Bruce Qu (Flickr)

After a year of process in 2015 and extensive (and often heated) public participation, Metro and Sound Transit mostly bit the bullet and forged ahead with aggressive bus restructures to feed the University Link extension. Storied routes such as the 71/72/73 were deleted or shortened, the once-highest ridership line in the city (Route 48) was split into Routes 45 and 48 at UW Station, and Capitol Hill workhorse Route 43 was mostly deleted, kept alive with a handful of vestigial peak runs.

County Council members understood the theoretical benefits of restructures, but also felt the heat from their constituents, who were apoplectic that Metro would “cut” some of the best-ridden routes in the city. Most notably, Councilmember Rod Dembowski’s demanded (and won) partial restoration of Route 71 between Wedgwood, Ravenna, The Ave, and UW Station. In return for sticking their necks out to vote Yes on the restructure, the Council also required a comprehensive report on the ridership changes caused by the restructure. Though that report will be released next month, this morning Metro released the first batch of systemwide ridership data.*

Though there is a ton of data to analyze, the system trends are unmistakable and encouraging:

  • Despite an aggressive ULink restructure, Metro ridership stayed flat, declining by just 0.2%. Despite Link’s rapid growth, Metro still represents 65% of all transit boardings regionwide (121m). Trolleybus ridership grew by 1.2% (19m), and diesel bus ridership fell by 0.4% (101m)
  • Systemwide ridership grew by 4%, from 178m annual boardings to 185m, inclusive of Metro, Sound Transit, Community Transit, Everett Transit, Pierce Transit, and the Monorail. This overall growth occurred despite ridership falling on Pierce Transit (-6%), Everett Transit (-5%), the Monorail (-2%), Tacoma Link (-4%), and the South Lake Union Streetcar (-16%).
  • Link ridership grew by 66% in 2016 compared to 2015, and that’s with only 9 months of ULink. At 19m boardings, in 2016 Link was equal to all 13 Metro trolley routes combined. [Edit: Metro notes that the 19m trolley boardings are on trolleybuses rather than trolleyroutes, so the 19m figure excludes weekend boardings on dieselized trolley routes.]
  • Total ridership on Metro’s restructured routes fell 10.1% – from 114k average weekday boardings in 2015 to 102k average in 2016 – as can be expected from a number of route deletions. But Link more than made up the difference, with average weekday ridership on restructured routes + Link growing by 13.3%, from 149k in 2015 to 168k in 2016.

Neighborhood trends after the jump.

Northeast Seattle

The most hotly contested by far, the NE Seattle restructure went big on a forced transfer at UW Station, dependent on a suboptimal situation at Montlake Triangle. But it appears to be working better than expected, even for routes with the furthest walk from the station via Stevens Way. The frequency boosts in NE Seattle have greatly paid off, with most north-south routes feeding the station seeing explosive growth in ridership. Route 65 (+58%), Route 67 (+200%), and Route 372 (+60%) are leading the pack. Total ridership on the former Route 48 corridor, now composed of Routes 45 and 48, is also up 8%. The major exception is the 15th Ave NE corridor, where Routes 73 and 373 saw ridership fall by 59%. It seems that most riders in the corridor have switched to Route 67, or Route 77 (+20%) during peak.

Capitol Hill Ridership

Ridership on Capitol Hill routes fell significantly (-32%) as riders replaced their slow one-seat bus rides with lightning fast Link trips. In the context of a 2-minute Link trip, the old 25-minute PM peak trips on stinky Breda buses are graciously fading from memory.

The vastly superior walkability of Capitol Hill Station also contributed to the bus ridership losses, as did the fare discrepancy between Metro and Sound Transit. Why pay $2.75 to transfer from bus to Link when you can just walk + Link for $2.25? This is especially apparent on Route 10, where ridership fell 31% from Fall 2015 to Fall 2016. The 15th Ave Business District is only 0.6 miles from the station, and it appears a good number of Hilltop residents are walking instead.

Route 11 initially picked up a ton of former Route 10 riders on E Pine Street, with ridership jumping up 40% between Spring 2015 and Spring 2016. As riders have had more time to compare bus and Link, ridership on Route 11 has fallen 16% from its peak this Spring, settling into a baseline that is still 3% higher than its pre-Link baseline. It would seem that lots of folks are walking across Cal Anderson Park rather than waiting for Route 11.

Route 49 is the exception here, with relatively flat ridership (down 6% from 6,800 to 6,400). Given the frequency boost (from 15 minutes to 12) and the replacement of the Bredas with the new trolleys, ride quality on Route 49 has significantly improved, more comfortable, less crowded, and more frequent than ever. So while the neighborhood has mostly replaced bus trips with rail trips, most riders are having a superior experience no matter their mode.

Northwest Seattle Routes

The changes to Fremont, Wallingford, and Greenwood have been a mixed bag. Routes 26 and 28 shifted from Fremont to their former peak-only variant on Aurora, providing faster trips to Downtown from East Wallingford and Carkeek Park. Route 62 provided new connections between the heart of Wallingford and Fremont. Route 40 shifted onto a new two-way transit lane on Westlake Avenue.

First, the good. The 3 primary north-south routes in NW Seattle are knocking it out of the park. Rapid Ride D (+21%), E (+8%), and Route 5 (+2%) now account for nearly 40,000 weekday rides on their own. The major cross-town pathway, Route 44 (+13%) continues to be one of the most densely ridden routes in Seattle.  And Route 62 jumped out of the gate with 7,400 weekday riders, picking up the slack on the Green Lake/Wallingford/Fremont/Dexter corridor.

