David Schaeffer - Founder, Chairman, Chief Executive Officer and President Thaddeus G. Weed - Chief Financial Officer, Principal Accounting Officer and Treasurer

Brian Hawthorne Colby Synesael - Cowen and Company, LLC, Research Division Scott Goldman - Jefferies LLC, Research Division Walter Piecyk - BTIG, LLC, Research Division Eric Pan - JP Morgan Chase & Co, Research Division Nicholas Ralph Del Deo - MoffettNathanson LLC James D. Breen - William Blair & Company L.L.C., Research Division Frank G. Louthan - Raymond James & Associates, Inc., Research Division Ana Goshko - BofA Merrill Lynch, Research Division Timothy K. Horan - Oppenheimer & Co. Inc., Research Division Michael Rollins - Citigroup Inc, Research Division Michael G. Bowen - Pacific Crest Securities, Inc., Research Division Georgios Kyriakopoulos - SunTrust Robinson Humphrey, Inc., Research Division Barry M. Sine - Drexel Hamilton, LLC, Research Division Michael J. Funk - BofA Merrill Lynch, Research Division James G. Moorman - D.A. Davidson & Co., Research Division

Good day, ladies and gentlemen, and welcome to the Cogent Communications Holdings', Inc. Third Quarter 2014 Earnings Conference Call and Webcast. [Operator Instructions] As a reminder, this conference call is being recorded and will be available for replay at www.cogentco.com. I'd now like to introduce your host for today's conference, Dave Schaeffer, Chairman and CEO of Cogent Communications. Please go ahead, sir.

Good morning, and thank you, all. Welcome to our third quarter 2014 earnings conference call. I'm Dave Schaeffer, Cogent's Chief Executive Officer. With me on this morning's call is Tad Weed, our Chief Financial Officer. We are relatively pleased with our results for the quarter, and we are optimistic in the strength of our business and the outlook for the remainder of 2014 and beyond. During the quarter, we experienced sequential revenue growth, EBITDA margin expansion, EBITDA expansion, sequential traffic growth. Sales rep productivity again was far above our historical averages. During the quarter, we returned a total of $29.7 million to our shareholders through a combination of dividends and buybacks. This $29.7 million return of capital to our shareholders includes a regular recurring dividend payment of $13.8 million or $0.30 per share and an additional $15.9 million paid under our return of capital program through the purchase of stock during the quarter. We purchased a total of 7 -- 476,000 shares of our common stock at an average price of $33.51 during the quarter. During the first 9 months of this year, we have purchased 1.4 million shares of our common stock for approximately $48 million at an average price of $34.24. So far this quarter and through yesterday, we have purchased an additional 218,000 shares of our common stock, spending $6.8 million at an average price of $31.27. We have a total of $40.9 million available under our authorized stock buyback program, and that authorization is allowed to continue through February of 2016. We continue to remain confident in the cash flow-generating capabilities of our business. As a result, as we have indicated in our press release, we announced another regular increase in our dividend, from $0.30 per share per quarter to $0.31 per share per quarter, our…

Thank you, Dave, and good morning, everyone. This third quarter 2014 earnings report and this earnings conference call discuss Cogent's business outlook and contain forward-looking statements within the meaning of Section 27-A and 21-E of the securities act. These forward-looking statements are based upon our current intent, belief and expectations. These forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. You should also be aware that Cogent's expectations do not reflect the potential impact of mergers, acquisitions, other business combinations or financing transactions that may be completed after today. Cogent undertakes no obligation to release publicly any revisions to any forward-looking statement made today or otherwise update or supplement statements made on this call. Also during this call, if we use any non-GAAP financial measures, you will find these reconciled to the GAAP measurement in our earnings release and on our website at cogentco.com. Now I'll turn the call back over to Dave.

Okay. Thanks, Tad. Now for some highlights from the third quarter of this year. Hopefully, you had a chance to review our earnings press release. As within previous quarters, our press release includes a number of historical quarterly metrics. These metrics will be added to our website. Hopefully, you'll find the consistent presentation of these quarterly metrics informative, helpful in understanding the financial results and the trendings from our operations. Our third quarter of 2014 revenue was $95.7 million. Our sequential revenue growth for the quarter was 1.1%, and on a constant-currency basis, our sequential revenue growth for the quarter was 1.8%. Seasonal fluctuations in revenue and -- usage have an impact on our revenues as well as the negative impact from foreign exchange contributed to our lower-than-historical growth rates in the quarter. We believe that our sales force reorganization initiatives will continue to deliver results and the investment that we have made in our sales force are beginning to generate results. Overall trends and highlights for the quarter, now pricing. Our most widely sold Corporate product continues to be 100-megabit per second non-oversubscribed and non-blocked dedicated Internet service, and our most commonly sold NetCentric product is a 10-gigabit port. We continue to offer discounts related to contract term to all of our Corporate and NetCentric customers. We also offer volume commitment discounts to our NetCentric customers. During the quarter, some of our customers took advantage of these discounts and entered into longer-term contracts with Cogent, representing over 2,800 customer connections and increasing their revenue commitment to Cogent over their contract life by $19.4 million. The average price per megabit in our installed base on our new customer contracts decreased during the quarter, both on a sequential and year-over-year basis. The average price per megabit of our installed base declined…

