Nurture your digital product profitability
After the start
Six months ago, you had a great idea. Something that your customers will love and that will bring them a lot of value.
You then work on the product proposition, business case and its profit formula and launch its creation. The proposition was clear : initial investment, break event point and revenue forecast.
But now, after the first launch, the situation seems less clear.
Your company is not a startup or a digital native business. Due to its size, the agility is lower. Like the image of the lotus-eaters, it sometimes seems that time pass quickly inside the execution machine of the company and keeping a good pace and Time to Market is challenging.
You like to have an agile approach of the development of your features, with MVP and incremental releases … But several business financial structures needs you to group features together to justify or receive budget : Multi-year budget (MYB), Self-Funded Research & Development budget (SFRD), Capital Expenditure (CAPEX) ..
You need also, in a certain way, to manage the time and schedule of the teams because processes like SAFE, even if they are based on agile methodologies, are often executed with a mix of projectized and functional organization. Due to that, the priority management and manpower plan requires you to anticipate the workload in regard to the product or total backlog.
Several teams like Network Operation Center (NOC), Maintenance Operation Center (MOC), Security Operation Center (SOC) … are assigned to a set a product and manage their time allocation with Kanban or other processes.
But, most of the time, the charge back is difficult to link to your product plan (incorporated in hour rate, total number of incidents …).
This create a “Hammer effect” where charges are assigned back to products based on company cost center more than real activity
Finally, it seems to you that delivery pipeline is leaking :
- Processing power provisioned but not used – Up to +30%
- Licenses spare provisioned but not yet used – Up to +10%
- Infrastructure spare (network, storage, …) – Up to +10%
- Wrong licenses type for development and test – Up to +13%
- Solution idle during non business hours – Up to +22%
- Computing power on demand insted of provisioned – Up to +30%
- Non negociated volume discount – Up to +10%
So, with simple math, you see that the cost of the product is almost twice the effective target.
But, as this challenge is intertwined with the changes in your deployment schedule, the comparison with your initial plan is more than difficult
So now ?
To address this challenge, you need to review your financial digital strategy.
For that, I recommend to correct the situation with 3 actions:
- Use a definition framework to recapture your needs
- Use an organization framework to replan your operation
- Use a model to drive by data
Let’s review then together
Recapture your needs
As the product is digital, I strongly recommend to use specialized framework like The Well Architected Framework of AWS that is focusing on 5 pillars :
- Operational Excellence
- Security
- Reliability
- Performance
- Cost
For this specific challenge, we will look at this cost part that is accessible at this URL:
https://wa.aws.amazon.com/wat.pillar.costOptimization.en.html
Along with an exhaustive description of the subject, the framework comes with a set of 10 questions :
- How do you implement cloud financial management ?
- How do you govern usage ?
- How do you monitor usage and cost ?
- How do you decommission resources ?
- How do you evaluate cost when you select services ?
- How do you meet cost targets when you select resource type, size and number ?
- How do you use pricing models to reduce cost ?
- How do you plan for data transfer charges ?
- How do you manage demand, and supply resources ?
- How do you evaluate new services ?
As you see, those questions will help you look at the source of the effects that you observe.
Replan your operation
Several dedicated processes have raised with digital like
- DevOPS for collaboration loop between operation and development
- SRE that use engineering technics to organize operation
- FinOPS that focus on financial accountability.
One this last one, the FinOPS foundation (https://www.finops.org/what-is-finops/) have a nice documentation on this.
The framework is based on 6 simples principles
- Teams need to collaborate
- Business value of cloud drives decisions
- Everyone takes ownership for their cloud usage
- FinOPS reports should be accessible and timely
- A centralized team drives FinOPS
- Take advantage ot the variable cost model of cloud.
If all those elements are new to you, I strongly suggest that you take the time to read the introduction.
I will however insist on two of them
Accessible and timely
Observability is key in digital, and financial data is no exception. Also, proactive behavior (look for data and decide with rules) is prefered to passive behavior (log and observe later) as the time to react will highly influence the size of each drift.
For each of your financial measures, your goal will be to use automated tools to create the measures that you want to use. For that, you can use native cloud tools from your cloud provider like AWS or Azure cost management, or multi-cloud solutions like CloudHealth or CWOM.
They will help you not only to measure but also to take actions when discrepencies are found.
Variable cost model of cloud
In digital, your direct demand (customer) or your response to the demand (processing), will vary overtime.
Any excess provisioning needs to be considered as a loss and avoided as much as possible.
Otherwize, as highlight in the first part of this post, your investement will not be efficient
Drive with data model
In the post From data to data model, I explained how data models are tools to move from history observation (measure) to forecast.
Your financial model is the profitability driver of your product and need to be based on the business data input of your product
To understand this, let’s look at this view provided by Aviation Week Network based on the IATA avionic organization forecast
As you can see, before COVID, the business projection of the industry was really promissing and as digital products were having their own high take rate forecast, many business models were having a break heaven point (BEP) really promissing.
With the new industry forecast, all business models need to be reeavaluated and the evolution of the expense realigned with them. But this can be challenging if you have not separated deployment, development and take rate impacts in your demand model
This is, of course, an unusual scenario that have strike the entire industry, but this illustrate the importance to have an automatic data model that can evolve after the initial product launch assessement and drive the execution and operation of your product
The financial model will define the target shape of each of the measure (average and range of uncertainty) so that you can detect anomalies and/or anticipate actions.
For example, if you add all product minimal computing consumption for the incoming year, you can define the needed reserved capacity that you can buy from your provider and reduce by 30% to 70% the associated expense.
What else ?
An expresso ? Maybe for your personal comfort, but more seriously, from those 3 first actions, you need then to be sure that this redirection effort is persistent.
In this post, I have not touched the others factors that concur to the customer satisfaction (user experience, functions, operation, security, reliability and performance), but they are of course also essential to the success of the product.
However, if we stay focused on the financial profitability, I would like to come back to on example of leak listed in the first part of the post.
On the example of the licences, the success of this management will rely on the knowledge of
- type of licence (functions availables)
- granularity of the licence (by year, by user, by instance …)
- usage condition (limits, reasignement …)
- ramp up/down speed (time to procure)
- …
We use generally the term “on demand”, but as you can see, the condition of the on-demand will generate different type of questions that you will need to answer with the model.
Also, as you have probably realize, you will provide the global direction but each of your team member will be part of its success and the communication is key.
This can be achieved by standard communication or by simply giving secured targeted access to the tool to them.