One of the unfortunate consequences of the NW Seattle restructure was the further disconnection of Greenwood from Fremont, with neither Route 5 nor Route 28 serving the heart of Fremont. Ridership on both Routes 26 (-20%) and 28 (-22%) fell markedly, though Metro tells us both routes still perform remarkably well in the peak period, and continue to suffer from peak overcrowding. Off-peak, riders are likely choosing the frequency of 5/40/62/D/E over these half-hourly neighborhood routes. Local trips between Fremont/Gas Works/Latona/UW also fell by a surprisingly high amount, with Route 31 down 12% and Route 32 down 7%.

South Lake Union

The March 2016 service change also split Rapid Ride C and D, moved Route 40 to Westlake Avenue in both directions, and installed a (quite successful) transit-only lane. Proving that frequency and broader connections matter more than rail bias, Rapid Ride C (+27%) and Route 40 (+8%) both grew healthily, while SLU streetcar ridership fell by 16%.

We also received stop-level data for thousands of stops in Seattle. We’ll soon be looking at specific stops at Capitol Hill and UW Station to discern any microtrends there.

*The data presented here, from Fall 2016, is accurate but partial. It reflects through the end of January 2017, rather than the end of the current service change period. Therefore, undersampling may cause final numbers to deviate somewhat from what is published here, though the general trends are accurate.  Take with a small grain of salt.

75 Replies to “The ULink Restructure Was a Bold Gamble. It Has (Mostly) Paid Off.”

  1. Great explanation of mostly good news. I look forward to the stop level analysis. Thanks, Zach.

    “frequency and broader connections matter more than rail bias”

    Shocker, who knew? Perhaps I can track down Mike McGinn and tell him the centerpiece of his transit plan — perhaps his mayorship — was a false premise that has amounted to a total waste of time. It won’t accomplish anything, but it might be cathartic.

    1. I agree — but don’t chase him down. At least he was write about the tunnel. Even when that thing is finally done it will be a huge waste of money. In comparison, the money wasted on the streetcar is peanuts.

  2. Coming from the strange area of Wallingford, me and many other riders do miss the ol’ 16. The 26X serves the low density single family areas with the fastest route while the highest density portions are served by the slowest milk-run. Sure, that 62 will be great in 2021 when Northgate Link opens but for now, it’s a super fast way to get between Wallingford and the bars of East Green Lake or Roosevelt since so few people ride it north of 45th.

    1. I actually don’t miss the 16 at all. It was infrequent, unreliable, and didn’t go where I wanted to go. The combination of 62 (frequent connections between Fremont and Roosevelt, where I actually do want to go), and 26 (fast way to get downtown, if infrequent) work far better for me. I’m lucky in that I live close to the 26/62 transfer point at 40th & Stone, though, so can choose which route to take if I want to go downtown.

  3. Is there a reason that the former route 25 was omitted in the NE Seattle analysis? It would be interesting to see how it compared to lackluster ridership on the 78.

    1. It’s hard to compare those numbers since the 25 served so many more areas than the 78. Anecdotally, I think all but a few of those 78 riders are waiting for the 65 or 67 for trips between UWS and U-Village, but the 78 came first.

  4. These data, taken together, send one overarching message: people ride more when there’s more frequency. Half-hourly routes are losing ridership to 15-minute routes. 15-minute routes are losing ridership to 10-minute routes and Link. People don’t want to wait and they don’t want to think about schedules. They are willing to walk a bit farther to transit that works that way. The policy takeaways are to invest in increasing frequency and convenience on core routes, and to make it as easy as possible to transfer between them.

    1. I agree completely. That is the big takeaway. The data supports it, and even stories I’ve heard heard support it. I know someone who would like to take the 76, for example, because it gets him to work faster than the alternative, which is the 372 plus a transfer to Link. But he never takes the 76. He was clear, saying, saying “that train is awesome, it comes so often” instead of “the train is fast”.

    2. Hint hint, Pierce Transit.

      (To be fair, their future route restructure is a step in the right direction)

  5. I’m sorry, but boardings does not equal ridership. For example, if every bus route was just a mile long, and everyone was forced to transfer to go further, the paper ridership using the methods you’ve presented would soar!

    A better analysis would be to take these numbers, and then remove the percent transferring from another route. I’m not sure how good Metro or ST data is that would enable this, though.

    1. The transfer analysis may be possible, but not with the data set we were given. The stop-level analysis should help, but that’ll take a while for us to digest. When the March report comes out it will also have customer satisfaction metrics. Metro told us that customer sentiment improved in 92% of the areas they asked about. It’ll be interesting to see which ones. In our August article Metro was claiming 3,800 bus-rail transfers occurring daily.

      1. The 92% satisfaction metric is great news, and suggests that the restructure really is a success. Even if some folks lost some time, they are still happy with the overall results. 3,800 bus-rail transfers seems ridiculously small, given the truncation of the 71/72/73 buses. That suggests that the bulk of the ridership comes not from people taking the feeder buses (those same buses, now renumbered) but walking. If that really is the case, then the big takeaway in my opinion is again, that frequency is the key. You can achieve that by cutting somewhat redundant service (even if it means asking a lot of people to walk a ways) or by simply spending more money.

    2. I will be interested in seeing these boardings with transfers subtracted when the data become available.

      I’m also curious about some of the Capitol Hill/CD routes not listed here changes. Route 8 and 48 have been shortened — however Link’s opening could either add or subtract boardings on them; I’m not sure if a comparison is possible for these. Finally, I’m also curious if Routes 2 or 12 lost riders to another route like Route 8 or 10 that goes to Capitol Hill Link station or not.

      3,800 sounds very low to me too. Are these new bus-rail transfers, bus-rail transfers at the new stations or net bus-rail transfers for the system? For example, I could see how a UW or SCC student would have previously transferred in Downtown Seattle prior to last spring, but now has a direct ride to UW or Capitol Hill for example. Link has eliminated the need for some transfers.

      Certainly, the 92 percent satisfaction may be the most important aspect of the entire setting! It provides some assurance to elected officials that restructures are not as horrific as some citizens may say that they would be.

