Another day, another set of data... I've been investigating what I call "peripheral" data sets in order to get a different perspective on how previously unseen or unmeasured activity affects the overall operation of our company. In my last post, I looked at how our busyness could be represented by the volume of communication over our internal project management system from one month to the next. In looking at that picture, I realized that the volume of activity is much more drastically affected by maintenance work for our clients than by new projects. I classify "maintenance" as any work done for an existing client- it's a pretty broad spectrum, but since our new project process is so regimented, the split in categories is pretty realistic as far as our company's day to day experience is concerned. When I noticed that October of 2008 had the highest volume of communication, I wondered what our maintenance sales were that month and how they related to new project sales. Sales data is the easiest information for me to dig up, but I wasn't interested in the particular sales totals as much as the relationship between the numbers.

This brings me to the graph you see above. As I said, I wasn't so much interested in how much we sold from one month to the next as I was about the breakdown of sales- how much of it was new business and how much of it was maintenance. So, I determined the percentage of each month's sales total for the past few years that came from new projects and maintenance. For example, the graph above shows that in October, 2008, 37% of the month's sales total came from new project sales while 63% came from maintenance. No wonder we had so many posts to our project management system that month! As you can tell from glancing at the graph, this is a relatively infrequent occurrence; more often than not, the new project sales account for the majority of the total. When I first plotted the data, I didn't add the percentage values because I was more interested in the general relationship, as well as any trends that might be perceivable from visualizing the data. Again, glancing at the graph seems sufficient to conclude that there are no obvious patterns, nor an obvious trend in any direction (i.e. maintenance percentages trending upward or downward).

Averages and Average Averages
Then I wondered about averages. The data set covers three years, but it isn't three full years. Additionally, the current year has a couple of extreme cases (January, in which maintenance accounted for only 19% of the sales total, and September, in which new projects accounted for only 2%), so I decided to look only at 2008's average.

In the chart above, I plotted out a spectrum displaying only the percentages of total sales accounted for by maintenance sales-- the lowest, 24%, came in July of 2008, while the highest, 71%, came in August. Two concurrent months bookending the spectrum seems to clearly show that there isn't a seasonal correlation. But back to averages, the average maintenance sales percentage for 2008 was 41%. What's interesting about this is that 8 months out of 12 were less than or equal to the average, leaving only 4 months in 2008 that exceeded it. If I isolate 2007, the average maintenance percentage for the 7 months plotted is 42%. If I isolate 2010, the average maintenance percentage for the 10 months plotted so far is 38%. These numbers are pretty close together. In fact, only 11 months out of the plotted 29 had maintenance percentages that exceeded 41%, which is a "score" of 40%. Maybe there is some significance to 40%...?

Ultimately, I'd love to see the percentage of maintenance account for more consistently higher amount. I think doing more work for fewer clients is to our and our clients' advantage- it fits in with my motto of what we want to do: Serve fewer clients at a higher level. I believe we'll get there.

One last thing: The graph above doesn't show the number of new projects sold on a month to month basis. In 2007, the average was 4.1. In 2008, the average was 4. This year, the average so far is 2.6. To me, that's the kind of decrease I want to see. It means that we're selling fewer projects on a monthly basis this year, but at higher costs each (fewer at a higher level). So, all in all, one more piece of the puzzle...

In my last post in what is becoming series on measurement, I started off with my hypothesis that our company is like an ecosystem, "comprised of many areas of unseen activity" in addition to the sort of seen activity you'd expect (sales, individual projects, relationships, etc.). So, in trying to verify my hypothesis, I've been gathering data representing all kinds of unseen and unmeasured activity to see how it relates to the big picture as I've understood it so far. I started with looking at our blogging activity over the past three years and noticed that the months where we posted less loosely corresponded to what we tend to think of anecdotally as "busy" times for our company. That made me wonder- how else could I measure "busyness"? Looking at sales data wouldn't quite do it, because those numbers would correspond to the beginning of a project, so the trendline of sales may not match up exactly with that of volume of work over time. However, looking at the volume of communication using our internal project management system might help me discern at trendline for "busyness."

