The Cloud is not a Place!

Wonder why some companies seem to "get" cloud while others struggle? Why do some companies seem way more innovative thanks to the cloud than others? Having spoken to a number of companies who have implemented some solutions on cloud or are implementing cloud based solutions, I see a few trends. Caveat before I start - I use cloud in this article to mean public cloud though of course in reality customers will usually implement some form of hybrid cloud.

1 - cost as a decision criterion still exists but is coming down in importance. For example, the 2016 Rightscale State of the Cloud report shows cost savings and moving from capex to opex as the last and 3rd from last reasons out of the top 10 reasons to move to the cloud.

2 - people are coming to accept that cloud security is better in many ways than they expected and in most cases have a sneaking suspicion that it is better than what they can do in their own data centers. This is also reflected in the Forbes cloud survey of surveys which shows that security concerns while still serious are no longer the #1 issue against cloud adoption in customer minds.

3 - data growth on the cloud is actually keeping pace with hype. A HPE study estimates that by 2019, 90% of all mobile traffic will be due to cloud apps while a Cisco study estimates that 83% of all data will be on public clouds by 2020.

4 - why is that relevant? Because, once a critical mass of data has moved to the cloud, what is often referred to as "data gravity" kicks in where the critical mass of data attracts more data in a virtuous cycle. A fundamental principle of data is that its value reduces as it gets moved and copies are made because then the currency of data and accuracy both get degraded. So using data where it is and bringing tools to meet the data rather than move data to the tools is a basic architectural principle that many companies have come to accept.

5 - the Rightscale report echoes what we see in other surveys and hear from customers daily regarding why move to the cloud. The top 4 reasons therein after which there is a big drop are :

• Faster access to infrastructure
• Greater Scalability
• Greater Availability
• Faster time to market (for IT changes)

From all of the above we can see that cloud is increasingly seen as more than just a cost saver. That brings me to the nub of what I wanted to focus on i.e., too many companies we speak to however still see the cloud as ways to improve IT but not to improve the business itself. A closer look at the top 10 reasons is very revealing in that except the point on "Improving geographic reach" (a lowly #7 in the report), not even one is about improving the business with say better analytics for decisions making or introducing new business capabilities like mobile apps for employees etc.

The inescapable conclusion therefore is that most companies still see the cloud as a "place" to outsource their concerns such as cost / skills / management headaches / infra reliability / performance issues / growing storage needs / etc., to and not as a way to reimagine how business can be done. Of course all the current IT related reasons are important and minus those table stakes, any innovation on the cloud using say Location Based Services would be a failure. So point is not to belittle existing services / reasons to move to cloud but to get companies to see how better to envision their business on the cloud.

Some interesting examples we are working on may help make the point. A large IT services company wants to move their HRMS application to the cloud. The initial trigger point is cost because today maintaining this system for over 50,000 people is verily a nightmare in an industry where 70% of the cost is people related. Ergo, most new innovations are sought in the area of people management / development but the current application cant provide the base for a modern EX (Employee Experience). After some initial meetings and discovery sessions with the users, it transpired that many of them want to use the inflexion point created by the cost pressures to take a fresh look at the way they currently manage their EX. A number of new initiatives have sprung up as a result such as innovative recruitment models, AI driven matching people to roles, continuous feedback loops between managers and teams etc. This is clearly a company that sees value in doing something different with the cloud other than just getting the cost monkey off their backs so to speak.

Another example is an automotive company. The auto industry has refined the supply chain a lot and while yes, IoT / GPS tracking etc., can be added to fine tune the same, it would be fair to say their supply chain practices are quite advanced. On the other hand, customer facing functions have some way to go at least in developing countries. In this example, they want to add innovative new business capabilities such as IoT based insurance pricing (using IoT data to assess driver performance and reward customers with better driving through lower premiums). Their targeting today is largely mass media based and that is both very inefficient and ineffective. They are therefore looking at much more precise prospecting using enhanced data from 2nd and 3rd party sources so that when even an unknown prospect searches for car reviews online, based on their digital signature and mapping the same to known profiles of customers, they can be shown very relevant ads/offers and enticed to the company's site. This is fundamentally different to mass marketing. This company clearly sees going to the cloud as a gateway to new ways to do business.