      1. I would imagine (and I have some anecdotal evidence to support this) that the 2 and 12 lost a small number of First Hill area riders that are instead walking across Cal Anderson to the station.

      2. Assuming those numbers are for Capitol Hill and UW Stations, which would make sense given the context, that would put the number of bus-rail transfers at the two stations at about 24%. Considering both UW and Seattle Central create a fair amount of traffic just from people walking to the station, that number seems about right to me.

        If Metro & ST follow through with truncating the 545 and 245 at UW Station I would expect that bus transfer number to go up dramatically.

    3. I don’t think that the number of the number of boardings minus the number of transfers is a good number for evaluating a transit system, at least in isolation. You’re assuming that all trips just have a single destination and transfers just slow down trips and create greater complexity for riders, but in a well-designed system, transfers can create greater connectivity and save time.

      I sometimes stop to buy groceries after work, but before going home, or run multiple errands in quick succession in a single trip on my days off- picking up a prescription, shopping at different stores, stopping at the library, getting coffee, meeting up with friends- boarding the bus or train 3, 4, or 5 times within a transfer window. A good, flexible, connected transit system should have the flexibility to make these short, quick trips possible, rather than force people to spread their errands out across multiple unlinked trips. Measuring the goodness of a system by the number of unlinked trips would make the first system look bad, and the second system look good.

  6. Any ideas what’s going on with the 31/32? These routes haven’t changed much! Possibilities I can think of:

    – Changes in UW through-routing: 31/32 used to through-route to 65/75, now through-route to just the frequency-boosted 75. This might result in fewer riders counted as using the 31/32 that were really on there for the through-route partner.

    – Relatedly, additional competition through UW from the 65/67 combo.

    – Additional competition for short trips between Fremont and some parts of Wallingford from the 62.

    – The one small change in the 31/32 themselves: switching from the Stone/40th route to the sparser, less interesting 35th/Wallingford route.

    1. Great question. I was wondering the same thing. I am curious as to which part of those bus routes has seen the decrease. That might give us some clues.

      1. OK, looking at the data on the 31/32 a bit more makes me even more curious. To begin with, not that many people take either bus, nor have the ever done so. In percentage terms this seems like a huge drop-off, but the numbers are so low, that this could just be statistical noise.

        32: 2,569 to 2,400
        31: 1,817 to 1,602

        So basically a few hundred people, which could be explained by the Husky men’s basketball team’s poor performance this year, or folks in Fremont deciding to drink beer in the beer capitol of the state (FreLard) instead of the U-District. Or maybe folks in SPU just aren’t going to the UW that much anymore.

        One thing that jumps out at me is that there is a big difference between the 31 and 32. Given the fact that they share about 3/4 of the route, I think that is interesting. 15th NW, SPU, Fremont, UW, U-Village, Children’s, Sand Point — exactly the same. But somehow the difference between serving lower Queen Anne (on a route exactly the same as the D) accounts for way more riders than a route that serves the most populous part of Magnolia. More than the combined drop-off. Weird, to say the least.

        The big takeaway for me is that folks aren’t fully taking advantage of Link. A trip from Capitol Hill to Fremont requires two seats. You can go clockwise, through downtown, or counter-clockwise, via the UW. Very few, apparently, are going counter-clockwise. Maybe the extra time walking over the triangle is enough to discourage people, or maybe the 8 — now that it is a little more reliable — just makes more sense.

        In the long run — less than five years from now, when Northgate Link is complete — I have to wonder: What happens with these buses? If you add frequency, will you see a big jump in ridership? Does it make sense to go up to 45th, and connect to the station there, or try and stay low? Is speed a problem, and if so, how can you make the route faster?

      2. “A trip from Capitol Hill to Fremont requires two seats. You can go clockwise, through downtown, or counter-clockwise, via the UW. Very few, apparently, are going counter-clockwise. ”

        So far as my observation and experience go, people don’t actually navigate using the bus network, they just punch the addresses into a trip planner and follow whatever directions it gives them. Picking one arbitrary Capitol Hill location and one arbitrary Fremont location, Google maps shows me three routes through downtown. Link is not listed as an option. I have to either place the departure point directly on top of Capitol Hill Station or move the destination into the south end of the U-district before Google starts suggesting Link.

        Personally, if I were thinking of going from Capitol Hill to Fremont, I might start out by thinking of going to the U-district via Link, but I’d have no idea what to do after that – I have no idea which bus routes go east-west in those neighborhoods or where you would find the stops. So, I’d end up using the trip planner, which would route me through downtown.

      3. @RossB
        I think the route 31/32 split is explained by 2 factors:
        -Only Route 32 runs past ~7pm.
        -Route 32 is easily the fastest way between Fremont and LQA and LQA is much more conducive to car free living than east Magnolia. It’s highly walkable, served by very good transit and I suspect has fairer (read more expensive) parking pricing. That creates a lot more demand for route 32 than 31. Anecdotally you can easily be the only person riding the 31 in Magnolia outside off peak.

      4. @RossB:

        > One thing that jumps out at me is that there is a big difference between the 31 and 32. Given the fact that they share about 3/4 of the route, I think that is interesting. 15th NW, SPU, Fremont, UW, U-Village, Children’s, Sand Point — exactly the same. But somehow the difference between serving lower Queen Anne (on a route exactly the same as the D) accounts for way more riders than a route that serves the most populous part of Magnolia. More than the combined drop-off. Weird, to say the least.

        LQA has significantly younger residents than Magnolia on average, and the 31/32 connect to “younger” destinations like SPU, the Fremont tech companies and UW, of course. And there is no easy transfer from 31 Westbound to D Southbound. It’s a 5-10 minute walk through a not-particularly-well-lit route

      5. @Alex — OK, thanks. Yeah, I know the southwest end of the 32 is way more popular than the 31, but was surprised it would make that much difference. The other destinations I mentioned — the areas the routes share — seem way more important. There are no other buses that serve huge sections of it. SPU to Fremont, Fremont to UW. I would have guessed that would be the lion’s share of the ridership. Even if the 32 is a way better route, the fact that it overlaps the D means that it shouldn’t make that much difference. If I’m at SPU headed to lower Queen Anne, I will take whatever bus comes first. If that is the 32, great. If it is the 31 then I transfer to the D.

