for searching and comparing products & services at different stages of sales funnel. Of course this must be further broken down, based on the company's customer segmentation strategy.
2. Existing customers visiting website for operational issues, post-sales processes, creating requests/complaints, finding out how to use the product, giving feedback or trying to be a part of the overall brand
3. Potential investors, business partners & stock market analysts trying to understand the business model, top management, press releases and the overall value of the business
4. Prospective (and current) employees of different departments such as Procurement , HR, sales, marketing, IT etc., trying to find the culture, values, career growth path, job opportunities and contacts within the company
    

There can be several other types of users of website as well, but they are not important from the point of designing the website (E.g. creating content brief, SEO requirement document, analytics requirements etc.). The basic requirements for each of these four types of users are mutually exclusive and hence this classification is a good way of creating separate pages or sections.

There is no one standard template or process framework or a to-do checklist which can be universally applied for all cases of website development, as it depends on so many factors - overall scope and timelines, development methodology adopted, legal agreements and payment milestones, external strategy and marketing consultants, internal IT team for testing & verification etc. Instead of trying to list down the complete process sequentially, I would rather highlight the following key points that, according to me, are extremely essential for any successful implementation - 

1. The business requirement document should not just have the list of all the content for the website, but must also identify the department (or person) who will own & update each of the content types going forward and the frequency with which it needs to be updated. It must also specify the internal process and approval chain by which any content will be updated on the website.
2. The business requirement document must provide clear definition of 2-3 metrics from the website, that each department need to monitor on a regular basis and the website needs to auto-generate periodically
3. There must also be a priority order assigned for each customer need (all the different types of users and their different needs) that the website needs to cater to, as per the business objectives of the company. For example, the priority assigned to a potential customer visiting the website for product comparison would generally be higher than the need of a perspective employee trying to know the company's culture.
4. The digital marketing team needs to provide a table in which website content types are mapped against different customer needs from the website. A content type can be a part of multiple customer needs
5. The SEO Requirement Document must have answers to all the following questions for each customer need, that the website caters to - 
1. What are the top 10 search phrases that a customer will use when trying to search online for his/her need ? 
2. What are the top 3 sites (not search engines) where the customer currently goes for his/her need ? 
3. What are the keywords which the company wants to monitor in real time (so as to take corrective action if required) ?
6. Point #1, #2, #3, #4 & #5 must be the basis on which website design brief (for developers) is finalized and signed off by all relevant departments. The website design brief must not just cover the main page but all secondary pages, all customer process flowcharts, communication message formats, SEO guidelines etc. The website design brief must not just cover the external user visible parts but also internal user dashboards through which any new content will be updated on the website, by different departments.
7. The entire content, compiled initially, must be fully redesigned based on the rules mentioned in the website design brief and SEO guidelines. The content must be written from the perspective of customer need and not from company's perspective i.e. instead of just talking about your product or service, help the customer find answers to his questions. Customizing content based on SEO guidelines is far more important than the look and feel of the website.
8. Most companies do not have dedicated IT resources for UAT testing. The departments who are responsible for updating content do not have the skills to do a proper and comprehensive UAT testing. The digital marketing team or the IT team doesn't have the required bandwidth to do this thankless and boring job. I very strongly recommend companies to get another IT vendor for doing UAT testing. This results in some conflict and competition between the vendors which needs to be carefully managed. This approach is rarely taken but yields much better results overall.
9. To maintain the designed look & performance of the website, there are several constraints added on the internal department-wise dashboards/pages through which content gets updated on the website. For example - (a) freezing the font and size for text content, (b) specifying the aspect ratio, size & resolution for pictures, (c) creating a max size limit on audio, video & flash files etc. The problem with these constraints is that it makes the departments who owns a particular content type on the website, dependent on IT team, to make any change. The correct approach is to integrate web tools in the internal dashboards so that images, audio, video or flash files can be easily converted to meet the design constraints.
10. Active monitoring tools, for higher management, need to be developed to auto-generate email alarms when any content type is not being updated on the website, as per the default frequency defined for that content type. Similarly there must be periodic auto-generated SEO updates (google ranking on search phrases, presence on key identified sites etc.) and analytics updates for the higher management and digital marketing team.