Thank you, Dave, and good morning again, everyone. I'd also like to thank and congratulate the entire Cogent team for their results and our continued hard work and efforts this quarter. We analyze our revenues based upon product class, which is on-net, off-net and noncore, and we also analyze our revenues based upon customer type. We classify all of our customers into 2 types, NetCentric customers and Corporate customers. Our NetCentric customers buy large amounts of bandwidth from us in carrier-neutral data centers, and our Corporate centers buy bandwidth from us in large multi-tenant office buildings. Revenue from our Corporate customers grew by 3.4% from the second quarter. These Corporate customers represented 47.4% of our total customer connections at the end of the third quarter, represented 53.5% of our third quarter revenue. Our Corporate connections grew at an increased rate of 3.8% sequentially for the quarter from 2.8% sequentially from Q1 to Q2 of 2014. Revenue from our NetCentric customers decreased by $600,000 or 1.4% from the second quarter primarily due to $700,000 of negative foreign exchange impact and the seasonal usage trends. Our NetCentric customers represented 52.6% of our total customer connections at the end of the third quarter and 46.5% of our third quarter revenue. Our NetCentric customer connections grew sequentially by 2.4% for the quarter. Revenue by product class. Our on-net revenue was $71.1 million for the quarter, which was a sequential increase of 0.9% and an increase of 10.1% from the third quarter of last year. About 90% of our new sales for the quarter were for our on-net services. On-net customer connections increased 3.1% sequentially for the quarter and increased by 15.8% from the third quarter of last year. We ended the quarter with over 38,500 on-net customer connections on our network and our 2,090…

Okay. Thanks, Tad. Now for a moment about our sales force activity and productivity. We began the third quarter with 337 reps selling our services and ended the quarter with 344 sales reps. We hired 63 sales reps in the quarter and 56% of the -- or 56 sales reps left our company during the quarter. Our monthly rep turnover rate was 5.5% for the third quarter, which is significantly better than our long-term historical average rate, which has been declining to 6.4%. We began the second quarter of 2014 -- or excuse me, the third quarter of 2014 with 310 full-time equivalent sales reps and ended the quarter with 329 full-time equivalent sales reps. We believe we are seeing the benefit from the increased tenure and quality of our sales force, and we anticipate that our additional sales resources will help us further see better improvements in productivity and accelerating revenue growth. Productivity on a full-time equivalent rep basis in the second quarter was 5.9% and similar, again, in the third quarter. This rate of organic rep productivity is substantially better than our long-term historical average of 4.7 units per full-time equivalent rep per month. As a reminder, our sales force rep productivities are not based on contract signing but rather are only calculated upon the delivery of service and the complete turn of a customer's order. Now to the scale and scope of our network. The size and scope of our network continues to grow. We added 33 buildings to our network in the second quarter of -- or excuse me, third quarter of 2014, and we ended the quarter with 2,090 buildings connected to our network. Our network consists of approximately 27,500 miles of metro fiber and approximately 60,000 intercity route miles of fiber. The Cogent network…

[Operator Instructions] Our first question comes from Barry McCarver of Stephens.

This is Brian Hawthorne on for Barry McCarver. My first question is about CapEx and kind of how are you thinking about that going forward. Should we expect this kind of $15 million to $16 million range?

Yes. Thanks, Brian. So we do expect capital to continue to decline. We did spend more capital in the quarter than we had expected, both for the expansion of the network into the 2 new markets, as well as the addition of expert data centers to our network. Now oftentimes, when we acquire these data centers, they are built out, but we are able to get tenant allowances or tenant improvement budgets from the landlords. While we will spend the capital in the quarter to modernize these facilities and bring them online, we receive the offsetting cash but have to amortize that cash as a discount to the lease over the life of the lease. We do anticipate slowing the rate of data center expansion as well as we now need to focus on increasing the utilization of our existing data center footprint, which is only running at about 35%.

Okay, that's great. And then the other one is, how are you thinking about revenue growth coming up here in the fourth quarter?

Sure. So as Tad mentioned, we have a significant headwind ahead of us in terms of foreign exchange. At least, we know where it is today, and we're 5 weeks into the quarter. We anticipate about $1 million or slightly more in FX headwind. With that, we do expect our revenue growth rate to accelerate in the quarter and on a constant-currency basis to materially accelerate, again, on a sequential basis and help us return not only to within our range, but our target is continuing improvement as well as consistent improvement in revenue growth that should allow us to return to the midpoint of our guidance range.

Our next question comes from Colby Synesael of Cowen and Company.

Just also touching on top line growth. So as we think about the expectations earlier this year, there's an expectation that growth would accelerate through the course of the year, ultimately, getting to, call it, 2.5% or 15% growth by the fourth quarter. I think you'll still do that, but certainly, the second and third quarter growth has been slower than anticipated. And while there's been a big focus on improving Corporate, which I think you've done, I think, where the slowdown has been relative to expectations has been more on the NetCentric piece. And while I appreciate the seasonality that occurs within the business, it seems like there might be something else there beyond that. And I'm curious if with the congestion you're experiencing at specific ports with specific ISPs, so recognizing your network is up, is underutilized, but you are seeing congestion at the port itself, are you starting to see customers change the way in which they use Cogent, recognizing that there might be a slowdown in how they're actually getting to customers in that ISP's network? So for example, if Netflix may potentially be the cause for that, other customers are arguably being congested as well and how that might be impacting total top line growth. And to that point, M-Lab came out with network information in terms of congestion last week. And I was wondering if you could touch on what your interpretations are of the information that they put out there.

Yes. Sure, Colby. Thanks a lot. A series of questions there. So first of all, we are seeing improvement in the Corporate sales force in terms of size of sales force, productivity resulting in accelerating growth in that Corporate business. And we have now been able to get our sequential growth rate in Corporate back above our long-term historical average, which is 3.3% quarter-over-quarter. We actually did 3.4% quarter-over-quarter this quarter, and we do anticipate continued improvement as that sales force continues to season and become more tenured and as a number of the initiatives and training that we put in place and management continue to bear fruit. So we actually think there's further improvement to come on the Corporate side. Now touching on the NetCentric side. We are disappointed it did not perform as well as we'd like. I would say the major culprit was actually FX, at least on a sequential basis. The $700,000 of FX headwind that we had entirely hit the NetCentric part of our business. Secondly, we are seeing some improvements in sales productivity on the NetCentric side, but approximately 5% of our aggregate revenues or 10% of NetCentric revenues are capped because of the intransigent of 8 ISPs to upgrade their connectivity to Cogent.

Specific to that though, I mean, is that becoming a bigger issue? I mean it seems like just a quarter ago, your expectations around that improving were better than what we're actually seeing. It's just seems like there's -- this is a very fluid, very dynamic aspect of your business. I'm just trying to get a sense of what's changing quarter-to-quarter here as we go through the course of this year.