        I have a feeling the other theory — the fact that there are a lot more 32 buses — might actually account for all of it. On Sunday there is no 31. The ratio of 32 buses to 31 buses the rest of the week line up pretty much exactly with the ridership ratio.

      6. “not that many people take either bus, nor have the ever done so”

        That can’t be. I’ve taken the 75 a lot which is paired with the 31/32, and the 75’s ridership is not that high but then you get to U-Village and campus and Campus Parkway and people start piling in and there are lot of standees — often as many as the 44. The 31/32 are popular, and were so before Link too. If they’re not showing up in the data then something is wrong there.

      7. Mars is definitely correct – how people react to restructures is heavily influenced by where their devices tell them to go

      8. @Ambarish And there is no easy transfer from 31 Westbound to D Southbound. It’s a 5-10 minute walk through a not-particularly-well-lit route

        Oh, very good point. I didn’t realize that, but I should have. It isn’t an easy problem to solve. You can’t put a stop close to 13th and Nickerson, because the bus has to be in the left lane to head towards Emerson ( You can’t put a stop close to the Ballard Bridge, because again, you need to be in the left lane to get to the overpass ( Not only does that make it difficult to transfer from Westbound 31 to D Southbound, but also to D Northbound. This means SPU to Ballard, for example, is a big pain. You could add a bus stop on the other side (on Emerson) closer to the bridge (say, here: but unless you have it actually on the overpass, it would be a hefty walk to the D southbound, and a really long walk to the D northbound.

        It’s a shame, really. The route looks just fine on paper. It’s not that anyone expects the same performance as the 32, but in theory — with the frequency of the D — it shouldn’t matter. But the transfers kill that connection, and thus a lot of the functionality of that route. I understand why you want to go that way — you serve all the big pieces of Magnolia along with the only service to Fisherman’s Terminal, but I have to wonder if there is a better alternative. This is certainly one of those routes where I would like to dig into the stop by stop data.

        If not many people use that Fisherman’s Terminal stop, I wonder if it would be better to simply overlap the 32 a bit longer, and split off at Dravus (where connections to Ballard are much better). That would give the folks in Interbay (an area that is growing rapidly) 15 minute service to Fremont, the UW, etc. It would mean that transfers are much easier. From SPU to lower Queen Anne you take the 32 or 31, whichever comes first. If you do have to transfer, then it is at Dravus, at exactly the same stop (no walking at all). It also would mean the connection between Magnolia (say, Magnolia Village) to Ballard is much better. Again, no walking for the transfer (either direction). The trade-off is less service to one small section of Magnolia (which is served by the 31) and maybe a bit longer walk to Fisherman’s Terminal. That seems like a small price to pay.

    2. I live in Magnolia and sometimes think about riding the 31 to the UWMC but always end up on my bike. I can’t justify walking a mile to catch a bus that will be slower and make me walk from Steven’s Way.

    3. I sometimes ride the 31/32 between Fremont and the U-district on weekends and more frequency would definitely be nice. I’ve tried taking the 62 a couple times after just missing a 32, figuring that walking half the way is still better than a 29 minute wait. It’s ok, but better frequency on evenings/Sundays would definitely be preferred.

      1. Where I live in Wallingford, I take the 31/32 a lot to get to Link (with its terrible connection on Stevens Way) if I miss the 26X. In the other direction, the 32 is very crowded from 6-7pm, often crushloaded leaving 12th and Campus until it gets to about Wallingford Ave. As the only route that goes from southern Wallingford to Fremont, it’s still very useful to us.

    4. I’d agree with Al. I don’t work in Fremont anymore, but when I did, I’d take a 31/32 (forget which) to through route to the 65. I was excited for the 62 instead, which would bypass UW. That kind of thing is probably enough to account for the ridership change.

      1. Heh, I actually didn’t even think about that use of the 62 for longer trips into the NE, but of course the 65 and 62 do cross at NE 65th/35th. The 62 sure is an interesting milk run…

    5. I think you’re right about 31/32 losing ridership due to moving off Stone. You can see that in the 62’s numbers too. I wouldn’t change the route alignments at all, too – Stone Way has enough development between 35th and 45th that it really needs a route like the 62 that runs the entire length.

      That said, 31/32 have *lots* of ridership at peak, to the point where buses are turning away riders. Metro might do well to run some more articulated coaches at peak. Off peak, though, they are quite a bit more empty.

  7. Great report. Fascinating stuff. I would say it does come full of caveats, though. First, ridership isn’t everything. If you take transit downtown, you are likely to take transit downtown no matter what happens. To really figure out whether a restructure has been successful or not requires very in depth analysis, essentially trying to figure out if time has been saved or not. But at a minimum, this is good news — if the opposite happened (less ridership) — then it would be time to worry.

    Second, since the restructure happened right when Link expanded, it isn’t clear at all whether the restructure accounts for a significant increase in ridership or not. Restructured routes + Link grew by about 20,000. That is nice, but hardly worth a parade. The three very urban, much faster, much more frequent trips might account for all of that increase. In other words, overall ridership might be just as high right now if they simply kept the existing routes as they were.

    To me, the big takeaway is that frequency matters, up to a point. An improvement from 15 to 12 minutes is pretty much meaningless. But 30 to 15 or 20 to 10? Enormous. That kind of improvement gets me out of my car and into a bus. Speaking of which, I don’t think it is a coincidence that the (3)73 has disappointing numbers. It is no better than it was before the restructure. Unlike some other routes, it didn’t see an increase in frequency. All that was done — and I appreciate this — is that the 73 doesn’t run at the same time the 373 does. But in the middle of the day if you want to get from Pinehurst to the UW, then you have a half hour wait if you miss the bus. Even at 6:00 PM, the frequency is a half hour. This makes it extremely painful for a transfer trip from say, Fremont. That is the kind of inconvenience that sends people to their cars (or keeps them there).