I have purposely not included any technical detail in the post as website development technologies (new frameworks, languages etc.) are changing & getting obsolete very rapidly. Choosing the right technology framework is an important decision that needs to be taken based on the overall requirement, the wire-frame design and the current trends in the market. However these are implementation details that are not important for the higher management, who are only concerned with the final result & impact on the business.

I hope this post can help companies get higher returns on their investments in creating corporate websites and not make the common mistakes that so many organizations currently make repeatedly.

A blueprint for creating Smart Cities

Before we talk about 'Smart Cities', let us first understand what makes cities so special. Why are cities the engine of modern growth? Why is it that even though the world has changed so much, the ranking of different cities has largely remained the same, for hundreds of years, for any nation or region? While there are several reasons for the same, the biggest reason is the basic economics of Demand & Supply and the formation of a "virtuous cycles". Due to higher purchasing power of people in cities, as compared to their surroundings, the demand for goods & services is higher, leading to corresponding higher supply. The higher supply of goods & services are largely met either within the city or its immediate neighbourhood (the Power of Proximity rule). This virtuous cycle results in higher job creation leading to migrations from rural areas to cities. Increasing wealth in the hands of a large percentage of diverse population, leads to higher revenues with the local government which in turn leads to higher investments in non-excludable common goods (such as infrastructure), better law & order and an inclusive political system. In this way, several virtuous cycles are created in a city. Moreover the bigger the city, the stronger and larger are the virtuous cycles. The city ecosystem is so strong that it is able to withstand dictatorships, wars, famines, technological changes, economic changes etc. by continuously evolving and adapting. 

'Smart Cities' is the latest buzzword in India, ever since the formation of NDA government and the promise of 100 smart cities. However the basic idea of the State being actively involved in fast-tracking the natural evolution of brown-field cities or creation of new green-field cities, is very old and popular. Various kings/queens in the past have tried to create new capitals/cities, with mixed results. In India, Fatehpur Sikri and Daulatabad are the prime examples of spectacular failures, even after huge investments by the State. Jawaharlal Nehru wanted to create scores of modern planned cities, unfettered by the past. Chandigarh is the prime example of Nehru's vision. Various communists/soviet governments in the last century spent huge resources in trying to create green-field cities. However creating a city ecosystem from scratch, with multiple virtuous cycles, is extremely difficult. The number of failures have been far more than the number of success stories.

There is no standard definition of a "Smart City". Broadly the smart city solutions across the world, have focused on the following domains - Infrastructure, e-Governance, Transport, Basic Amenities, Security, Waste Management, Healthcare, Education etc. Most experts agree that creating a Smart City is more like a journey rather than an end goal. I am not going to try to create my own definition of Smart City or create a new framework for understanding/analyzing it. There are enough research papers and articles on it.  However I would like to point out one important and key difference between the modern concept of Smart City and earlier attempts (mentioned above) of creating planned brown-field or green-field cities. The key difference is the role of digital technologies, digital infrastructure & related concepts such as Big Data, Internet of Things, Predictive & Prescriptive Analytics etc. While the scope of 'Smart Cities' definitely includes a lot of other things, it is the digital infrastructure, digital technologies and using IT as an enabler to solve various problems, which makes its potential so huge. In this blog post, I will be focusing on these only.