Sure, Colby. So a couple points. First of all, let's talk about seasonality. We generally have weaker traffic growth in the summer months. Q3 has 2 summer months, 1 fall month, that being September. We actually saw the amplitude of traffic growth from August to September greater this year than last year and are encouraged by record traffic on our network continuing to increase so far this quarter. Secondly, we have had very limited but some success in pressuring those ISPs to increase port capacity but at a much lower rate than is necessary. So 62% of our exiting traffic goes to Cogent customers. That traffic continues to grow unfettered. Another significant portion of our traffic, almost 20%, goes to the 32 peers that routinely upgrade and have been upgrading at an accelerating rate for us. Unfortunately, 18% of our exiting traffic goes to those ISPs whose ports are congested. Our customers understand that. Now unfortunately for those customers, they have few alternatives. So while our ports are congested, so are the ports of every other major transit provider. Therefore, they are left with 1 of 2 alternatives: to either not have their applications run at the optimal level going to those customers or be forced into a monopolistic relationship with the ISP, paying a higher price per paid connectivity, i.e., paid peering. We have seen the FCC take a fairly aggressive stance and change its position on this. I think we are at the cusp of the commission coming out with some new rules as a result of their proposed rule-making that was issued earlier in the year. Customers understand this, and it actually ties into the M-Labs discussion. So our Corporate customers represent 3% of our traffic but 53.5% of our revenues. We have prioritized or made sure…

So to get back to the midpoint of your guidance that you continue to reaffirm. Is it that the NetCentric business accelerates and improves just beyond, I guess, these ISPs or what you're seeing the congestions of the other ISPs or the growth comps ? Or is it really more an expectation that the Corporate side of the business really sees continued acceleration, kind of get the aggregate companies to that midpoint of the range?

I think it's a combination of both, Colby. So I think we will see improvement in our average sequential growth rate in the Corporate part of our business. It's already above the long-term average at 3.4%. I think we'll continue to see that improve materially from where it is today due to more productivity and more sales rep selling those Corporate products. On the NetCentric side, it will be more volatile. We do need, I think, the regulatory solution to be able to get back to above the 2.7% sequential growth rate in that business. But I do believe we will see improvements there as well. And I think I do want to comment on your M-Labs question. M-Labs was treated as a Corporate customer on our network. It was very visible in the data that they made public that there was no congestion to our customer network such as Cox, Charter, Cablevision, some of the smaller ISPs. But the larger ISPs, like AT&T and Verizon, Comcast and Time Warner, abused their market power. You saw heavy congestion on those ports. When we effectively partitioned our Corporate customers and isolated them from that impact, you saw an improvement in the M-Lab's measurement, but it was only for our Corporate customers. All of our NetCentric customers continue to suffer from this abuse of market power by the ISPs. We believe that the FCC is on the verge of strong regulations, and they've, I think, been pretty vocal in saying they are going to come up with a scheme to support the Open Internet.

Our next question comes from Scott Goldman of Jefferies.

Dave, I guess, a couple of questions. First, on the sales force productivity side and its implications for growth, I guess, a 2-parter there. And one is maybe you could just talk about sales productivity in dollar terms. I feel like we've seen this elevated sales force productivity in terms of orders per rep per month being at the 5.9, 6 level. We haven't really seen any acceleration in the revenue growth coming out of that. So maybe just talk about what it is in dollar terms. Or is this just the order side? And as you look into 4Q and the implied growth to get to that 10% growth level that you've talked about, how much of that is sort of seasonal increases in traffic versus needing increased productivity or increased size of the sales force to hit that number? And then I have one quick follow-up question on the regulatory side.

Yes, sure. Thanks for the question, Scott. So let me start with the rep productivity and revenue growth. So while rep productivity has improved by about 25% from historical averages, we have also seen a decline in ARPUs. So our on-net ARPUs year-over-year declined 5.5% and on the off-net side, about 8%, 8.1%, in large part due to the lower cost of loops. So we have seen an improvement in revenue growth on a constant-currency basis. Because we had under-invested in the sales force and we're not growing the sales force for 4 years, we saw our growth rate, on a constant-currency basis, decelerate to about 8.8% at the trough. It has improved to 9.3% as a result of this additional sales force hiring and additional productivity, even offsetting the declines in ARPU. Also, usage does play a role here. Approximately 5% of our revenues or 10% of our NetCentric revenues are burst-related, meaning there are customers using more than their commit. And because of the limitations on access to these 8 ISPs and the constraint on that approximately 10% of our NetCentric business, we have seen some pressure, further pressure on ARPU and a lack of revenue growth from the NetCentric side of our business. I think you will see improvements in both areas. Now in terms of improvements in the fourth quarter and continuing on through next year, we have a long-term growth target of 10% to 20%, with the goal being at the midpoint of that range. We will see significant FX headwind and will not deliver a full 10% year-over-year growth rate this year because of that foreign exchange. We will see acceleration in revenue growth in the quarter due to all of the sales force initiatives on the Corporate side, which has no seasonality, as well as 3 strong months of traffic growth, which will help both the burst usage and ARPUs on the NetCentric portion of the business. But to hit our long-term goals, we need to continue to improve the sales force both in terms of productivity and size, and I think you're seeing that. And that over the past year, we have increased the number of quota-bearing reps by almost 28%, and over the same 1 year period, we've seen that productivity increase, offsetting these ARPU declines.

That's helpful. And then just following up on the last question. You made a comment about needing regulatory intervention to return the NetCentric back to that historic 2.7% average. Just wondering, assuming the worst-case scenario where you don't get any -- where regulators don't step in to help out on that front, how did your business model change on a go-forward basis?