    1. 12 minutes means “we wanted to do 10 minutes but couldn’t quite afford it”, and same with 20 minutes vs 15. The 49 was going to have 10-minute headways but that got lost in the resurrection of the 43 and 12.

    2. 30 minutes to 15 minutes makes the marginal difference in whether people will take the bus. If people know it will come in 15 minutes they’ll just go there and wait for it, but if it’s 30 minutes they’ll want assurance it hasn’t just left, and if such assurance is not available they’ll drive, postpone their trip, or switch to an alternate destination. That’s what I always used to do with these routes that were 15 minutes daytime, 30 minutes evening & Sunday; I’d take them when I knew they’re frequent, and avoid those areas otherwise.

      15 minutes to 10 minutes has a similar but smaller impact. It sometimes makes trips more feasable and convenient, and if you’re going on several errands it allows you get more accomplished in a day because you’re not waiting as long for buses. That’s another factor drivers cite for not taking transit: they can get more things done in a day in their car. Getting transit to 10-minute frequency or better at least minimizes that discrepency. But I don’t think 15 minutes to 12 makes much difference; it’s still “almost 15 minutes”.

      1. I personally don’t believe that very many people would just walk out and wait for a bus that comes every 15 minutes, especially if the route is not perfectly reliable. You could be standing outside for an entire 15 minutes, or longer if buses were bunched. In my view (and maybe there is research that supports a specific standard) 10 is the minimum headway to consider just walking out the stop, but I actually don’t do that unless headways are more like 6 minutes. At 10 minutes on a route with significant bus bunching, you could still be waiting 10-20 minutes. So I check OBA for almost everything outside of light rail or buses along 3rd/4th/tunnel within downtown.

      2. A 10-minute walk and 10-minute headway is the most common threshold where average Americans will walk to and choose transit, but other changes also affect people’s choices. It can be difficult to say whether one change (30 to 15) has more or less effect than another change (15 to 10) because different people react differently, and it can be difficult to capture all those choices and correctly quantify them. So I may have been overgeneralizing or gotten it reversed when I said that 30 to 15 has a greater impact than 15 to 10.

        However, I stand by my belief that 30 to 15 has a greater impact for at least some people, and that it triggers some factors that aren’t present in a 15 to 10 minute change, at least for some people. The effect may be different for transit fans vs non-transit fans.

        For me, as a transit fan, 30 to 15 makes a major difference in whether I’ll use the evening C, E, 5, 10, or 11. All these routes got evening frequency between 2012 and 2016 except the 11 which still doesn’t have it. So I no longer avoid those areas in the evening. I don’t like waiting 15 minutes but as a transit fan it’s tolerable, and I’m very grateful the 30-minute limit is gone, which lessens my impatience.

        Non-transit fans may not tolerate a 15-minute wait as much. (And the wait exists whether or not you use a schedule. If you use a schedule, you can wait at home rather than at the bus stop, but you still have to watch the clock and time your activities to the bus cycle, which often leads to wasted time.) But when the A replaced the 174, thus going from 30 minutes to 15, it lead to a 50% increase in ridership. That may not be the 100% that one might hope, but it is more riders, and a more convenient experience for them.

      3. 30 to 15 minutes is a much bigger improvement than 15 to 10. 15 minutes is when passengers first get comfortable going to a stop without checking the schedule. Compared to 15, 10 hardly gets noticed by a large number of passengers.

        In 2006, the WTA in Bellingham had the biggest ridership increase in the entire nation. What happened? Simple, they introduced a series of 15-minute corridors, and people flocked to them.

  8. It seems like the charts above really should take into account the deleted routes when considering total ridership.

    For instance, the new 372 presumably has all old 372 riders, plus some from the 68 and 72. And the new 67 has all old 67 riders, plus some from the 66 and 68.

    An example, based on 2014 ridership data at, 25% of ridership on route 66 was between U District and Northgate, and 40% of ridership on route 68 was to Roosevelt/Northgate. So assuming no growth at all, the new 67 “should” have ~3300 riders each day. Which means an increase of ~50%. Quite good, but not tripling ridership.

    For the 372, assuming it “should” have 60% of the old 68 and the 35% of the old 72 that wasn’t going to downtown, one would expect ~7700 riders each day. Which means that the 372 actually stayed flat rather than gaining 60%.

    ULink has been a huge success, but it’s worth paying more attention to the bus numbers.

    1. Right – but it’s hard to tell exactly how much that “sum” is. A good chunk of the former route 72 riders were just going between the Ave and downtown. Now, these people are riding either Link or the 70, and would not be on the 372 at all.

      Anecdotally, I’ve ridden the 372 the most during the Saturday period when it runs every 15 minutes. On evenings/Sundays, I’ll check OneBusAway, but if it’s not coming very quickly, it is usually faster to just start walking than it is to wait for it.

    2. As a former 72/73(& 79) rider, I can tell you that the above is not the case for all of us. The Link transfer is much slower than before, and the 67 is worse still. What I see in my neighborhood (75th & 15th) is folks making the long walk to LCW to pick up the last boarding on the 312 or 522. Even with a >0.5mi walk, it’s faster than any other option (for able bodied adults, mind you). The drawbacks are: often riders are left at the stop when buses fill, and occasionally buses pass without stopping.
      I’d further add that ULink is terrible coming back from a show on a Saturday night for folks on the old 72/73 line up 15th. After 9:30pm plan on the 40 block walk. Next time I’ll 41 to Northgate and 67 back south (absurd given the Link frequency into the night).