A third example is of a financial services company which is looking to move from high net worth customers to the next segment but then cant expand the salesforce proportionately for cost and productivity reasons. They are looking to a very digitally driven lending process that goes all the way from targeting to smarter analytics to do more precise underwriting to document management for the paperwork to very intuitive mobile solutions for the salesforce down to learning algorithms that look at loans that turned sour to see what went wrong to refine the targeting and underwriting process. Such a project would mean reinventing their existing manual intensive and error prone processes for the new segment. They had a choice to simply expand their current IT set up for the new segment but realized that such an approach wont cut it. So they are embarking on a fresh lending platform ground up on the cloud.

Can all these example be done on premise? Very unlikely because companies have realized the limits of attempting such projects in house. What are these limits? I wont go into the obvious ones about elastic capacity, speed to provision, opex versus capex, skills sets, security etc. I would want to instead focus on 4 reasons that are not so evident at first blush:

Data Driven Decision Making - As companies increasingly cope with increased transaction volumes thanks to omni channel operations, big data etc., there is a drive to increase the reliance on data in decision making. May seem obvious at the outset but it is a little more complex. In the past, the time it took to release a new product or tweak existing offers / pricing or alter packaging etc., was driven largely by constraints in the "real" world such as time to get dealers / employees trained, do small test marketing followed by surveys to see what is working / not, ensuring physical availability of product etc. The other "real" world issue was the time it took to get the data from test market surveys, dealers, employees etc., back, to analyze and tweak the prices / offers etc. To get to market faster, customers therefore usually took a look at whatever data was readily available even if it was aged / incomplete and took decisions based on experience even if that was sub-optimal. With the cloud and related digital technologies this process can be reimagined substantially. For example, with many buyers (not just millennials) turning to online channels for opinions / info / comparisons etc., tapping into their digital behavior (even if anonymized for privacy reasons) throws up a lot of data in a very short time to be able to test new ideas. Analytical tools allow you for example to compare multiple versions of offers on slightly varying web pages to which you randomly send online shoppers / browsers to see which one is more sticky, which one tends to draw the most leads / checkouts, which one seems to be most attractive to your preferred customers etc. Point being that bringing in data driven decision making is premised on rapid ingestion of data and analysis failing which people will fall back on gut based decisions. Not to say that gut based decisions will need to go away or will disappear. Au contraire, in many use cases human intervention is still essential. However, by the same token in most decisions, data can help humans be more precise, a point that more experienced but less digitally savvy people tend to ignore / deny. But then what has all this to dow with cloud? Why not just do all this with on premise tools? Because, cloud based decision tools are a lot faster in this regard for 3 reasons.

One, provisioning the resources to quickly ingest data from a new source is a lot easier on cloud than on premise. Two, many new tools coming from IT companies are developed with a cloud first mentality (they may in some cases provide on premise versions too but with a delay if at all). So if having access to the latest intuitive visualization tools / self learning algorithms is essential in that specific use case, then the cloud is a better bet. Lastly, most of the analytics will have to feed back into operational systems either in batch mode (to help create new underwriting models or better segmentation models) or in real time (intervene with the best offer even as a customer is browsing a site or use a chat bot to suggest an answer even as a chat is in progress). Either way, with SaaS increasing rapidly (the Cisco report shows SaaS going from 45% of the cloud market today to 59% in 3 years even as IaaS drops from 42% to 28%), data gravity in CX and EX apps is increasingly moving to the cloud. So it is inevitable that the analytical tools will follow the money (or data in this case) to the cloud. That is a nice segue into the second reason.

