Commentary Archives - Assert Digital Ventures

How can we balance restricting harmful/fake content with privacy?

December 9, 2019December 9, 2019
by Andy

There have been two major issues in tech over the past couple of years which are highly related yet I haven’t seen much talk about their interplay: how to rid (primarily social media) from harmful/fake content and maintain people’s privacy. The former coming to prominence after the 2016 election “influence” campaigns and subsequent revelations around Facebook, YouTube, Twitter and others trying to deal with fake accounts.

Allowing users to mask their identities has been a hallmark of social networks and user generated content (UGC) sites since their formation. After some egregious cases of harmful and hateful content arising early on, each made tacit attempts to ensure identities. Bots and incentivitized users easily worked around them. Few — notably Airbnb after some high profile problems — went to a REAL ID system. This requires users to present multiple forms of government issued ID to establish their identity. REAL ID isn’t perfect, but it certainly reduces the number of fake/anonymous accounts.

On the other side, there has been concern around user privacy and the use of personally identifiable information (PII) that these sites collect. Once again, Facebook unfortunately took the lead here in allowing data to be used in malicious ways demonstrating how dangerous companies controlling PII can be. Since then, there have been many voices, including governmental, calling for significant privacy constraints and for users to “own” their data and identities.

These types of controls are easier said than done. However, it demonstrates that people are not comfortable with big tech (and others for that matter) having this information and, either through malicious or incompetent behavior, may misuse it. This, unfortunately, is in direct conflict with the desire to have more transparency and identity on social media to attempt to limit harmful/fake content so as to better police it. What to do?

My proposal is for a multi-layered system which addresses both issue:

Any internet property which allows users to post content (e.g. social networks and UGC sites) must verify their identity using a system like REAL ID.
The company may only verify the identification documents but may not store any PII information associated with them. They must keep their own internal ID for each users which has no way of being reverse engineered into PII.
Each company must ask (and give access to change at any time) users what information about them that they’re allowed to store. This includes what public persona/name/nickname the user wants displayed along with their content.
The government will establish a system for REAL ID based on the issuing entity. When an ID verification request is sent to the government API, valid requests will have a unique set of characters returned to the requesting company (we’ll call this encrypted ID) which they MUST store with the company’s internal unique ID for that user. The government entity must keep storage of all encrypted IDs and their association with REAL IDs. Each requesting company must get a different encrypted ID but a consistent one for the same user on subsequent requests.
If there is suspicion of a crime, a law enforcement entity with a subpoena can request the encrypted ID associated with any account. That can then be used to reveal the REAL ID already stored in the government database.
Each REAL ID API call by companies will incur a small fee so as to pay for scalable and secure data systems that the government must maintain.

This system isn’t perfect but will a) curb bad behavior online with the knowledge that if you cross a line, you could get a visit from law enforcement b) greatly restrict bots and malicious actors from gaining accounts c) support a government program to protect IDs but associate them with real identities they already issue and protect d) do not allow companies to hold PII which could be used maliciously or hacked.

Are there people who would lose their important anonymity under this plan? You don’t have to keep any PII with the company, only the government (and you do this today)?

What about authoritarian governments and how they could abuse this? Yes, that’s a concern but is also a larger issue. In any case, this does not address private messaging apps and other types of sites, only public content ones.

Thoughts?

My appearance on Reuters CCTV talking about next day…

December 3, 2019
by Andy

I was recently interviewed by Reuters CCTV for their segment on Ohi, a startup building micro-warehouses in urban centers to provide non-Amazon ecommerce with next and same day delivery.

Reuters CCTV

How Apple, the ultimate product company, is a functional…

September 17, 2019
by Andy

The org-du-jour for modern companies is rarely a functional one. Considered outdated as traditional 20th century tops down management, functional organizations are an outgrowth of the need for large functional workforces. When you needed to produce widgets at scale, you needed large numbers of workers, managers of those workers, managers of those managers and so on. The hierarchy and its function reinforced the structure: do what you do efficiently and leadership will ensure it comes together as an end product.

The rise of the knowledge worker, analog to digital transformation and automation changed that. Suddenly, businesses could achieve massive scale with a relatively light workforce. The new worker was not a heads down widget assembler but a highly educated person with ample tools and abilities to create. Thus, the hierarchy was both torn down not only by workers not wanting to be a cog in the wheel but also by managers cut from the same cloth willing to experiment to drive more innovation out of their quite able workforce.

Tech companies tended to lead this charge as massive margins and scale allowed them to toy in areas away from the core business but also retain highly valued employees by affording them flexibility and discretion, to a degree, over what they worked on. This tended to flatten organizations, push decision making down and tear out the base of the traditional hierarchy, mostly to great success, reinforcing its use. Many traditional companies — under pressure of and trying to replicate the innovation within tech companies — are wrestling with their own hierarchies. If not to bring this culture but also to attract and retain the highly valued top-tier knowledge worker.