Unfortunately, there was no simple way to do this. Our system doesn't have reporting tools that would tell me how many logs were completed for a particular period of time. However, since I receive an alert every time a log goes through our system, I realized that I could isolate those messages in my Gmail account and... count them one by one. Actually, I ended up counting them 50 by 50, since that's the page limit in Gmail's search results (see image at left). Needless to say, it took a while, but I did uncover some interesting things.

First, let me describe the metrics shown in the graph above. The vertical axis represents the number of individual logs posted by Newfangled employees to our project management system (these might be messages checking in on production progress, updating the task description, asking or answering questions, posting files, etc.). As you might imagine, a project of even minimal complexity would have many such messages, particularly as we have people collaborating on projects who work in separate offices. More obviously, the horizontal axis represents months over the past two years.

The first thing I noticed was the dramatic increase between August (1191 logs) and October (3132) of 2008. To put the number in perspective, October's total averages to about 136 logs posted per day (there were 23 business days that month)! That's in addition to all the phone and in-person conversations that occur here each day. In fact, we haven't had a month with that level of communication volume since. I cross referenced this number with our sales from October of 2008, and it turns out that, in addition to the two new projects we signed that month, we also did 57 different new functionality upgrades to existing client sites. That was the highest number of upgrades in one month for the entire year. With that in mind, 3132 project logs makes much more sense. We were busy.

Here are some Newfanglders. They may not look like it, but some of them were probably communicating using our project management system the moment this picture was taken.

However, I then thought to cross reference the lowest month shown, June of 2008, with the sales from that month to see if they were correspondingly lower. While the number was less (3 new projects signed, 43 new functional upgrades), the overall communication volume wasn't proportionate. It should have been a higher number if there was a direct correlation. Especially since the previous month, May of 2008, we'd signed 7 new projects and done 55 functional upgrades. There would certainly be some bleed from May to June in terms of project communication. While October, 2008 is the peak, the numbers tend to level off in the mid-2000's after that, though there is another peak in October, 2009. I don't think sales are the complete explanation for this, but I do have a couple of ideas about other factors that could.

The first factor is the number of employees using the system. Between May and October of 2008, we hired 4 new employees- three Project Manager Assistants and one Resourcer. These roles were essential to a new system we'd been establishing to make sure our service remained excellent while the complexity of our work increased. The Project Management teams use this system constantly to communicate and log project progress, and the Resourcer is constantly checking in on every task to watch progress and utilization. No wonder the number jumped so drastically! The other factor helps to explain the pattern, in that we are again seeing a jump this October despite not having a corresponding growth in personnel- that factor is the pre-holidays rush. This happens every year, where existing clients and new prospects are eager to get work completed or scheduled before the holidays and especially the new year. There's something about these calendar landmarks that put the pressure on, not to mention the common need to allocate funds before the end of the financial year.

This was an interesting exercise. Little by little, measuring these "peripheral" data sets is giving me a much better sense of the big picture.

A Recent Example

On a current project for SteriPEN that is now in the whitescreen phase, the client's marketing team and I discussed how to display external links to online retailers that may carry different combinations of SteriPEN products. Before two internal review sessions, on the Product pages, links from sidebar widgets went to a Buy Online page where users could filter the online retailers by the category and product.

Here is a Product page at that stage of the prototype, with Buy Online in the sidebar...

...linking to this page, where the correct retailers were to be pre-filtered.

It took a fresh reviewer to suggest pulling the filtered retailer links right into the sidebar of each product page.

In a third review, another set of fresh eyes suggested that the retailers maybe should not be placed in the sidebar, since the sidebars could be reserved for more universal callouts rather than the page-specific function of displaying retailers for this individual product. So we placed the retailer links across the bottom of the page.