The nature of COTS (Commercially Off The Shelf) solution implementations is changing - In the past if a company was implementing an on premise COTS application like ERP or CRM or HRMS, they identified ways in which it had to be modified to suit their needs (statutory / best practices / "that is just the we were and will always be") and then customized the core of the application in most cases. Some smarter companies used safer customization techniques to minimize the impact on the core but in all cases it was a surgery at a minimum (open heart in the first case and with local anesthesia in the second case but a surgery nevertheless). Extracting data from them for analysis became tougher. Tougher still was to push data back into these applications to suggest new segmentation / underwriting models or upsell offers in real time as mentioned above. As a result, analytics tended to become a separate function in the IT department (or implanted within user departments but as a separate group) peopled with quant gods whose world and that of their users intersected only minimally. Therefore the impact of the analysis was limited at best. In the cloud, a basic reality has been altered making the potential impact higher and faster. What is that?

SaaS applications cant be customized like on premise COTS applications. So the surgeries that were performed in the past on premise are not possible anymore. (As a side note, some degree of personalization etc., is possible in elegant SaaS solutions within limits of reason but nothing that alters the core of the application but that is a separate blog topic!). So customers have to perforce rethink a few things. For starters, is the function being covered by the SaaS critical to my value chain or only a supporting function? That will decide if capabilities beyond the SaaS are really needed or they can make do with what is in the core SaaS. Even if changes are deemed necessary, how do you go about adding the changes?

SaaS solutions will require you to create separate bolt on applications (using PaaS / IaaS) for such changes which can then talk to these SaaS pieces without altering the core of the SaaS application. Why does this matter? Because such an architecture ensures that the SaaS pieces remain pristine which makes it relatively straightforward to extract data from them for analysis or push data back into them for interventions. That way the time it took to alter the heavily customized on premise COTS apps to extract or push back data is dramatically reduced. Better still, business users can see analytics as part of their standard application screens reducing the ivory tower nature of analytics in the on premise world. For example, let us say your SaaS Procurement application does not provide out of the box analysis of the quality of material from the vendors being compared. With a BI PaaS standards based platform, you can create a small quality comparison tool that can be inserted into the standard quotation comparison screen being used by the buyer with minimal effort and more importantly creating minimal roadblocks for future upgrades. The key here was the use of a standards based platform. That brings us to the third reason.

Standards on the cloud - In on premise applications be they COTS or your own custom built applications, how you got them talking to each other was very company specific or in some cases not even that. While that may be fine for business as usual up to a point, integrating newer capabilities then becomes a nightmare because you don't know how touching something may affect other integrations. This to a large extent is because not all on premise applications followed industry standards for integration. Even where they did, in many cases custom extensions were built without following good practices with the result that integration became complex. In the cloud especially where SaaS pieces are concerned, they tend to have clearly defined interfaces through which other SaaS or custom PaaS pieces can communicate. No back doors are possible. This makes integration quicker, simpler to maintain and less prone to errors. Point being that the cloud was an inflexion point that gave vendors a clean slate to start with, thereby allowing them to take standards based approaches to the development and deployment of services.

Frequency of upgrades - In the on premise world, we frequently see business users very upset that IT does not upgrade regularly to take advantage of new features. The root cause for such delays is that the customizations that have been done mean that any upgrades have to be done carefully to avoid undoing the customizations or acting at cross purposes to the customizations. An ERP customer that I know of has long wanted to take advantage of our transport management capabilities but because IT has refused to upgrade for a while now, they are stuck as the capabilities are only available on more recent releases. IT in this case accepts there is an issue but cant do much as the legacy of customization means that most of their time is spent keeping the lights on leaving little time to even contemplate leave alone add new capabilities through upgrades.

With the cloud, that is a far easier proposition because upgrades are far simpler on SaaS pieces in particular since they are not customized. This means they can be upgraded with little or no impact on the custom add on PaaS pieces (though yes in some cases, some testing would be required to check that performance is not affected etc.).

Smarter companies have noted these 4 factors that make the cloud more than just a "place" to dump their problems. They have used various inflexion points in their IT / business like an impending upgrade or a new business venture or a slap on the wrist from a security auditor or a competitive threat to revisit basic assumptions about how they do business. They are as a result benefiting immensely. Amazon Web Services (AWS) conducts excellent workshops to help customers plan transformative projects on the cloud. Please contact your AWS sales rep for more details.