Wednesday, 20 of March of 2019

Why LinkedIn is Losing Value for Content Creators

LinkedIn used to be a great tool to attract an audience for my articles and talks. Two or three years ago, an investment of less than an hour on LinkedIn would bring me a substantial number of views, and the viewers were interested in my field. Those days are over. Today, that effort brings perhaps 10% of the viewers it once did, and I expect it will only get worse from here.

First – how do I know this? Being a data analyst, I have made an effort to track views over the past 4 years. I use tracking links and observe the response to my posts. My approach isn’t the best controlled or most sophisticated available, but it tells me enough to let me know which posts attract readers, what channels draw viewers, and what times are best for posting.

It might not surprise you to know that overall response to posts has been declining. After all, the number of posts and emails, even the number of channels is increasing all the time, and human beings have only so much bandwidth. But LinkedIn has made a number of changes that make the problem even worse for the everyday content creator, and many LinkedIn users are unaware of them.

For example, have you noticed that LinkedIn routinely reduces the frequency of emails from your groups – even though you opted in for those emails and did not request any changes? So, those who don’t open group emails frequently receive fewer and fewer messages, and may even stop getting messages at all from some groups. No message means no chance of a message getting read, and no chance of a reader taking notice of your post.

Another thing – are you sure that your posts are really getting posted? LinkedIn automatically diverts some posts and comments for moderation, even when moderators don’t request this. Since many groups don’t have active moderators, those posts are never seen. Recently, I looked back at the groups where I post, and found that, in some cases, as many as four or five posts made over the course of several months were still “waiting for moderation”. (I suspect some others may simply have timed-out and disappeared, though I don’t have sufficient records to verify that.)

It appears that LinkedIn flags some people, in some groups, for moderation. This may be hard to detect, as the same user may be affected in some groups, but not others. Or the new posts may be diverted, but not comments. If you can spot the pattern, you’re a better woman than I. I’ve found that this is happening to many people. So, a lot of posts are just going down a black hole.

But the posts are just diverted for moderation, so they should be moderated and appear in the group in a little while, right? Best of luck. I tried writing to moderators about this. Many never responded, some responded but weren’t sure how to address the problem, and one told me point blank that he had no intention of actually moderating his own group.

And then, it got worse.

Once upon a time, any entry in the LinkedIn status box would bring me a couple of interactions with other human beings. I’d get a note, a call, or perhaps a comment from someone I spoke to for another reason. And I knew that, for every person who reached out to me, perhaps ten more saw the post. Then, LinkedIn integrated status updates with twitter. Ugh. The feed was soon jammed with oodles and oodles of twitter posts. Even now that they have eliminated that particular twist, I see lots and lots of links to mainstream media content, with no meaningful comment from the person who posted. That’s mixed with notes about every group my colleagues join or company they follow. The small fraction of updates that are actually about what people are doing are lost. I may get a few “Likes”, but I don’t get real interaction.

A few months ago, LinkedIn decided we needed “Thought Leaders”. Only certain people can be thought leaders. They started with already well-publicized figures like Richard Branson. After a while, LinkedIn opened a path for applications to become a thought leader, but quickly closed that channel. So now, the top of the page is crammed with mainstream news updates, Thought Leader posts, and crap. The thought leaders I see in my feed are all men. Out of curiosity, I alphabetized the thought leader list and checked out the first 100 people. About 90% men. Mostly white and Asian men. I think one black guy managed to make it in there, and good for him.

Bottom line – if you’re a content creator, and you’re not, say, Richard Branson or Jack Welch, LinkedIn has become a very poor vehicle for cultivating an audience. Time to develop other channels. I, for one, have decided to stop posting about new articles in LinkedIn groups, and to cut back my time spent on LinkedIn. It’s just not producing for me any more.


The FTC will change the way you do business

New piece on All Analytics:

FTC & the Data Brokers: Why You Should Care

Comments Off on The FTC will change the way you do business

Risks of product loyalty

3 Tips for Sustaining Your Analytical Software, an All Analytics post by Bryan Beverly, left me with some doubts. Speaking as one who has interacted with many highly product-partisan software users, I must suggest a few cautions here.

Referring frequently to a product name, rather than using common technical or plain-English terms, has consequences. For one thing, it is often difficult for people who are not experts or familiar with your subject to understand what you are talking about. Use of such jargon also contributes to the analyst’s image as a geek whose concerns are not important to the bottom line.

Product-loyal analysts take other risks as well. They may become so attached to their tools that they are unaware of, or unwilling to use, easier or more cost-effective alternatives. They may appear inflexible to current employers, untrained or untrainable to prospective employers. They may have difficulty collaborating with others who use different products.

I have worked with organizations burdened with managing multiple products and multiple versions, creating serious inefficiencies in purchasing, training and support. Product loyalty implies much more than just believing a certain tool is the best one to do your job. It’s also a roadblock to interacting with others who have differing needs or preferences. Snobbery tied to product preferences is a common thing. It is the profession’s class warfare.