For all the buzz surrounding 'Smart City' in the world, the success so far has been rather moderate for both brown-field and green-field cities. Though few people will doubt the huge potential of smart city initiatives, the on-ground implementations have not resulted in the expected efficiency, cost-reduction or usage. Citizen portals & mobile apps of city corporations for making complains, requests, bookings etc. are a good example. Their usage is less than 1% of the total population, even for the best cities in the world such as New York, London, Berlin etc. The apps/websites for public transport systems are only marginally better. Smart metering & billing solutions for electricity, water, gas etc., have been successfully implemented in several cities, but they have not had the expected change in user behaviour. Same is the case for smart initiatives in Education and Healthcare sectors, which have very low usage. Video analytics have been successful only in a few cities such as London. Same is the case for sensors, which generate a huge amount of data but their practical usage in decision making have been very low. Smart city solutions were supposed to bring high efficiency & savings, through strong feedback loops and help generate insights by combining learning from diverse systems. The most disappointing aspect of smart city initiatives so far, have been that different successful system have not able to talk to each other or learn from each other. 

Implementing smart city solutions requires active cooperation of multiple government departments and a central coordinating agency is required to be created for the task. Ministry of Urban Development, India is creating a special SPV (Special Purpose Vehicle) company for each city, to plan, align incentives, get initial buy-in from all agencies, appraise & release funds, implement, manage, operate, monitor and evaluate the Smart City development projects. Historical evidence suggest that it is very critical to make one IT vendor responsible for overall product management of the entire suit of smart city tools. This chosen smart city IT partner of SPV, must be responsible for handling project management and coordination with all the other companies and for developing customized middle layers. Without a capable smart city IT partner, investments in separate stand-alone systems/tools lead to either very poor integration or painful deadlocks & delays.

For any city, there are multiple levels of governments such as city-level (municipal), state-level and federal/national level. Their jurisdictions are separate, their budgets are separate and they are independent of each other. Many times they are controlled by rival political forces. In such a case, it is quite natural that there is lack of cooperation & planning between them. In citizen's mind through, "the government" remains one entity. Having multiple independent and unrelated customer counters, websites, portals, apps, customer care numbers, collection centres and other citizen touch points, results in low awareness, low usage, high cost and high dissatisfaction with each of them. The high level design principle for any smart city, must be to try and have, as far as possible, uniform & integrated citizen facing front-end system(s) - single window/counter for all city issues, single city website, unified mobile app with sub-apps for different services, unified smart city card etc.

The same design principle must be applied for creating front-end system(s) for all the different departments (such as urban development, environment, transport, water, medical, power/electricity, PWD, education, police, public grievance, resident welfare societies, etc.). Government departments are independent of each other and they do not want to deal with multiple tools or any additional processes or IT settings which are not directly related to their narrow scope of work. Hence there needs to be a consolidated and integrated home page, with links to all relevant tools, especially customized for each department. Creating a single login across all tools on the main home page for a department, is highly desirable but not always possible. All the IT complexities between the customer facing systems and the department facing systems need to be taken care by different smart city solutions, customized software and smart process engineering. Department hierarchies, functions, jurisdictions, rules & regulations are different for each city. Hence off-the-shelf products do not work and a majority of smart city solutions need to be custom developed for most cities. Moreover even if off-the-shelf products are used, they can hardly ever be updated to their latest version over time, given the complexities of the overall system.

Making even a very small change in the standard processes of any large sized department is very challenging. This is especially true for government departments which have a huge legacy/old issues and are constantly under the scrutiny of Judiciary and auditing agencies. Moreover in case of smart cities, the change need to be driven by an outside agency which is even more challenging. It is far easier and cheaper to make changes in software suite, even if they have to be reversed later on, rather than make any change in the existing processes of any department. No smart city solution has the luxury of starting from a clean slate (not even for green-field cities or private cities). Hence proper change management is very essential. The key is to have only very small incremental changes and to make sure that all open and legacy issues of the old system are carried forward with each incremental change. Any solution which will not take care of all pending legacy issue of all impacted departments, at each incremental step, will fail miserably. The process changes must be small and frequent and must be very carefully planned and monitored. If there is more than 5-7 weeks gap in between any two incremental changes for a department, the compliance level to the latest change goes down drastically. It is very essential to keep re-enforcing the complete long term vision and how it will help all departments in the end, when implementing each incremental process change.