It's no news that objective critique can aid any creative process, but I am still amazed by how little time it takes a short review session to produce significant ideas for improvement.

I have a hypothesis that our company is much like an ecosystem--comprised of many areas of unseen activity in addition to the very visible activity. With that in mind, I've been collecting data for the past few weeks that is a bit different from what I might normally look at (i.e. website and financial data). I want to see what unknown connections there might be between what we do intentionally and what we do unintentionally.

The graph above, containing one set of data from my "peripheral data collection" of late, shows the number of blog posts we've published since starting the Newfangled blog back in October, 2006. Throughout these three years, we've never had any established quota for publishing frequency, so I wondered what conclusions I might be able to make from looking at post frequency from the beginning until now. One immediate conclusion I can make is that this is not a large enough sample of data to support identifying significant cycles. There are only two full years represented, and the truth of the matter is that our blogging was fairly inconsistent during those three years for pretty discernable reasons. The first is due to population. From 2007 through 2009, we added 9 new employees to our team, all of whom have contributed to the blog. We also lost a few who blogged from time to time. The second is due to a sense of purpose. When we first started, many of our post were culturally oriented, "look-what-I-found" kinds of posts. It wasn't really until July of 2008, when I published a post called Newfangled Blogging 2.0 that we really began to focus our efforts. In fact, July of 2008 was a time when we were focusing on defining a web content strategy in earnest, blogging being just one piece. After that, we started to plan our writing- identifying topics we wanted to see covered in the blog, making the frequency more consistent and encouraging more people to write. We've been moderately successful in that; our production schedule does make it a challenge to do all that we want to do. But, we're getting the hang of it.

Note, for example, the May-June-July pattern that shows up in 2008 and repeats in 2009. That's an interesting trend. It used to be that summers were a slower time at Newfangled. But since 2007, summers have been the opposite. They've been very, very busy, so I'm not surprised to see the decrease in blogging at the outset of that season. I also wonder if we'll uncover a similar pattern in October-November-December. Time will tell.

"Optimizing" is an interesting word. It is important to ask the question, "For what, or whom, are we optimizing?" Is the ultimate goal of your site to be Search Engine Optimized, or User Optimized? Oddly enough, most people get stuck on the SEO part of this but don't stop to think about what happens if the SEO actually works and brings great prospects to their site. Your site's success isn't measured in traffic; it is measured in conversions.

We believe that focusing on the actual user experience is the best approach to optimization. If users can get to your site easily, intuitively find the content that they are most interested in, and easily read it, then Google will be able to do the same. The reason that Google has become synonymous with the word search itself is that Google is the best at mimicking human behavior. Since Google reads through web pages and sites like we do, we people consider it to be the best search engine. Makes sense, right? 

We have seen time and time again that Google bait pages aren't great at bringing people in and keeping them. You may get a #1 ranking on Google for a certain phrase, but chances are that few people will actually search for that phrase and most of those that do click through to your Google bait page won't stick around for long once they get to your site. Why? Simple. It is because the page wasn't written for them, it was written for Google. Once they get to the page it just doesn't really seem all that interesting to them, so they bounce.

Plan your site and write your content for your prospects, make it an educational resource they can't live without, and I assure you that Google will respond in kind.

"What is this worth?" "Is this working?" "What should we change?"

Ok, so you should be pretty psyched about data. If you're not yet, I've done you a favor and gathered together nine posts from the past year that are all about data. Don't worry, they're not terribly technical, and each includes a helpful visualization that is created to help communicate the key point concluded from the data collected. Go nuts!

The list below is ordered from newest to oldest.