      1. It’s important to put things into perspective and remember that in the old days, riding the 72 from Ravenna all the way to downtown wasn’t that great of an option either. At its best (after prop 1 took effect), the 72, by itself, ran only 30 minutes, and was prone to occasional pass-ups at Convention Place Station following major events. It was also not unusual for the #72 to arrive in the U-district 20+ minutes late, following a major event, simply as a result of people squeezing onto the bus in the downtown tunnel.

        At least you have the 522 as mitigation, which is much faster than the old 72 ever was. And, when Roosevelt Station opens in 4 years, things should start getting much better.

      2. That is a good perspective, I was never a fan of the old frequency of the 72; although, I could also catch the 73, which helped a great deal.
        I am looking forward to the opening of Roosevelt Station. But I am concerned about peak AM commute trains leaving Northgate completely full when they eliminate the 41 (and others?). <5min frequency is needed. For context, when I lived in Stockholm the AM Tunnelbana had 3min frequency and was still pretty packed (for neighborhoods more dense than Ravenna/Northgate).

      3. In the gap between 2021 and 2023, Link will have all 4-car trains, compared current service is mixed 2- and 3-car trains. One car can hold one busfull, and the 41 AM is approximately every 6 minutes peak the same as Link, so the capacity should be just about right in terms of no more crowding than currently. The problem may come if Roosevelt and U-District stations push it over the edge. But again, you’re getting one car for the 41 riders, and half of another car (every other run or so) for the people at Roosevelt and U-District. North Link might also spur additional reverse commutes, but they’re going the other direction so they’re irrelevant to this problem. Then in 2023 north Link’s capacity will double.

  9. I have a request for Oran, if he is out there, watching the blog. This would be a great time to somehow resurrect the old data that you had for a while. For those who are unfamiliar, the best transit map in the area is here: For a while, you could toggle between the service map before and after the restructure. It made it really east to see what changed. I don’t know of anything that compares, really.

  10. It would be interesting if you could use the stop level data to tease out the effectiveness of the replacement of the 43 with additional trips on the 8 and 48, particularly off peak and Sundays when there’s significantly less service than before.

    1. Seconded! Taking the Link to the U District just to have to wait for 20+ minutes at the cold, dark, deserted bus stop to get home on Sunday evening is… almost enough to make me want to transfer to the 8 at Capitol Hill and walk for 12 minutes instead, except then I might also be waiting for the 8 for 20 minutes.

  11. Have the 26 and 28 really lost ridership over the portion of the routes they still serve, though? It doesn’t seem like it to me. The Fremont/Dexter to downtown portion of the ridership shifted to the 40/62. The relevant metric would be use of stops north of 39th (for the 28), the 26 is harder to compare because of the new routing.

    As a semi-regular mid-day North Seattle 28 rider, it feels about the same to me.

    1. So do I! That is one route that should never have been deleted, especially with the growth of the SLU area.

    2. The Roosevelt Rapid Ride is essentially a recreation of route 66. It’s supposed to run from Northgate Transit Center down 5th Ave NE to Roosevelt and then downtown via Eastlake.

      SDOT’s shiny new “Draft Seattle RapidRide Expansion Program Report”, dated February 2017 shows that the segment of the Roosevelt Rapid Ride from 45th to downtown should start operating in 2021. The segment north of 45th is “Subject to Further Analysis” in 2021. On the “2024 Seattle Rapidride Network” page it still shows the Roosevelt RapidRide route north of 45th in “Subject to further analysis” mode. There’s no indication when it’ll start running north of 45th.

      Why not just bring back the 66 for the next ten plus years (or more) until the Roosevelt RapidRide route is up and running in it’s entirety?

      1. “Subject to further analysis” means it’s unfunded. They found that Eastlake is in worse condition than expected and Fairview is not an option for a cycletrack, so that sucked the money that would have gone to north of 45th.

        Deleting the 66 was done by Metro, and adding Roosevelt RR is being done by SDOT. They don’t always agree on what the best network should be. Metro did not favor the Madison RR corridor or the SLU streetcar, and it may not have liked the Eastlake/Roosevelt corridor either, but it acquiesced to them because Seattle was going to build them anyway.

        Roosevelt RR as it’s now planned is not the 66 but the 70. It serves Fairview instead of Eastlake south of Mercer.

  12. Route 70 is omitted from this data pull (or I’m just blind).

    Previous reporting here is that ridership is way, way up on the 70.

    Many from the north end of the U Dist appear to have simply replaced their 71/72/73 trip to downtown with a walk+70, rather than the 45+Link.

    Others are transferring at Campus Pkwy, from 73 type routes, rather than staying on board to go to LINK.

    Ridership has grown a crazy amount in the AM peak, when, presumably, these are not UW students from the dorms on Campus Parkway heading downtown. I understand the job growth in SLU and the deletion of the 66, but, that doesn’t account for all the growth. Metro has added a couple of trips in the AM peak to deal with overloads.

    1. You’re right. Route 70 should be included. It didn’t change routing, but it did gain trips, and it did take a good chunk of the former 71/72/73 riders.

  13. Lets tell the truth on why Rod Dembowski wanted the # 71 to continue and it wasn’t because he was looking out for his constituents or the riders of the route. Heck no, the only reason he didn’t want to cancel the # 71 is because he and his family ride the route and that is only reason why.

    He didn’t care one iota about any other routes and I know that for a fact because I was in communications with him during the time when Metro wanted to cancel route # 72 and he assured me and the rest of us who didn’t want the route cancelled but at least routed to the UW Light Rail Station that he was looking out for us and working with Metro on this.

    But when it got down to it he was only concerned about saving his own bus route and didn’t give damn about anyone else. In other words he is just another typical politician looking out for own interest and the interest of special groups and in this case his own family.

    So Rod and his family got to keep their route and an easy transfer at the UW station while those of us from the # 72 now get to ride the overcrowded # 372 and have to walk to and from Stevens Way so that we can get to downtown.