Given the dynamic nature and pace of innovation in the smart city domain, it is impossible to come up with a comprehensive list of all requirements at the beginning of the contract. While the citizen facing requirements don't change very frequently, the department specific requirements change very frequently (sometimes due to new regulations, judicial decisions, RTI requests etc.) and need to be fixed very quickly. Hence it is extremely crucial that the turnaround time, for new 'Change Requests' to vendors, is very low (approx. 3 days). The initial contract for all vendors need to be very carefully drafted. It has to be a mix of milestone based and effort based contract. The formula(s) for pricing future change requests must be defined in very unambiguous terms for each vendor.

The real benefit of smart cities tools and big data analytics comes when all the different smart city systems such as complain/request ticketing system, billing, collection, asset management, websites, mobile apps, call centers, CRM, sensors, command centre, video management systems, GIS, energy/water meter monitoring systems etc. are strongly inter-connected and use the same underlying data layer. The underlying data layer can be logically separated into two categories - entity-based (citizen database, department employee database etc.) and GIS-based (housing database, camera location database, electricity grid, water pipeline grid, road & sewer network etc.). The cost, both monetary and performance, of creating customized software to interconnect different systems is quite high, but very essential and unavoidable.

Data consistency is a far greater challenge for smart city solutions unlike any other software implementation. There are huge data errors (missing data, wrong data, formatting issues etc.) in all legacy stand-alone systems. Proper handling of all inconsistencies during one-time data transfers from legacy system is very challenging. Also for most of the smart city solutions, there is never a clean one-time transition from one system to another. Maintaining data consistency during long and multiple transition periods is important. Implementing smart city solutions will always involve multiple data clean-up exercises at different stages of implementation. Without advance manpower planning, periodic data-cleaning exercises will become a big bottleneck in implementing smart city solutions.

Citizens physically interact with government departments (municipal, state or central) only when they have a particular need such as paying electricity/water/gas bills, getting NOC certificates, registering a complain/request, booking government amenities, buying or renting a new apartment etc. All these touch-points should be used to add/update the same back-end city resident database, especially photographs. All forms, that are required to be filled by the customer, at each of these counters, must be pre-filled with the personal customer data, as available in the database. Pre-filling form and asking customer to correct any mistakes, is the most in-expensive way of database cleaning and pro-actively updating city database of changing mobile numbers, address, email-ids etc. One of the big advantages that Indian cities have over many other cities in the world, is the availability of Aadhaar database, which ensures that the duplication error is only possible for corner cases, where linking with Aadhaar is not possible.

A significant percentage of human dwellings in any city are either temporary or illegal or under ownership dispute. There is no unique number or identifier given to each human dwelling in a city, similar to Aadhaar numbers for every person in India. All legacy systems have a unique, non-standard and non-compatible way of storing addresses. However the address list in the electricity billing module for any city, is a very good substitute, which covers almost all human dwellings. Anything which is left can be clubbed under 'homeless' category. Electricity bill address database must become the address data layer for all smart city solutions. Data clean-up exercises need to be taken to make sure that address field in all other databases such as water billing, cooking gas connections, landline bills etc. are mapped to a unique smart city address as far as possible. The smart way of doing this is to make the residents do the change themselves when paying different bills online. The more difficult part is to save latitude and longitude for each address. This remains a manual door-to-door effort. 

Even though we can get a fairly accurate database of residents in a city and all addresses in a city, the mapping between the two is almost impossible to maintain, especially since a majority of population lives on rent under formal or informal agreements. Similarly mapping residents into distinct families is impossible to maintain. These mappings should not be a part of the basic data layer to be used in all smart city solutions, but should rather be left to each application which needs this information. Private cities where all apartment are sold by a single (or a few) companies, can maintain a much more accurate databases. Through proper planning of NOCs for resale & tenant registration process, they can track addresses of both owners and tenants. This does have a major advantage in creating customized applications for citizens on top of smart city software layer.