3 Years of Sales Data	2 Years of Communication Data	3 Years of Blogging Data
6 Months of Referral Data	6 Months of Campaign Data	Top 3 Analytics Metrics
Impact of Specific Referrals	2 Years of Newsletter Data	1 Week's Email Data
1 Year of Campaign Tracking Data	1 Month of CTA Data	1 Day of Timesheet Data

Less frequently but more consistently, I edit the Newfangled newsletter each month before it goes to press. Several factors make this process different than a one-off sentence or paragraph correction:
-I'm working with an entire document rather than just an isolated section. Seeing the big picture gives me the chance to make general structural suggestions or get a better sense of the overall tone.
-I get the draft a little in advance of the publish date, which makes it possible to address those kinds of upper-level issues when needed. There's time to sit down and talk through any points that are unclear.
-I'm working with the same writer every month, which has allowed me to develop a sense of what kinds of feedback will be useful. For instance, the introduction is often one of the last things written (which is a very common writing strategy and is generally the approach I would recommend), so it's usually a key topic of discussion.

Because my background is in teaching rather than commercial editing, the editing style I've learned focuses heavily on getting people to develop and express their own ideas. One of my tendencies as an editor is therefore to want the writer to do the rewriting, rather than swooping in with a red pen and doing it all myself. The earlier in the revision process, the stronger that tendency is. If I find an idea unclear, I'll usually ask questions about it that lead to a discussion about ways the writer can clarify; if the wording is unclear in a certain spot, I'll write out an alternate wording or two but leave it up to the writer to choose which to use; if I see a misspelling or punctuation error, I'll usually just take care of it. Again, how appropriate this hands-off style is depends on the situation. I like that it allows authors to retain a sense of intellectual ownership over their writing and to know the final product still feels like their voice. (Ending up with a document that doesn't sound like me is one of the things I want to avoid when getting my own writing edited.) On the other hand, it would probably be good for me to develop more comfort with just jumping in and taking care of everything when the situation calls for it.

What about you--do you collaborate on writing, either as an author or an editor? If you seek out editing for your writing, what kinds of responses do you find helpful? If you tend to be the one doing the editing, how do you approach the situation?

Our building was actually the first public building in Carrboro. It was built because the train tracks happened to end at this location. They were supposed to continue on a half mile into Chapel Hill, but they ran out of funding so they stopped where they happened to leave off. The original use for the rail line was as a drop-off point for the local farmers to bring their cotton and tobacco. The man that happened to own the few hundred acre farm that abutted the tracks decided to build a building (our glorious office) to hold the offloaded wares overnight. Before long, he decided to outfit the building with the area's first steam engine. This was first used to power a cotton gin and later a grist mill--both of which resided inside the building. At some point between then and now more funding was acquired and today the rail continues on that half mile to a plant that powers UNC. The line is mostly used for coal transport to that plant.

Until I get seduced by the perfect reading device, I'm pretty convinced that there is untapped power and potential in print-on-demand (for example, check out the Espresso Book Machine. Russell Davies, the person who, believe it or not, first exposed me to the idea of print-on-demand, says it perfectly:

My favourite example is this: Things I Word Rather Read On Paper. Is it combines what the web does well; publishing, gathering, discovering and curating content (via instapaper) with what print does well; being readable, durable and portable.

As I read that, I realized I should probably put together a post that outlines my process of gathering, discovering and curating content- it's a fairly complicated one when you take into account all the various channels for finding, experiencing and sharing information. But, it is one way of being a human synthesizer- a necessary discipline for people in our industry. For now, though, I wanted to again share the printed side of it- particularly because I'm in awe of how good of a job Lulu.com does. I submitted my files and had my book within several days; the quality of the book itself is very, very good.

The image below shows an interior spread of my "A Year of Ideas" book, which includes an image by Lauris Paulus and an article titled The Street as Platform by Dan Hill. (Lauris and Dan, don't worry, I'm not selling this book. It's just for me.)