    Thanks Rod and I will remember you NOT at the next election.

    1. There’s a silver lining though because resurrecting the 71 and 73 restored 15-minute service on 15th to 65th. This is a more important corridor than Metro realized.

  14. Great. Metro bricked the 28 and now as ridership has predictably fallen, we’ll hear sometime in the next few months or years that it has to be cut entirely, because god forbid we allow a “low ridership route” to survive. Same with the folks who use the 73.

    The problem with the U Link restructure was that it was built on a false choice that assumed you can have frequency or coverage but not both, and that frequency improvements can only be achieved by cannibalizing routes. The end result of this thinking will be that large swaths of North Seattle will be cut off from transit service. There is no way that this is good.

    I’ve been around this site long enough to know that most regular readers are fine with this. I don’t think most Seattle residents are, but they don’t think they have any other choice (thank you Rod Dembowski for showing that they do), so they muddle along as best they can.

    The climate won’t. We need to replace single occupancy vehicle driving with transit everywhere we can. We also have to do that within the existing land use patterns, even as we work to change zoning. So that means there will be milk runs serving low density neighborhoods as well as high frequency runs serving high density corridors. This is fine and good and as it should be.

    I’m glad that supporters of this restructure got what they wanted out of it, but it did damage to the system and to the city. Let’s hope that we are better organized next time Metro proposes something like it so we can stop it.

    1. “The problem with the U Link restructure was that it was built on a false choice that assumed you can have frequency or coverage but not both.”

      How is that a false choice? Show me the money that buys your idealized coverage without trading off frequency.

      1. Right. Either you have new revenue, or you have to chose between coverage and frequency. Metro generally does not have the luxury of adding significant service hours to restructures, unfortunately..

    2. The 28 is not going away. It still carries more riders than many other routes in the Metro system.

      But frequency does matter. Just yesterday, I took at Car2Go from Carkeek Park to Fremont in direct response to checking OneBusAway and seeing a 24-minute wait for the 28.

    3. So that means there will be milk runs serving low density neighborhoods as well as high frequency runs serving high density corridors. This is fine and good and as it should be.

      Except that infrequent milk runs through low-density neighborhoods don’t replace driving because they’re so much slower than driving. In fact, they’re almost always slower than biking, and if they’re infrequent and indirect enough, often slower than walking over surprisingly long distances! If a bus route is cut in the forest and nobody was riding it, the bus and its driver are making a bigger sound elsewhere in the network.

      The restructure managed to put frequent service within reach of vastly more people than before. That means service that people will actually use! Unfortunately a lot of that service isn’t as useful as it will be after stations at 45th and 65th open. Some origin-destination pairs are less convenient than before (and many of these will improve post-Northgate Link), and some are more convenient. Very, very few areas were “cut off entirely” from service, and if every person that used to ride the bus in those areas switched to driving a coal-rolling F-350 we wouldn’t notice because the number of people is so tiny.

      1. I have a milk run pretty close to my home in Issaquah. I’ve literally never taken it because it’s nearly 40 minutes to Bellevue transit center. It’s always faster to drive to Eastgate P&R or S Bellevue P&R to get anywhere. It only exists to serve the few people who don’t have access to a car.

        i can’t wait for driverless vans to replace all these milk runs…

    4. Some numbers are relative, others are absolute. The worst performers were killed off in the 2014 cuts. the U-Link restructure, and the C/D restructure. What remains is higher-performing and therefore less likely to be deleted. Those purges were one-time, because those routes had lingered for forty years or longer even though they had long become practically ghost routes. So there will be buses on 8th Ave NW and Latona Ave NE; the question is whether they’ll always go to the same termini and neighborhoods. But we don’t have to guess what will happen to them; it’s in Metro’s long-term plan. The 2025 plan keeps the 28 and turns Latona into a Frequent Northgate-UDist route. The 2040 plan keeps that Latona route and turns 8th NW into a Frequent Magnolia-145th route. So unless the economy crashes again or there’s huge opposition in the restructures, that’s what you’ll be getting.

    5. I wouldn’t be too worried about the 28. It serves a corridor that doesn’t overlap with other routes (closest route is the 5, and that’s an arduous, 10-block hike uphill. Now if you were a 78 rider, on the other hand…

    6. “The problem with the U Link restructure was that it was built on a false choice that assumed you can have frequency or coverage but not both, and that frequency improvements can only be achieved by cannibalizing routes… I’ve been around this site long enough to know that most regular readers are fine with this. I don’t think most Seattle residents are, but they don’t think they have any other choice (thank you Rod Dembowski for showing that they do)… The climate won’t. We need to replace single occupancy vehicle driving with transit everywhere we can.”

      That’s a different kind of change and a different kind of advocacy. The restructure is just transit housekeeping, like spring cleaning a house. You’re trying to turn it into a campaign to change tax policies, state laws, and the state constitution. That’s a much larger effort. It would take much longer than a year, it would require convincing politicians who don’t believe in raising taxes. Or replacing them with politicians who do, but that requires changing the minds of voters in conservative district. Most people here think that’s a good idea but it’s not realistic in the short term, and we shouldn’t hobble our transit reforms by making them dependent on unrealistic long-term goals. If we did that, King County would go to the legislature and say “We want to raise taxes so we can implement a U-District restructure without tradeoffs”, the legislature would say “No”, and the whole restructure would fall apart and be nothing. That might help some people, but it would force other people to continue enduring infrequent long milk runs, or taking two buses to get to Link, when they really want frequent service consolidated on an arterial and a bus direct to Link. I’m talking about the original 16, 26, and 72 which frustrated me to no end and made me long for something better.