Sure. So I think there are 2 inevitable consequences. First of all, approximately 10% of that NetCentric revenue can't grow. It's 5% of total revenues, so it means that NetCentric revenue growth average would be reset to 2.5% from 2.7%. Now I think we can offset some of that with more salespeople and better productivity, but I think the full potential of the Internet and the full potential of our NetCentric business will not be realized if these large ISPs are allowed to remain as gatekeepers to their customers building walled gardens. Secondly, we will emphasize the Corporate part of our business, and we will do that in 2 ways: one, continuing to sell more dedicated Internet access and probably accelerate the growth of our VPLS- and MPLS-based pseudowire products, i.e., point-to-point and point-to-multipoint services, which have generally been an add-on to our Corporate customers and not a big part of our business. I think increasingly, we will focus on those Corporate customers that are probably a bit up-market that have multiple sites and use our network to allow us to drive revenue growth on that Corporate side hopefully above even 4% sequentially.

Our next question comes from Walter Piecyk of BTIG.

Dave, can you give some sense if there was any increase in churn with the Corporate customers on a sequential basis related to the fact that -- and I think this was quoted in the press. I think I actually sent you this link. Some of these customers were experiencing some issues on the Corporate side from the stuff that was going on with the increased traffic from the NetCentric business.

Sure. So we basically survey our customers twice a year, and we actually survey every customer every time there is a trouble outage. At the -- earlier this year, in February, because we were not seeing regulatory relief as quickly as possible, we ended up effectively prepping our Corporate customers up, meaning that 3% of the traffic always got through and there was adequate capacity. There was no uptick in Corporate churn as a result of this issue. Now what has been a result of this issue is even if one customer or a series of customers is causing the problem, all of our NetCentric customers are suffering. And it's an unfortunate need for us to manage the traffic exiting our network as we have no capacity constraints on our network. Witness the fact, at peak, we're only 32% utilized, yet we were forced to use these extraordinary traffic management tools to differentiate our customer Corporate customers. And quite honestly, a lot of those Corporate customers run applications like VoIP that are particularly latency-sensitive, but we saw no uptick in Corporate churn. And in fact, that's been reflected in the fact that our on-net churn was flat for the couple quarters. Our off-net churn did spike up from 1.5% to 1.6%, but that's noise, it's -- the trend line has been consistent.

In fact, the Corporate churn was 1.1% in Q2 and 1% in Q3, so it actually slightly ticked down.

So the issue then is on the units. Getting the units back up to maybe 1,600 a quarter is just getting the salespeople to do a better job and deliver a little bit more. I mean, I figured if you hired salespeople, if we look to those on-net customer additions, we would have seen more of an uptick on a sequential basis. I think on the customer net addition, you're down 15% year-on-year despite adding all these salespeople.

Yes, and I think you're right. It takes a while for these salespeople to become fully productive, and we are continuing to see improvement but not as quickly as we'd like.

Understood. So one last question is -- Hank Kilmer was quoted in Ars Technica, if I'm pronouncing that right, article, talking about exactly what you were saying as far as prioritizing, I guess, or whatever word you want to use, network managing the traffic of your Corporate customers. And I guess, I'm curious, when you made this change in February, do you think that, that specifically contributed to the jump in speeds that M-Lab had recorded? Because obviously, there's been a lot of focus on that chart, and I guess the theory has been pushed that, while it was effectively you guys, by basically prioritizing the traffic that made things look better. It wasn't because Comcast and Verizon all of a sudden made things better by signing these interconnection deals with Netflix. So just curious, since Hank has basically admitted that there was this change in February, and you did as well on this call, do you think that's what contributed to that change in speeds that were on the network that everyone's been focused on?

Absolutely. So it had nothing to do with the direct connect agreements that were signed. In fact, Netflix, I think, acknowledged that they hadn't even moved most of their traffic as of those dates. And we were forced to do this because our latency-sensitive applications that we're running on our customers networks were being impacted. M-Labs is not a NetCentric customer; they are a Corporate customer, and their ports were treated as such. So the improvement that they saw concurrently to all 4 networks on the same day occurred because we effectively prioritized there traffic out of necessity, not out of desire.

Because the thing that's interesting that John Oliver makes this huge deal about how -- oh, look, right when Comcast and Verizon signed these agreements, look at how the traffic improved. And the reality was it may be had more to do with how you're prioritizing traffic at the time.

Well, 2 different things. Traffic improved for our Corporate customers. I will say that when these direct agreements were signed between Netflix and the ISPs, Netflix's experience did improve as a result of having more port capacity available; therefore, their bps could flow. Previously, their bps were constrained going through Cogent and every other trans or provider. They continue to be constrained going through us and every other transit provider, but they have an alternate, more expensive direct route, euphemistically called a fast lane, which is probably not a -- it's not what the FCC wants to see.

Our next question comes from Eric Pan of JPMorgan. Eric Pan - JP Morgan Chase & Co, Research Division: A little more on CapEx. You previously talked about plans to moderate your CapEx spending as a result of your building expansions going forward. This comes at a time when a lot of your peers, large and small, are adding buildings on it almost as their budgets will allow. Can you give us some color on the defensive strategy and the economics?

Sure, Eric. So we added buildings at a consistent rate to what we have been doing over the past several years. We do continue to expect our building addition rate to moderate on the Corporate side as we are completing the construction to all the buildings on our target list. On the data center side, we actually have seen some slight acceleration in the number of carrier-neutral data centers that are being constructed, and therefore, we connect into those facilities to remain able to serve the entire addressable market. Now I think the bulk of our higher CapEx came actually from things that were different than those 2 causes. They came from, one, the addition of long-haul mileage into 2 new markets that we had been planning and had mentioned to investors, but were waiting on regulatory approval to be able to do that, that is Albania Montenegro completing our network footprint in Eastern Europe and the Balkans; and then secondly, we have been opportunistic about acquiring Cogent-owned data centers. And in the past year, we have increased the amount raised for square footage in our network by 25%. That has resulted in us spending CapEx to take what were completely vacant data centers that have been mothballed but constructed and bringing them back online and into current compliance. Now in every instance, we've been able to get the landlord, the owner of the facility to pay for those costs, but we record the CapEx up-front and then depreciate it over the life of the various improvements. And the landlord's cash payment to us is amortized over the life of the lease as a rent discount. So it has distorted our capital and made it look larger than it probably really was. Eric Pan - JP Morgan Chase & Co, Research Division: So on the Corporate side, would you consider expanding your addressable market into buildings with much fewer tenants if you're able to get a contract in place beforehand?