- - -

Here are the other two books I've created using Lulu.com. The book on the left was the first one I created last March. The book on the right is the second one I created shortly after in April:

"It's 5:30 a.m. on a Saturday, and a white delivery truck for the New Jersey Record has just pulled into the parking lot outside the Plaza Diner in Fort Lee, New Jersey. The driver -- Mike, who asked that his last name not be used -- is at one of his 130 stops on an eight-hour shift that began at midnight. Mike's job, which takes less than a couple of minutes per stop, entails filling the coin-operated machine with the day's papers, collecting unsold copies and emptying the machine of its coins. Even though Mike has a full schedule and lots of stops, it doesn't equate to pushing as many papers as he once did. Mike loads 15 copies of the Record into one machine -- and that's a good load, he says. Other locations receive only five to seven copies. He's also tasked with filling machines for USA Today. Though he's been on this job for only two years, Mike has been on the route long enough to know business is down. He says newspapers sell better at train stations than from the street machines he services."

The Solution?
The image below represents the solution to the inefficiencies of the vending box model. TimesReader is an application that pulls New York Times content to your computer every day in a more "traditional" reading format than the New York Times website. It also archives up to a week of back "papers" and allows all kinds of unique navigation options. In addition to more content, subscribers get an ad-free interface. For now,though, the website, which provides all New York Times content unrestricted will be the "solution" for people like me. But at some point, The New York Times, and other leading newspapers, will figure out a paid model and the TimesReader will be there to pick it up- particularly if additional devices (such as the rumored Apple tablet) strike deals with newspapers-- see the video with Bill Keller at the end of this post for more on that subject.

The Real Problem
The real problem isn't really the method of delivery. See, the vending box method may seem like an inefficient vestige of the past (which it is, in some respects), but we shouldn't necessarily consider the online solutions as preferable on the basis of perceived simplicity. I think that the contemporary delivery methods are probably dependent upon significantly more complex systems. With the print model, there is one "template" and several delivery methods (homes, businesses, boxes, and particular vendor outlets). With the online model, there are multiple templates (website, emails, unique content channels, advertising, mobile applications, etc.) and the massive conundrum of delivery (from once daily to constant delivery to a multiplicity of formats).

The real problem is regaining a perception of value, which, in part, requires an accurate perception of complexity. I'm not sure how likely that is- it's going to depend upon the unique perspectives of individuals everywhere. (For example, consider the perspective of the vending box delivery driver once that system is phased out.) But that aside, the current climate of content consumption is driven by our expectation that it's all free.

The graph shown to the left (courtesy of The Awl) highlights an overall diminishing circulation trend among the major U.S. papers (the Boston Globe sank below the 400K mark, which is why it does not appear on the chart). However, the Wall Street Journal appears to be enjoying a healthier reality than the others. Because of the overall decline in circulation, though, advertising is waning and news offices are laying off many journalists (and plenty of other positions integral to the production process). None of this is going to be solved (even by a micropayment scheme) until we recognize the value of the systems required in order to produce this content and act upon that recognition.

The Real Solution
So, I don't exactly know what the real solution is, but I do know it has everything to do with perception of value. With newspapers, the value hierarchy begins with the inherent value of the information itself, followed by the value of the thinking behind it-- the writers, editors, producers, etc., followed by the value of the organizational systems, followed by the delivery systems, followed by the value of the corporeal product itself. It's clear what's expendable. I also know that many other industries besides journalism are facing this very same issue right now, including our own. For us, and companies like ours, the value hierarchy is very similar to that of a newspaper. Much of our costs have more to do with planning than implementation, and despite our knowledge that the planning is the foundation of any product, getting the customer to recognize the value of it a challenge. That's why we've spent the majority of the past year highlighting the value and necessity of planning, slowly changing the culture of our small corner of the web. I hope that a similar progression occurs in other industries as well.

In the video below, Bill Keller speaks to the digital group at the New York Times, and speaks to many of the topics I've mentioned above-- potential pricing models, new delivery methods, and new technologies to fulfill them...

- - -

Bill Keller speaks to the digital group at The New York Times from Nieman Journalism Lab on Vimeo.