      Accomplishing your goal would require: repealing the constitutional limit on property taxes, repealing the Eyman initiatives (and legislative substitutes) that lead to a deflationary reduction in governments’ budgets and require votes for certain kinds of taxes, changing the tax structure by adding an income tax or carbon tax or something else, and convincing politicians that transit should have a higher priority and our minimum level should be like Canada or Europe. These are all worthy goals but they require a direct and sustained long-term campaign, they can’t be done as just a trivial part of a transit restructuring.

      There have been a few positive signs. Prop 1 and Move Seattle are voter-approved transit increases within the current tax structure, as was ST3. However, comparable bus-service propositions in King County and Pierce County have failed. So you not only have to change the state tax structure, you also have to convince voters in the rest of King County and Pierce County to go along with the bus increases.

  15. I realize this is anecdotal, but I go downtown now a lot less often.

    From the upper u-dist in the mornings I used to be able to hop on a 71/72/73, spend 10 minutes going down the ave, and then another 10-12 minutes (assuming decent traffic) getting to Westlake station. During the travel I’d sit or stand and get lost in reading or stare out at the nice scenery, etc.

    Now though I hop on a 45/7X, spend at least 12 minutes getting to the montlake triangle area, spend a minute or two using the stairs/elevator and bridge to get to the other side of the street, take an elevator down, wait 0-10 min for Link, and then spend 8 minutes getting to Westlake.

    Doing the reverse is more annoying, especially in inclement weather. Waiting to cross Montlake, waiting at the bus stop (which has nothing around it, no place to grab a snack or even quickly pee). Yes, the frequency is greater, but I’m trading off a lot for that.

    Instead I sometimes just spend time walking to the 70 and taking that ride.

  16. I’ve been living in the eastern half of Ravenna for about two years, and the U-Link restructure has overall been a massive improvement for me. Prior to the restructure, off-peak, I had the 65 and 71, each running only every 30 minutes. Missing a 71 meant either waiting another half hour, or a fairly long walk to catch another bus.

    Now the 65 runs every 15 minutes for most of the week, and I have the 62- also running every 15 minutes- for going to Green Lake, Wallingford, and Fremont, and providing connections to the 73, 45, 67, 40, and 44.

    And for better or worse, I still have the 71 every half hour for most of the week.

    The frustrations I now have in my neighborhood are:

    1) The 65 is an OK way to get to the UW campus, and it probably works fine for getting to the UVillage or Children’s hospital, but its convoluted route makes it a slow way to get to the U-District, especially when classes are changing.

    2) The transfer from the 65 to Link is absurdly awful. I walk fast, and it still takes me about 3-4 minutes to make it off the bus and to the UW station entrance, plus another 1-2 minutes to get down to the platform. For someone who walks slowly or has a mobility impairment, just getting from the bus stop to the Link platform could take 10 minutes or more.

    If the Roosevelt and U-District Stations were open now, I would almost never use UW station.

    3) The reliability and bunching of the 62 feels like it’s significantly improved since it’s inception, but transferring to/from it can be iffy- there are days when I’m heading to the U-District and potentially could transfer to the 73, 45, or 67, and somehow manage to just miss all of them by a minute or two. Moving these routes to 10 minute headways could greatly improve transfers.

  17. I really miss the 73 one-seat ride to and especially from downtown. The 73 took the biggest hit and the plummeting ridership reflects that. It has to backtrack to reach the UW station. There is a significant transfer penalty, especially coming from downtown when you might have to wait along a high-traffic street with no amenities for half an hour. They axed the handy loop through my neighborhood. They axed Sunday service though later restored it. I would be better off with the old 73 that went straight into the bus tunnel and incidentally stopped right in front of my house. The new 73 doesn’t even bother going all the way to Husky stadium on game days. Clearly Metro is trying to kill the 73, which honestly I wouldn’t mind, as long as they waited until the light rail reached Northgate.

    1. Metro did try to kill it but now that kept it and repeatedly expended it due to persistent demand, it’s not going away. Metro originally wanted to remove all-day service on 15th and consolidate it all on Roosevelt. I was always uncomfortable with that because 15th is still pretty dense south of 65th; i thought it should still get 15-minute service. The reason Metro kept the 73 was both community demand and the awkward transfer between the 347/348 and 67 in Northgate: the 347/348 turn west and get caught in the Northgate traffic. When Metro then kept the 71 too, it had the side effect of preserving the 15-minute combined service south of 65th that I thought was important. Unfortunately it also precluded the possibility of putting those hours into the 73 to extend the 15-minute service further north. That would also have helped those who are missing the 72.

      Husky games is not just a 73 issue, it affects all northeast Seattle routes. All of them are routed away from UW Station to the shuttle stop at 15th & Pacific. That goes far beyond Metro, it has to do with the city’s refusal to have dedicated transit lanes to UW Station during games. Without it the buses can get stuck in gridlock and thrown off schedule. Metro moved most buses off Montlake Blvd years ago because traffic bogs down most afternoons — it would be like that only more so.

      Metro’s LRP in 2025 has a Frequent route on 15th from UW Station to Richmond Beach (73/348).

  18. The 26 and 28 are strange routes, so strange factors affect their ridership. Whenever I go to Cartridge World I check the 26’s schedule to see if I can use it, but because it’s half-hourly it’s hit or miss. The 71/72/73X were express and the most frequent routes between north Seattle and central Seattle and ran 24 hours with evening locals and the 83 because of the 35,000 students at the UW and the second downtown in the U-District and because people from other places would go there to catch the expresses. The 26 and 28 are express because… it doesn’t make sense. They’re on small residential streets with only a few people in walking distance, and they run half-hourly so they miss people who won’t bother to check if the half-hourly run will fit their schedule. You should put an express on on the highest-ridership street and make it frequent, as Metro did with the 71.72/73X and CT did with Swift. Instead you have these not-very-frequent expresses going through not-very-dense neighborhoods while people in higher-ridership areas are left out, most obviously the D and, if you consider its entire length, the E.

Comments are closed.