Probably not. So our average building is approximately 550,000 feet. We, today, have 12 customers per building. That's the highest penetration we've ever had out of the 51 opportunities. I think without a material revenue commitment from a customer, we would not build into those additional smaller buildings that have fewer tenants. We have been pretty disciplined about where we build, and today, there is no tenant that's ubiquitous and offer large enough to give us that kind of anchor commitments. So I think we expect the pace of about 15 Corporate buildings a quarter if tax rate continued to decline. Eric Pan - JP Morgan Chase & Co, Research Division: Got it. That's very helpful. And then as your building expansion slows and your CapEx comes down, have you considered -- instead of returning the capital, redeploying that spending to perhaps acquiring dark fiber assets or getting into the fiber-to-the-tower business?

We do a limited amount of fiber backhaul from our buildings. We do not view that as our core competency and have no intention of either constructing or acquiring companies to be in the dark fiber business. We view our skills in selling Internet connectivity and IP connectivity, and we leave that to others. Eric Pan - JP Morgan Chase & Co, Research Division: Got you. And lastly, have you thought about expanding into the consumer broadband market?

So we indirectly serve a number of consumers. First of all, we sell bandwidth to all of the large content generators that are targeted to consumers. Secondly, we sell to over 90 international PTTs and thousands of regional consumer-focused service providers. So we indirectly, through our NetCentric business, serve those market segments. We have no desire to become a residential ISP, and you've seen, I think, the 2 competitive overbuilds of fiber, that being FiOS and Google Fiber both struggling and probably generating no return on capital.

Our next question comes from Nick Del Deo of MoffettNathanson.

Dave, 2 questions. First, looks like your revenue from Netflix was down modestly in Q3 versus Q2 despite their European expansion. And I think why the number of networks connected Cogent fell to 5,100 from 5,200 in the quarter, which I think is the first time you've ever had a decline. I was hoping you could comment on those items.

Absolutely. So first of all, with regard to Netflix, they, like all other NetCentric content companies, have some seasonality. People just watch less hours per day of video in the summer. In Q2, you really have 1 summer month, June. In Q3, you have 2: July and August. Their launch in Europe was in 2 phases. At the beginning of the year, we saw a significant turn-up in Northern Europe, the U.K. and Scandinavia. Their French and German launches were actually later in the year and actually not until -- a little later than they had originally planned, that being September 15. So we only got the impact of those markets in the last 2 weeks of the quarter. It did result in a very slight decline, but very slight, in their revenue, but they have a take-or-pay, and we expect that revenue to continue to grow. Now with regard to networks connected, we have gone back to a number of our network operators that buy upstream from us. Many of these companies operate multiple networks, i.e., more than 1 autonomous system number. We have had a program to go through and try to convince all of those networks that have more than 1 AS to concentrate traffic on one of those networks as opposed to having multiple ASs connect to us. There are 2 reasons for that. The first one is actually for the entire Internet. Many of you may remember earlier in the year, Verizon had an issue when the routing table, due to the way they were configuring their BJP sessions, actually went beyond 512,000 entries. It's, today, hovering at around 480,000. By forcing these networks to groom and consolidate, we can effectively extend the time at which some of those older routers can continue to operate. Now we, ourselves, have none of those devices on our networks, so we have no issue, but it doesn't do us any good if the routing table gets bigger than our peers' routers can support. So we've been pretty active in trying to get people to groom down their networks and get to as few of the ASs as possible. And that's really all that's going on here.

Our next question comes from James Breen of William Blair. James D. Breen - William Blair & Company L.L.C., Research Division: Just around the margin you had addressed a little bit. When we look at sort of where potential numbers are for 2014 for EBITDA and you adjust for the equipment sales as well as legal fees and currency, it seems like EBITDA margins are only up modestly, maybe 30 or 40 basis points just based on the actual growth rate. Can you just talk about what's the impact there and what gives you confidence that the EBITDA margins continue to grow 200 basis points a year as we move through 2015?

Sure, Jim. So 3 points. First of all, we have very high contribution margins on incremental sales: on-net, 95% EBITDA margin contribution; $0.45 off-net. We have 2 extraordinary expenses that have hit us. The first being totally self-inflicted, that being our decision to ramp up the sales force. And when we made that decision, we concurrently decided to sell off surplus equipment to help offset those costs. We will continue to see equipment sale gains for the next couple of quarters. We also anticipate the rate of sales force expansion to begin to moderate. Now we expect the lower turnover, coupled with a more moderate rating growth to do 2 things: one, moderate the growth in SG&A, maybe even have it flat to slightly declining; and secondly, accelerating revenue growth. So we invested in our sales force, that was planned, and we plan to offset that with these equipment sale gains. That's the first half of the story. The second half was unplanned. It was -- we did not want to be a crusader for the Open Internet. We did not want to spend $5 million fighting a regulatory battle. We had to react to the fact that a large number of companies were acting in a cartel-like manner to retard the advancement of the Internet, directly opposed to our interest to consumers' interests. That is a temporary expenditure. So we will see 3 things occur next year: one, by mid-year, these equipment sales will taper off, but they will continue into the beginning of the year; two, we will see a dramatic and material decrease in our expenditures for legal and economic analysis; and then, third, we will see a moderation in the rate of sales growth that -- and a continued seasoning of the sales force that will accelerate revenue growth with the high contribution margins. So we feel very comfortable that our long-term average of 200 basis points a year can continue for the foreseeable future.