I love the ideas of creating in-house user groups, letting management know about the value of your tools to the organization, and doing all we can to learn and help others learn about good tools. In the end, though, a tool is just a tool.

Noreen Seebacher asked: “What if an analyst has a very specific reason for favoring one piece of software over another? Is it fair for him to promote that product or does it create risks for the organization in terms of cost controls, etc?”

Analysts always have very specific reasons for preferring one product over another!

It’s important to separate the reason – what we need to accomplish – from the product. Often, analysts become attached to specific products, believing them to be better than all others for performing specific tasks. When that happens, and it happens often, there are many possibilities to consider.

    The product may not actually have been superior in the first place
    A real product advantage that existed at one time may not persist as other products change and improve
    A new product may have better capabilities, or eliminate the need for certain tasks or methods
    The benefits of standardizing methods or tools across an organization may outweigh the specific benefits that drive the preference for a specific product
    Challenges of using some products, or integrating them into the business, which affect many stakeholders, may outweigh advantages to a subset of the organization
    The costs (think total cost of ownership as well as pricetag) may not be justified by the reasons behind the analyst’s preference
    and so on…

If you have a good reason for preferring a specific product, you must state it in terms of the business. What do you need to do, why is it necessary to the organization, how does it translate into dollars or some other meaningful business metric? If you can only state your preferance in statistical jargon, you won’t be persuasive.

Let me give an example of an organization that was deeply affected by this issue. A state government agency was planning an organization-wide operating system update. The staff used several different products, and often several versions of each, running on several operating systems. Through many lengthy and detailed discussions, we found that all the analyses that they needed to perform could be addressed with just one product family.

Standardization offered an advantage in purchase cost, but it also addressed many other issues. Technical support would be simplified. Training weaknesses could be addressed in a manageable way. Users could understand and help each other more easily with a common platform. Sharing of work would become practical. And the new tools offered valuable capabilties which had not been available to the organization before.

Some of the staff was quite open about their dislike of the change. They were certainly faced with some legitimate challenges – they would have to learn to use a new product, and they may have had code written for their old tools which would now have to be replaced. But in my discussions with them, those concerns never came up – instead they grumbled – publicly -that the old stuff could do everything the new stuff could do. That simply wasn’t true, and since they constantly made such claims during public presentations, I was forced to contradict them in front of their coworkers. It made them look stubborn and foolish.

This wasn’t an isolated case. Very similar things go on with every organization that explores standardizing tools. The people most resistant to change are usually the analysts with the most sophisticated statistical training – that is, the same people who should be the best equipped either to make a good business case for their preferred tools or learn to use the alternatives, and learn well. When they choose to do neither, they look like highly educated babies.

Comments Off on Risks of product loyalty

Is There Madness in Your Methods?

New article on All Analytics: Is There Madness in Your Methods?

Comments Off on Is There Madness in Your Methods?

Selecting Big Data sources for predictive analytics

New article on Smart Data Collective:
Selecting Big Data Sources for Predictive Analytics

Comments Off on Selecting Big Data sources for predictive analytics

Lauren Bacon is really on the money

Once And For All: Tech is Not a Meritocracy, a recent post by Lauren Bacon, is really on the money. A lot goes into the mix that leads to success in industry, and the loose collection of things we call “merit” are only a bit of that mix.

It’s hard to imagine that anyone would believe otherwise. Consider a few of the people you know who are in positions of power and influence. Are they all fireballs of talent? Does each and every one seem to be the best person for the job? Don’t you also know many people whose jobs seem unworthy of their merits?

Lauren’s post focuses on biases – unrealistic views of ourselves, as well as others, that stand in the way of progress for women and others. She points out that we must recognize the existence of unconscious bias as well as overt discrimination.

Ian Muir responded with a question – what can allies be doing to help? Lauren says she intends to write about that, and I’m looking forward to reading what she has to say! In the meanwhile, let me make a few suggestions of my own.

Every day, each of us has the opportunity to do a little something to help others advance in their professions. If you simply make a point of doing one thing each day for someone in a group that is underrepresented in tech, it will add up to a lot over time. Don’t think you have to be a big shot to make a difference. Every bit helps, and little each day helps a whole lot.

Let me start with an example. Do you know of a conference that is coming up, one that is a good showcase for the speakers and a worthwhile networking venue? Do you also know someone – perhaps a woman or an African American who is not [yet] well known- who would be a good presenter? What can you do to help get that person on the platform? Don’t just say, “Hey, so and so, you ought to apply.” That’s not what the good ‘ol boy network does, and we have a lot to learn from the good ‘ol boys. Instead, call the conference organizers, and plug that person. Do what you can to get them motivated to call your colleague and invite her or him to present.

Taking the conference theme a little farther, next time you are thinking about a conference – look over the list of speakers and take note of the level of diversity. If it isn’t what it should be (for example, 25% of the people in tech are women, so 10% female speakers should set off your bullshit* detector. Try suggesting some qualified speakers. If that doesn’t work, make noise about the problem in public – write a blog post, tweet, put the word on the street. Use whatever channels you like, but call them on it, and do it in public.