One of the most basic smart city solution is a ticketing tool for all non-emergency services in the city, such as complains, request, amenity booking, events registration etc. In many cities in USA, these are called 311 services. In India, most people prefer to physically go to government offices for their problems. They don't trust that registering complains through call center, official website or official mobile app, can get their problems solved. Complain counters in government offices or call centre have a very high operating cost and are highly inefficient in accurately recording the actual customer need for back-end fulfilment teams. Citizens are not good at typing the exact problem and hence they should be allowed to voice record their issue and subsequent comments on those tickets. To avoid the problem of tickets been marked resolved, without actually solving the issue, getting confirmation code from the ticket raiser, must be made compulsory for most cases. It is surprising that such simple common sense features are missing in most smart city solutions in the world. Of late there has been a growing tendency to use social media (especially twitter) to get the attention of governments. Modern smart city ticketing tools need to have plugins for popular social media sites as well. No citizen facing work that gets done by a department, should be allowed to by-pass the ticketing system, else the ticketing system cannot be used for measuring performance of departments resulting in a failed IT project.

Setting up and managing CCTV camera network is very costly, with very low returns on investment. As the number of cameras increase, it becomes virtually impossible to monitor them in real time, even when advanced video analytics are used for generating alerts. Image processing technology still hasn't matured enough, especially for countries such as India with high population density. The biggest disappointment with CCTV network is that in many cases, it is not even able to help ex post facto, as the camera was damaged or there was a connection problem with the fibre optic network or there was an obstruction in front of the camera or the resolution was too poor to be of any use etc. A good solution to this maintenance problem is a video management system which automatically creates tickets in case of any issue with the camera and shows the location of the faulty camera on GIS maps in the command centre.

The number of types of sensors have increased dramatically in the last few years. Commonly used sensor by cities, are sub-station monitoring sensors, gas leak sensors, fire alarms, water flow sensors (both drinking and flushing), parking sensors, street-light sensors etc. One of the biggest challenge with sensors is to keep the number of false or non-actionable alarms to a minimum, by correctly configuring/calibrating them. This takes a lot of time and effort. It is essential to separate warnings from critical alerts. Many sensor implementations fail to make any difference because of the sheer volume of low priority alerts on the dashboard, generating multiple alerts for the same issue and lack of a short SOP for dealing with each type of alert. All sensor inputs must map to only a few common screens (city GIS maps) on the command centre. 

In many cities there are completely non-personalized smart cards for various public transport systems (E.g. Delhi metro smart card). In most cities one card is used at multiple transport systems, making payments at various merchant locations, making online payments, loyalty programs etc. Non-personalized cards makes administrative tasks easy, but have a very limited account balance limit (only Rs. 10,000/- in India) and cannot be used for a number of other applications. Oyster cards in London or Octopus card in Hong Kong are great examples of personalized contactless smart cards for all residents of a city. Apart from payments, such personalized cards can also be used for access control for commercial and residential buildings, implementing fines & penalties etc. The long term goal for smart cards should be to have only one unified card for the entire city which is linked to as many applications as possible. Printing photograph, full name or any other personal detail on the smart card, makes the issuance process very difficult, costly & can lead to customer inconvenience. The technology changes in contactless smart card domain are very rapid (there are 8 categories of MiFare cards alone) and backward compatibility is not always guaranteed. There are multiple types of POS devices, turnstiles, access control devices etc. that need to be compatible with the smart card. It is a challenge to make sure that all systems & applications, using the smart card, are upgraded to the newer technology, after every 5-6 years.

It is not possible to cover implementation challenges & learning for all the different smart city domains in one blog post. Smart city solutions in education and healthcare domain have little or no integration with the rest of the solutions. Other smart city solutions such as utility billing, paid parking, smart traffic lights, fibre optic based metering, IOT applications, big data analytics etc. are tightly coupled with each other. Apart from creating these basic smart city solutions, an open-to-all free smart city platform needs to be developed. The platform should enable anyone to use the APIs and data exposed by the smart city platform, to develop their own applications, tools, analytics engines, research papers etc. In this way smart cities can crowd-source ideas, incubate start-ups and co-create application with a large world-wide developer community. A good example to emulate is Singapore’s open data policy[1] and one map[2] initiative.