Our next question comes from Frank Louthan of Raymond James. Frank G. Louthan - Raymond James & Associates, Inc., Research Division: I want to just go back on the net neutrality thing, and maybe if you could just be a little bit more clear exactly what you're trying to accomplish. Because when we talk about net neutrality, a lot of things tend to come to mind. But I think to myself and most investors, your cushion is a fairly unemotional rational -- economically rational company. In most net neutrality commentary, it tends to be the antithesis of that. So what exactly are you trying to accomplish within that sort of movement? Are you pushing for Title II regulation? Is it just simply looking at the smaller aspect of it that affects you on the peering side? How broadly are you supportive of net neutrality? And what would you see as -- how would you define being successful with this expense that you're going to in pushing this issue?

Okay, Frank. So thank you for the compliment that we're rational and nonemotional, and I hope that is true. Cogent was founded with the fundamental premise that the Internet was the network of the future, and I think that assumption has come to fruition. However, there are legacy networks that are trying to protect their vertically integrated business models. So at one level, anything that retards the advancement of the Internet, its usefulness to consumers, is ultimately bad for Cogent. That's a little too indirect for us to invest $5 million. Now net neutrality means a lot of different things to different people. It's interesting, the ISPs that are violating net neutrality all publicly state that they support net neutrality, yet their actions speak the exact opposite. Net neutrality means every bit that a consumer pays for can be delivered to them equally, and they get what they pay for, and the bits that they get over the Internet are comparable in quality and quantity to what they get from a closed wall network, and the customer is free to pick the applications that they choose. Now unfortunately, because the FCC's authority to use Section 706 of the Telecom Act was upheld by the courts, the ISPs realized that the only way they could degrade Internet traffic was doing so through the interconnection points, i.e., directly where Cogent feels the pain. And it's not only Cogent; it's every other backbone provider as well. So we have experienced behavior that has effectively impacted 5% of our revenue, saying they can't grow, and 10% of our NetCentric revenues can't grow. We're spending $5 million to allow that 5% of our revenues to grow at our historical average. With a 95% EBITDA margin contribution, if we're successful, we get a payback on that…

So we are supportive of Title II. We actually think that's the right answer. Two, I do not believe Title II will disincent investment. I think that the disincentive to invest comes from the need to protect one's legacy businesses and preserve the functionality of an obsolete outside plan. We are an all-fiber end-to-end company. We think that the value and functionality of the Internet will continue, and as long as innovators are free to come up with new applications and deliver those to end users, the demand will be, for the foreseeable future, insatiable. So I am certain there will be some unintended consequences of any regulatory action. If people behave correctly and abided by the law, this would not be an issue, but what we have seen is companies that have a long history of using market power to circumvent the law. They are continuing to pressure competitive companies, such as ourselves and ultimately the users of these companies, into a monopolistic relationship. It's a mess, and we've got the best solution to what is an imperfect situation.

Our next question comes from Ana Goshko of Bank of America.

It's been a very lengthy and thorough call, so I'll keep it brief. Dave, you do have some bonds that are very high-coupon over 8%, and they become callable in February. So I want to understand what your plan is and timing for addressing those, so we could get a better understanding of how much free cash flow savings the refinancing could result in for next year.

Thanks, Anna, for the question. So I think Tad mentioned it in our prepared remarks. We are currently evaluating that. We do not plan on, I think, trying to buy the bonds prior to the call date. At the call date, we will evaluate the market, and it is almost a certainty that if the market remains as it is today, we will go ahead and refinance those bonds, i.e., pay the May call, exercise our call and lower our interest rates. I think the cash savings could be pretty material, and the payback on that May call is probably about an 18-month payback. So I think those holders should expect us to refinance.

Okay. And then would you refinance with secured debt?

We're going to look at our cost of capital and the various instruments that are out there, whether they be floating or fixed, secured or unsecured. But I think we will be very focused on our absolute lowest cost of capital. And for that reason, it will probably be secured, and it will probably be of relatively short tenor.

Our next question comes from Tim Horan of Oppenheimer. Timothy K. Horan - Oppenheimer & Co. Inc., Research Division: So Dave, in your opinion does -- is the FCC, FTC or DOJ, are they on your side with this argument? Do they understand the argument at this point? And I guess, clearly, you want the last small provider to add ports. Is there a referee or someone that can really force them to do this?

So 2 questions. Ultimately, you need to ask the regulators. I think what we have seen publicly, which is what I can comment on, the FCC has come out, along with the White House, and supported a definition of net neutrality that we think is accurate and gets us to where we need to be. Two, the FCC has, I think, modified its position on the bifurcation of the problem and really understands now that interconnection is part and parcel of net neutrality, and to be able to deliver an end-to-end user experience, you need to regulate all or, at least, make sure all parts of the traffic path remain uncongested. In terms of who's the referee, I think it's very clear, under the '96 act and the '34 act, that it's the FCC. Now we can get into a legal debate whether that's what's allowed under 706, Title I, Title VI or Title II. I think it will be a hybrid of all those. And I do believe after the commission received 3.7 million written comments on its proposed rulemaking, the preponderance, over 99% of those comments in favor of net neutrality. I think the public has spoken. I think the regulators understand what they need to do, and there is now a process going on to try to craft those regulations to make them as sustainable as possible to the legal challenges that there almost a certainty. Listen, Verizon was so concerned that it preemptively told the FCC it's going to sue it before it even knew what the rules were. Timothy K. Horan - Oppenheimer & Co. Inc., Research Division: I know it's only 5% of your revenues, but it seems like it's impacting a good piece of the other business because it's 18% of your terminating traffic and it -- and now the customers must be worried about it. Can you talk about maybe how much overall revenue impact it truly has been? Because it does seem to be much more than what your basis at this point. Or said another way, how much faster could you be growing if it wasn't for this?