Amongst tech companies, Apple is a notable outlier. They have retained and thrived under a mostly functional organizational structure and hierarchy. They have even resisted the trend to split into divisions with functional structures like Microsoft did. Some of this was born of tradition being one of the old guard tech organizations. However, how can a company thought of by many as the ultimate tech product organization, thrive with a structure many of their peers consider outdated?

One could argue that Apple is the ultimate manifestation of the functional organization due to their focus on a seamless customer experience. Many tech companies expand from a core base product (e.g. Google’s search, Facebook’s feed, Amazon’s e-commerce store) through ancillary products and services that leverage the network effect. As they experiment with other products, they build up mini-organizations that are mostly self-contained for each of those products. That allows them to iterate and evolve faster and somewhat independently of the core. This is a mostly successful strategy (esp for the three examples above).

However, there is a downside. If your core and side products evolve via largely independent teams, their experience will also diverge creating a disjointed customer experience. Google realized this a decade+ back as Gmail, Maps and GDrive/GDocs became major products. A user moving across them (if they even seamlessly could) might as well have been moving across lightly integrated offerings from different companies. Recognizing this weakness, a team of designers formed a group to unify the experience. It took years and I’m sure a lot of internal arm wrestling but the result was Material Design and now a (more) unified experience.

Apple is the epitome of the unified experience. They marshal a massive, highly skilled workforce to create focused products. They are rigorous about seamless integration through the whole experience which is often their only differentiation. The way they achieve this is at the top of the functional hierarchy. In many ways, this is Steve Jobs’ greatest legacy for the company. His, at times ruthless, focus on experience, unified his executives and thus his teams. They built highly focused and skilled organizations (e.g. chip engineering, user interface design, etc). Because no one (but the execs) had responsibility for the overall experience, they didn’t afford anyone to go off the farm from the core design. In fact, they often kept other functional teams in the dark about what the others were working on only agreeing to limited information and “interfaces” between them for necessary integration.

Thus, one could argue that Apple’s products are what they are due to the functional organization as opposed to being hindered by it. Is it replicable? Perhaps, but the argument that this is Jobs’ greatest legacy has merit. Over the decade or so since his direct involvement waned, the organizational design and culture continues but one could argue that the lack of innovation from Apple since — with mostly incremental and accessory products & services — is due to not having him drive the integrated experience over a functional organization from the top.

How Each Big Tech Company May Be Targeted by…

September 15, 2019
by Andy

Of these four:

Amazon: Favoring its own products?
Apple: The power of the App Store
Facebook: Consolidation of social media
Google: What appears in search results

Google really presents the biggest concern. For better or worse, search is still the gateway to much of the monetization of the internet. Yes, Amazon exploits is master view of its marketplace to sell their own products. However, Amazon branded products are still a tiny % of the overall market and often are the generic to other sellers more specialized products. Plus, no one has proven that Amazon is favoring their own.

The Apple AppStore is powerful and there have been abuses (takedowns, not allowing products to compete with Apple’s) but, worldwide, Apple is still a small % of overall app sales.

Facebook is more compelling. However, there are alternative social media (its mostly but not exclusively winner take all). And there is likely a simpler fix to right this by unwinding the Instagram and Whatsapp acquisitions. Imagine Facebook proper (and Messenger) having to compete there?

Google owns, controls and can shift whole businesses and industries with an algorithm change (see what the Medic algorithm update did to the natural health and wellness space). Even if you buy their “don’t be evil” mantra, they simply have too much power and depend far too much on algorithms to make decisions. Talk to any advertiser who’s received an ad takedown from one of Google’s algos. You contact support who can’t tell you why your ad was taken down and they spend their time trying to help you trick the algo to get your ad back up. There’s a problem here.

Amazon, Apple, Facebook and Google have been the envy of corporate America, admired for their size, influence and remarkable growth.
Now that success is attracting a different kind of spotlight. In Washington, Brussels and beyond, regulators and lawmakers are investigating whether the four technology companies have used their size and wealth to quash competition and expand their dominance.
The four firms are lumped together so often that they have become known as Big Tech. Their business models differ, as do the antitrust arguments against them. But those grievances have one thing in common: fear that too much power is in the hands of too few companies.
The attorney general of New York, Letitia James, said Friday that the attorneys general in eight states — she and three other Democrats, plus four Republicans — and the District of Columbia had begun an antitrust investigation of Facebook.
https://www.nytimes.com/2019/09/08/technology/antitrust-amazon-apple-facebook-google.html

Commentary: ROI is the only KPI

August 18, 2019August 18, 2019
by Andy

Key Performance Indicators (KPIs) are one of the hottest topics in product management (well, management in general in a well run company). However, many struggle in defining meaningful ones. I assert there’s only one that matters: return on investment (ROI). Using this will help simplify and clarify all your activities and planning. I have found few things that can’t be defined in terms of ROI.

