What are these "direct" visitors doing on your site?

Hitting a specific page?  Loading an image?  Robots crawling?  Robots checking pages for change?

I am suspicious that these are not genuine human visits.

If you have a kickass price, SHOUT it in the <title>tag.</p></title>

How much ranking power do you think that an optimized URL delivers to a page?  Consider that title tag is enormous, links are enormous..... URL is probably tiny tiny tiny.

Now consider that the authority that these pages carry will pass through your site.

So, if this was my site I would be asking... How much of my external linkjuice is hitting these 20 pages.  If none then I would not expect to lose very much by redirecting the pages.  But if these pages had lots of offsite assets pointing to them the URLs would have to be really fugly before I redirected them.

1. Do this query:  site:www.nyc-officespace-leader.com

2. Start drilling down the SERPs.  One page at a time.   Look for content that you didn't make.   Look for duplicates.

3. When you drill down about 44 pages you will find this...

In order to show you the most relevant results, we have omitted some entries very similar to the 440 already displayed.
If you like, you can repeat the search with the omitted results included.

The bad stuff is usually behind that link.  Google doesn't want to show that stuff to people.   It could be thin, it could be duplicate, it could be spammy, they just might not like it.

4. Find out what is in there.

Possible problems that I see....

I see dupe content like this and this.  Either your guys are grabbin' somebodyelse's content or they are grabbin' yours.  Can get you in trouble with Panda. You need original and unique.  Anything that is not original and unique should be deleted, noindexed or rewritten.

A lot of these pages are really skimpy.   Think content can get you into trouble with Panda.   Anything that is skimpy should be deleted, noindexed or beefed up.

I see multiple links to tags on lots of these posts.  That can cause duplicate content problems.

The tag pages are paginated with just a few pages on each.  These can generate extra pages that are low value, suck up your linkjuice or compound duplicate content problems.

You have archive pages, and category pages and more pagination problems.

hmm...  It could be someone manually grabbing your content.... or a robot such as qualys that is visible to your analytics.

I've had both.

Questions received from customers by email and phone are the most important driver of the content plan.   At the same time, you must know the products well enough that you can identify the things that the customer needs to know, but is not asking.

We pay no attention to content length, other than telling enough to convey the information.  I bet we don't have a single 3000 word article on our retail sites.

On a lot of my pages, I have an

title at the top, often with

below it.   Below them, spanning the entire width of the page, I will have a wide format image with a relevant caption.  I think that it is important for the visitors to have an interesting image at the top.  If I don't have one spanning the entire width of the page, then I have my best image at top of the right column.  We often spend good amounts of money on these article-opening images.  We also use these images on category pages and recommended readings lists to advertise the articles to all visitors of the site.

OMG!  No!

If you would have earned #1 position from the beginning of Google, that would have been your best opportunity to have organic traffic that matched what you see in Google Trends.   HOWEVER, Google has become, a better webmaster, more concerned about meeting shareholder expectations, and has begun modifying the format of the search results pages to keep you on their search pages for more page views, display more ads, display more ads at the top of the SERPs, increase shopping results income, make more money.   So, if the #1 organic position, would have remained at the tippy-top of the SERPs for all of those years, then your traffic graph might be similar to Google trends.  Instead, the reality is that your traffic graph would have shown either a much steeper decline or much less dramatic growth.