Well, I actually think my characterization of the revenue is pretty accurate. Listen, it's always difficult to do a counterfactual on a what-if. We are growing -- on a constant-currency basis, we grew 9.3%. There's no other telecom services company or datacom services company that's growing anywhere near that rapidly. So I think we're doing pretty well. I can't tell you how the lack of connectivity impacts a venture capitalist decision to invest in the next Facebook or Snapchat or Dropbox. Are they concerned that they can't get access to the customer so they don't fund the entrepreneur to come up with a new application that drives consumers' desire to see more or use more Internet? Does it slow down someone's desire to put more video on the Internet? These are things that I just don't have great visibility to. What I can tell you is the things that are under our control, our sales force, our Corporate business is accelerating its growth. The NetCentric business -- 90% of that NetCentric business is growing, and it is growing on a constant-currency basis as well as a nominal basis. But there's clearly an overhang here, and I think it's probably far broader than Cogent in terms of its impact on society. And I think -- listen, the FCC is not acting to protect my 5% of the revenue. And quite honestly, if that was my only argument to them, they shouldn't listen to Cogent. They should be looking at this and understanding is the Internet transformational for our society? And whether it be the proposed mergers or net neutrality, I don't think anyone wants to preside over the creation of the next AT&T or the death of the Internet.

Our next question comes from Michael Rollins of Citigroup.

Maybe just to change topics for a moment. Just curious, Dave, if you can talk a little bit more about your Corporate business. Maybe give us a range of experience you have with penetration, your buildings and, as you look at the competition on the ground today, what you think a fair average penetration makes sense for the business to aspire to over the next 3 to 5 years.

Sure, Mike. So when we look at our Corporate buildings by age, we actually continue to see the addition of about 1.8 customers per building, whether the building has been on-net 12 years or 1 year. Now there's a fair amount of variability in there. Our best Corporate building has over 200 connections in it, but that's an outlier. Our average Corporate building has 12 as we migrate to selling some of these additional services like VPN services, VPLS and disaster recovery to customers. It actually increases the size of the addressable market above the 51 tenants because the tenant can actually buy more than one connection, one to the Internet and one to a remote site. And I think it's reasonable that we can expect to see, at least, a doubling of our penetration. Over time, now, I don't think that's over 2 to 3 years, I think that's probably going to take 7 or 8 years for us to ultimately reach that penetration. And because of the types of Corporate buildings we focus on, both size and location, we have not seen the type of competition that cable has given to the more suburban, smaller single-tenant buildings.

Our next question comes from Michael Bowen of Pacific Crest.

I think you're setting a record here for length.

I'm trying to keep it short, Mike, but I'm only answering people's questions.

No, it's fine. One quick one and one little longer one. I guess on -- since you've telegraphed that you're going to spend $5 million in legal fees this year, and you don't expect to spend any next year, is there anything that can delay this process, either by the ISPs or the FCC, which could potentially force you to spend more into -- or additional funds in 2015 that you're not contemplating right now? And then maybe a little bit longer one, I understand the sales force churn is still down from a historical average, but it did tick up to 5.5% from, I think, 5%, if I'm not mistaken, last quarter. I'm trying to get a feel for what type of reps are leaving and how that might be affecting your on-net revenue growth or your on-net ARPU. Can you help perhaps connect the dots there? Because I think ultimately, we're trying to figure out how you're going to get back to that 10% to 20% range. And then finally, any thoughts on tightening that range from 10% to 20% maybe down to a smaller range?

Sure, Michael, a number of questions here. So first of all, on the legal fees, it's always harder. It always takes longer than you expect. We have been encyclopedic in our delivery of data and statistics to the various regulators. I do not believe there is more that we have to give. Most of what is going on at this point is now just interaction with those regulators. So while the process may get delayed, the large expense and our response to the civil investigative demands are complete at this point. And I'll be honest, I don't see what else they can ask us. So I think that number is probably a good number, and I do see it going down even if the issue is not resolved. Two, to your second question on sales force churn, the answer is it did tick up slightly from 5% to 5.5%. Part of that is as the sales force hiring that we did a year ago continues to mature and these people continue to have larger quotas, they're measured against a higher bar and some people that made the initial cut aren't necessarily making a long-term cut. I do not believe that the sales force changes have had any impact on the type of customer we're selling or the ARPU. The ARPU changes are really coming from dynamics in the market declines, on price per megabit, on the NetCentric side, average port size and contract length. No real impact from the type or turnover of the sales force. And then finally, on your range question, we are not in the business of giving very specific guidance but rather think it's in our best interest to give general broad guidance, and we feel very comfortable that there's ample addressable market, ample ability for us to hire and train salespeople to achieve the midpoint of our range. And by keeping that range broad over the long term, I think it gives investors the ability to value Cogent not on a short-term basis, but rather over the long term against the potential of the business to produce cash flow. So we're going to stick with that wide range.

Yes. And if I could, I just want to make it real clear here. Because I think on -- one of the first questions that was asked was -- I think Colby actually said something about -- and you thought you might exit the year still at the midpoint of that range, and I wasn't -- I may have missed it. I wasn't sure whether you affirmed that, whether that included FX or not. I just want to make sure we put a fine point on that.

Right. So let's be crystal clear. First of all, we will have significant FX headwind. Two, we will be slightly below our full year range; three, we will see a sequential improvement in our growth rate. And what I have said publicly, and we still support, is by the end of the year, we will be on a sustainable path to deliver sequential growth rates that will get us to the midpoint of the range. That does not mean we will be at the midpoint of the range this year. It means we will build a foundation of number of salespeople, productivity of those people, to get us there next year.

Our next question comes from Georgios Kyriakopoulos of SunTrust.

Given that you increased your minimum capital return to $12 million per quarter and also moved dividend higher by $0.01, can you discuss your 2.5x leverage target and, more specifically, the flexibility you might have with the model to expand beyond the target?

Sure, sure. George, first of all, pleasure to meet you over the phone. So we are committed to producing an increasing amount of cash and returning that cash to shareholders. We set a target that was at the low end of any of the companies in our comparable universe. We will continue to evaluate our cost of capital and what is the appropriate amount of net debt that the business can support. We are, today, under-levered and will continue to be in that. I think once we reached 2.5x, the board will evaluate that, but at this point, I'm not in a position to give you any additional targets other than the ones we've laid out.