Effective, timely and detailed citizen communication is very important for successful implementation of smart city projects. Traditional marketing medium such as billboards, print, television & radio are not sufficient. To get effective feedback from early adopters (NGOs, active citizens, 3^rd party contractors etc.), two way communication medium such as discussion forums, email groups and social media, needs to be adopted. For any effective communication strategy, smart cities must have a fairly accurate (it can never be 100%) database of email-ids and mobile numbers mapped to each citizen. Smart cities cannot rely on citizens informing them when their email-id or mobile numbers change. A permanent call-center is required to continuously call or email citizens and proactively keep the database updated. Every time a communication message by any smart city application to an email-id or mobile number fails, they must be added to the list of faulty email-ids or mobile numbers that need to be corrected.

Smart city applications and tools stack on top of each other. Certain basic applications and data layers need to be developed first before the rest can be built. However the exact sequence, in which applications/solutions must be developed and implemented, should vary from city to city based on the needs of the specific city. For example there is no point in having a smart card, if you don’t have a ubiquitous public transport system(s) in the city. In cities where the population is less than 20k, smart city systems such as RFID tags parking systems, smart water meters etc. will not result in appropriate return on investment. The smart city solutions which solve the most pressing need/concern for the citizen of that city must be prioritized over others. The need can be security (especially for new green-field cities) or transport or governance (for brown-field cities). 

As mentioned earlier, cities flourish because of virtuous cycles of high demand & supply within and around the city. The city must provide jobs and all basic minimum resources, which will then attract others to migrate to the cities. Very often city planners neglect the aspect of job creation within the city and give over emphasis on creating amenities. Moreover they focus on malls, shopping complexes, hospitals, colleges etc. but neglect basic facilities such as last mile transport (auto, rickshaws etc.), schools, play-schools, crèche, laundry shops, household helps, local medical centres, groceries, electricians, plumbers, local wine shops etc. Many of these basic amenities require high subsidization at the initial stage. Investors in housing projects, who do not wish to live in the city, are a huge risk in creating city ecosystem. If the right city ecosystem is not created, it will end up creating ghost cities such as Kangbashi and Thames Town in China. New suburbs (such as Bandra Kurla Complex, Mumbai) or cities (such as Greater Noida) existing close to an old city are highly dependent on good and multiple connectivity to the old city. Independent green-field cities (such as Amravati) are highly dependent on governments to create new jobs and kick-start migration into the cities. Smart city solutions can only act as an enabler for achieving higher efficiencies, but they cannot be used to create city ecosystem. This is the reason why most of the smart city solutions work better for existing cities rather than new cities.

Once a smart city solution has been successfully deployed, monitored and stabilized, it needs to be handed over for operations to the respective departments, except the department for data correction which need to remain centralized across all smart city solutions. Some solutions such as waste management, smart electricity & water meters etc. come fully under the domain of a single department and cost/benefit allocation is very simple. However most other smart city solutions such as smart card, mobile app, website, GIS systems, RFID parking solution etc. cut across multiple departments. Special accounting mechanisms have to be developed to distribute revenue, manpower required, operational costs etc. of these solutions across departments and governments.

Even through currently the ROI (return on investment) on smart city projects do not meet expectations, I have no doubt that investments in smart city projects will only increase in foreseeable future. Many companies will come up with products which are specially customized for city use cases and not corporate use cases as most IT products are currently. The recent surge of small townships and private cities built by private developers in India, will lead to greater development of smart city applications, as decision making in a private company is always faster than any government agency. There are several other aspects of smart cities that I have not been able to cover in this blog, especially non-IT related areas. In this blog I have mostly tried to cover aspects of smart city which are different from traditional city planning. A lot of work needs to be done before the dream of 100 smart cities in India can be realized. I am eagerly looking forward for the day when I can enjoy most of the smart city solutions described above myself.

---

[1] https://data.gov.sg/

[2] https://www.onemap.sg