The next day, do another thing.

Need ideas? Here are a few. I hope you’ll respond not only by doing some of these, but also by posting some suggestions of your own.

More things you can do to help others advance [I’m going to say women here, but of course, this isn’t just for women]:

    Encourage your local professional group to showcase women speakers, especially new speakers, and to place women into chapter office and key committee positions

    When somebody’s hiring, suggest a women for the job

    Ask others to inform you about talented women they know, or know about

    Ask women what kind of help they need. If you can’t provide that help yourself, look into your network – perhaps you have a contact who can.

    Prepare yourself to call others out when you hear false statements. This might mean emotional preparation, getting ready to disagree with someone in public, or research, such as learning a few relevant statistics, or other facts, that will help to make your point.

    Read a good article written by a woman? Share it with others.

    Maybe you know someone who has every advantage in the world already. He still needs help. Do you know a woman who can help him? Make the introduction! People appreciate those who help them, and often return the favor.

    Don’t let anybody convince you that it isn’t your business to help. Yes, there are people so small and so stupid that they will try to do that. If you encounter such nonsense, just shake the sand out of your shoes and move on.

*[with apologies to]

[bool-shit] noun, verb, interjection Slang: Vulgar.
unconscious bias.
verb (used with object)
conscious bias
verb (used with object)
3.something rotten

1910–15; bull1 (perhaps reinforced by bull3 ) + shit

Related forms
bull•shit•ter, noun


Powering Predictive Analytics with Big Data

The recording of my All Analytics seminar, “Powering Predictive Analytics with Big Data” is now available ( You must register (free) to hear the recording. Listener questions and my answers are also posted.

[The following little note is just for the benefit of Technorati: M7X95QSKFVE6]

Comments Off on Powering Predictive Analytics with Big Data

Bluffer’s Guide to NoSQL Databases

Bluffer’s Guide to NoSQL Databases

New piece on All Analytics today.

Comments Off on Bluffer’s Guide to NoSQL Databases

O’Reilly Strata: Where are the broads?

Lutz Finger posted a word cloud of job titles for the upcoming Strata conference, and asked, “Where are all the needed Data Scientists?” A reasonable question, although it has never been my impression that Strata sought to attract analytic talent. It’s a commercial event, designed to attract the corporate buying power that their sponsors desire. My question is the same as last year’s: Where are the women?

Last summer, in my post “O’Reilly Strata: Deluded About Diversity?”, I pointed out that representation of women in the Strata audience and on the speaking platform is pathetically low, despite these two important facts:

1) There are as many women in analytics professions as men. (Doubt that? Read this: “The STEM Profession that Women Dominate”)
2) O’Reilly claims to actively pursue diversity among speakers and posts a lengthy and blusterous diversity policy.

Last year I found that only 12% of the speaking slots at Strata went to women. That proportion actually dropped from 15% in 2011.

Think there were no qualified women to speak? Remember, there are as many women in analytics as men. Think no women applied? Nope, that’s not it either, since I personally know several qualified women who did so, and not one of those talks was accepted. By the way, Strata doesn’t have to wait for applications. They could just pick up the phone and invite some women to speak. The grapevine informs me that they have been known to actively invite speakers. Male speakers, anyway.

So how did they do this year? I just tallied it up: women make up 12.6% of the slots at next week’s Strata conference. What progress.


Do data scientists really need IT? Yes!

My post, Kiss & Make Up With IT (, has kicked up a lot of discussion.

One reader asks “Do data scientists really need IT?” He thinks not. He’s dead wrong. You can read the article and discussion on All Analytics. Here is a bit about why it is important for data analysts to work constructively with IT:

Let’s consider what data analysts in corporations are expected to do. It is their responsibility to provide corporate management with information that supports decision making. These executives are empowered, and obligated, to manage the business in the interest of shareholders, and within the law.

The most relevant data to any business is internal data. This data is the private property of the business, and not available through any other source. The responsibility for maintaining and controlling such data falls to the IT organization within the business. This is no trivial matter. There are significant legal and financial concerns tied to the handling of data.

If the data is properly managed, there will be no route to obtain it other than through the proper channels, and those channels are controlled by the IT organization. This is not merely the way data is managed in the business world. Nonprofits and government agencies use similar processes.

I have encountered many data analysts over the years who resorted to any means they could devise to avoid dealing with IT. They often get away with this, because their results are viewed only by their own sympathetic team members and by executives who do not question them on matters of data management. That is, they get away with it until the work comes under serious scrutiny. That scrutiny commonly happens when the business becomes involved in a lawsuit or other legal action, or when a new manager enters the business asking questions and taking no prisoners.

Early in my career, I discovered an issue with some of the data that I and my team depended on. My manager was unwilling to take action to correct the problem. There was a loud public shouting match between us over that. But the problem got fixed in the end. Why? Because an auditor came in to examine our records, spotted the issue and brought it to the attention of upper management.

Comments Off on Do data scientists really need IT? Yes!