Okay. So then you would expect to reach the target before you make any decisions. Is that correct?

I think that's correct, sir.

Okay. And then lastly, Dave. If I step back and think about your NetCentric addressable market, which is somewhere between $1.5 billion and $2 billion, it would imply that your market share is around 10% and growing, given the overall flat growth profile of this market. Now what do you think is a realistic market sales target for Cogent? And how fast do you think that you could get there?

Sure. So first of all, your correct in sizing the market. While we are about 12% in the market by dollars, we are over 20% of the market by bps. I believe we can realistically get to be 1/3 of the market by dollars by continuing to grow faster than the market and gaining share due to the value we deliver to our customers and our pricing model. I think we've got 3 more -- we're going to -- this will be a record call here, but we're trying to get these done quickly.

Our next question comes from Barry Sine of Drexel Hamilton.

Another net neutrality question for you. I thought it was an interesting comment you made that on both Google Fiber and Verizon FiOS, they're not getting any return on capital because they spend obviously on what they're charging. So if the future comes to pass as you hope on net neutrality, it would seem to me for networks like that for end household networks with fiber, there's 2 ways to raise that return of capital, either charge the incoming content or bps coming in, which, I guess, they're trying to do through prioritization, or raise the access rates that they're charging customers for monthly subscription rates. So is that the route that you expect the industry to take? And if so, does that not run into yet another perhaps even more important public interest issue, which is greater penetration broadband, especially the lower-income households?

Sure. So a couple of points. First of all, the commission, I think, is evaluating and probably close to redefining what broadband is and raising that bar from 4 megabits to 25. Two, the existing outside plans, that being twisted pair and HFC coax, were never designed to carry Internet are inadequate and eventually do need to be replaced with fiber closer to the PRAM, the ideal being fiber all the way to the end-user. Third, it is a certainty that people will pay for what they use. Usage-based pricing is fair. That doesn't mean prices need to rise. It means that the legacy providers will have to not continue to subsidize their legacy businesses from their broadband business but rather take the profits of those broadband businesses and reinvest in modernizing their outside plant. The challenge for FiOS and Google Fiber is that they are overbuilders with low market share. So you can't confuse the fact that local axis is a national monopoly, and any competitor has a disadvantage in that it has low market share compared to the incumbent. What really needs to occur is a set of incentives to incent the incumbent operator to invest, and I think you will see that with later phases of regulation. Listen, local axis is a monopoly, and the only way you can ensure the monopolies ever invest is through regulation because they never have an economic incentive to invest.

Our next question comes from Michael Funk of Bank of America Merrill Lynch.

Two quick questions, I think, that haven't been asked yet. First, how are you thinking about the dividend payout ratio longer-term as a percentage of free cash flow? And do you have like a high-end target for that? And then second, you did show some improvement in working capital this quarter with the accounts receivable improvement. Where are you driving there? Is there room for more improvement in working capital in the next few quarters?

Sure. So first of all, let me take the payout one. I'll let Tad take the working capital one. In terms of -- our goal is to return capital. If we have internal places at the business to generate returns significantly above our cost of capital, we're going to go ahead and invest in the business. We've been pretty clear that we cannot find enough of those opportunities. We also do not see inorganic opportunities that make economic sense. For that reason, we're committed to returning capital. We do need -- have no desire to hoard capital on our balance sheet. So we will return that capital, right now, more than 100% of free cash flow to investors by levering up. That makes economic sense in a low interest rate environment. If interest rates remain low, we will reach an optimal leverage target and then return the vast majority of our capital either through a dividend or buyback. I think the combination of those 2 tools is more powerful than just a dividend. So I don't think the dividend will ever approach 100% of free cash flow, but it will be an important part of our return of capital strategy. And I think you've seen in the last couple of quarters, we're buying back about 1% of the company, and we're giving a dividend equal to about 1% of the market cap of the company, kind of a balanced approach. That seems pretty reasonable at this point in time. I can't predict where the stock is going to go or what the cost of capital is going to be. What I can say is we're committed to making sure investors get the capital. We don't hoard it. And then Tad will take the working capital one.

Yes, on working capital, I mean, first off, the payments that have the most material impact on our working capital each quarter, as I mentioned in the script, are the timing of the interest payments. I know that wasn't your question, but I just want to point that out again, and as a reminder, on the $240 million of notes we paid that in February and August and then on the new $200 million notes, we pay that in April and October. So those statements have a material impact on the total working capital. With respect to accounts receivable, we did a great job of lowering the days sales from 28 to 26. That's probably a historical low. I don't anticipate any material deviation from that. We can get in any one day in our lockbox north of $3 million. So every customer being billed monthly, the timing of kind of that last receipt in the lockbox can have a relatively material impact on what our ultimate days sales and working capital is with respect to accounts receivable. So our group continues to do a great job on monitoring that and making sure customers are living up to their obligations to pay us, but I don't anticipate a material change one way or the other in days sales.

For our final question, it is from James Moorman of D.A. Davidson. James G. Moorman - D.A. Davidson & Co., Research Division: I'll keep it real quick since I'm the last one. On Netflix Europe, with their launch of additional countries in September, I realized there's probably not much data from this quarter. But just what have you seen so far and maybe what to expect for Q4?

Sure. So I think they're growing. I want to be careful about really commenting on the business model of a specific customer. Netflix has been the largest -- it's just become the largest distributor of video content around the world, and I think their penetrations outside the U.S. are very minimal. So I expect rapid growth from them. I do think there are some unique regulatory requirements, and some of the European countries are different than in the U.S., and we expect continued growth from Netflix. But I wouldn't feel comfortable giving an exact prediction of how quickly they'll grow.

Thank you. That's our final question. I'd like to turn the call back over to management for any further remarks.

I just want to thank everyone. Hopefully -- it was a very long call but a complete one -- and take care, and we'll talk next quarter. Thanks. Bye-bye.

Ladies and gentlemen, thank you for participating in today's conference. This concludes today's program. You may all disconnect. Everyone, have a great day.