In my work across a variety of companies large and small I typically see one of these situations:

KPIs inherited from what finance/sales (or whomever the revenue generating owner is) says measured as pure P&L and not often reconciled with the total investment (and often missing secondary KPI product and user success measures)
A massive number of intricate measures that no one can get their head around nor really knows how or why each matters for the business/product
Not much at all

In each case, there needs to be a KPI reset. I encourage most teams I work with to have primary and secondary KPIs. Primary are what really defines and measures success. Secondary are more intricate that focus on more detailed aspects of a product or business. I have arrived at the point — after testing this across many different types of products, services and organizations — that ROI (broadly applied) is the one all products should set as their primary KPI.

In my teaching, we discuss these KPI categories:

Acquisition: how do customers discover the product
Conversion: how to we get them to pay
Engagement: how regularly and for how long your product is used
Retention: when time to return/renew, do they
Satisfaction: how do customers rate the product
Defect rates: what is the rate of defect/failure in the product
Cost savings: how can this product make the business more efficient

There are a few KPI models but most overlap with these groups. These are excellent secondary KPIs and should be measured in every product. In addition, we always want to measure our KPIs in terms of our customer needs, not just our business interests.

Amazon provides one of the best example here: typically, warehouse logistics would measure success as the time from order received until order shipped. Amazon measures from order received until customer receipt of the package. If they are quick in the warehouse but the shipper is slow their customer still is not satisfied. This is also why you see Amazon taking over the last mile of deliveries more and more.

All traditional KPIs can be translated into ROI. Let’s look at some examples.

Example: Social network engagement

Social networks are all about engagement, right? Well, yes, that’s a key secondary measure: typically daily active users (DAU) over monthly active users (MAU). However, every user and every user interaction has a value. Facebook makes money on ad display and interaction. They know the values of those and a bit of math can peg a value on each user engagement. They then need to know both what are the upfront and ongoing R&D for the service plus other operational costs providing the ROI denominator. That might seem simple and obvious but there’s a knock-on effect: by tracking ROI, they can actually see variations that engagement alone does not show. What if ad rates fluctuate? What about bandwidth or R&D costs? What if we now need to hire 20,000 people to moderate our content? Engagement alone is too simple.

Example: Subscription service retention and LTV

Its less expensive to keep a customer than acquire a new one goes the saying (and the data most often supports this). This one should be pretty straightforward as any good acquisition engine inherently uses ROI to evaluate their marketing spend efficacy. But what about retention? How much are you willing to spend (often in discounts or other incentives) to keep a subscriber? Companies typically know the lifetime value (LTV) of their customers (especially if you’re in the subscription game). Its the average purchase * typical retention period purchases (e.g. $30/mth over 15 months or $450 LTV).

However, are you tracking lifetime net profit (LNP) of your customers? LNP takes LTV and considers margin. Continuing the example above, if we consider that we have a 33% margin on those subscriptions, our LNP is actually $148.50.

Lifetime Net Profit per User
I. Avg Purchase	$30
J. Average Purchases per month	1
K. Typical retention (months)	15
L. LTV (IxJxK)	$450
M. Average Margin	33%
N. LNP of a user (LxM)	$148.50

When the marketing dept tells customer service that they can spend $200 to incentivize a customer to stay, that might actually be a negative ROI.

Example: Real customer acquisition cost

Any good marketer (esp digital) will track the ROI of all their placements across media buys. Customer Acquisition Cost (CAC) is typically your marketing spend (for a given channel/placement) over your number of conversions. But just looking at CAC might not show the full picture. Consider the example below:

Real Customer Acquisition Cost
A. Marketing Spend	$10,000
B. Clicks	2,000
C. CPC (A/B)	$5.00
D. Conversion Rate	2.5%
E. Conversions (BxD)	50
F. CAC (A/E)	$200
G. Discount on 1st purchase	$50
H. Real CAC (F+G)	$250

The conventional CAC does not capture the fact that there’s also an incentive discount of $50 making the real CAC $250.

If you want to take it further, the LTV/CAC ratio is often measured as key to tracking unit profitability. Combining our two examples above, this would be 2.5x ($450/200) under the conventional model. But considering LNP/rCAC, that now is a negative 0.59x ($148.50/250) return. It looks like you have a nice ROI on your marketing spend for your service and a good retention plan, but maybe not.

Thus, using ROI (or its proxy measures) is going to give you a clearer picture of:

Is what I’m doing operationally with my product profitable?
Is my acquisition funnel profitable?
Should I build this?

I now always ask this question and drive to find these ratios when I engage companies on product strategy.