Behind the plan

How my Cowtown plan
was built.

Every choice in my 23-week plan, why I made it, and where it came from. Where a choice comes from a published training plan or a study, it’s cited. Where it was a judgment call for my body and my life, I say so. Last updated Nov 21, 2026.

Start here

The short version

My plan is a 23-week build from the San Antonio half (Sep 27) to the Cowtown Marathon (Feb 28, 2027). I built it with Claude, then checked it against published plans from the B.A.A. (the Boston Marathon’s organizers), Nike and Jeff Galloway’s runDisney plans, and had it audited a second time. You can see that comparison on the training guides page. The core ideas:

Starting point

What shaped every choice

Goals

RaceDateGoal
Sugar Land 10KOct 11 (done)Fun run. Finished in 1:00:59
Run for Breast Cancer HalfDec 122:40 to 2:45. Anything under 2:50:05 is a PR
Cowtown 5KFeb 27An easy, walk-heavy shake-out, part of the Cowtown Challenge
Cowtown MarathonFeb 28Finish before the 7.5-hour cutoff, that’s the real goal. Aiming for under 6:00 (13:30 to 13:45 a mile). Reach: about 5 hours

Where I started from

My week and my preferences

The big picture

The phases, in order

PhaseDatesWeeksJob
RecoverySep 21 to Oct 113Recover from the San Antonio half. Short, easy runs, plus the Sugar Land 10K as a fun run
BaseOct 12 to Nov 226Rebuild the running habit. Long run grows from 4 to 10 miles. Short hill repeats start
Marathon blockNov 23 to Jan 3110Long runs from 11 up to 20 miles, mid-week medium-long runs, the Dec 12 half, race rehearsals, two fartleks
TaperFeb 1 to Feb 213Long runs step down 14, 10, 6 while the legs freshen up
Race weekendFeb 22 to 281Cowtown 5K on Saturday, marathon on Sunday

Why this order: recovery first, so the half’s soreness and my knee settle before anything builds. Then a base of easy miles before the long runs get long, and race-specific work only after that base exists. Last, a taper so I start fresh.

The B.A.A. plans use the same order in four phases: a 3-week prep, a 6-week building phase, a 9-week marathon-specific phase, and a 2-week taper. My blocks are similar in length. My base starts gentler and my taper is a week longer, but its first week (a 14-miler, 3 weeks out) does the same job as the end of their marathon-specific phase.

Day to day

The weekly shape

A typical week in the marathon block:

DayRunWhy
Mon3 miles easyShort, loosens the legs
Tue3 miles, group runSocial, easy pace
WedMedium-long run (5 to 8 mi), sometimes with hill repeats or a fartlekThe second most important run. Builds strength for the late miles
Thu3 miles, group runSocial, easy pace
FriRest, optional easy walkFresh legs for Saturday
SatLong run, early morningThe key session of the week, at about race-morning start time
SunRest, optional easy walkRecover from the long run

Why five short days instead of four longer ones: the group runs are a standing commitment with friends, and I wanted to keep them. Running is also how I clear my head. Spreading the miles over more, shorter runs keeps each run easy on my knee. The early Base weeks used four run days, and the fifth came in on Oct 31 when the group runs started.

Hard and easy days alternate. Nike’s plans say to space out hard efforts like speed runs and long runs, with recovery runs and rest days in between. The only hard days in my week are Saturday’s long run and some Wednesdays, with easy days around both.

Extra activity counts. A long dance night or an unplanned hard workout can take the place of a planned run, and a skipped run comes off the week’s total instead of being made up later. What matters is the total load on my body, not what the calendar called it.

Lifting stays. Strength work happens at midday, not right after runs. A 2014 review of randomized trials found strength training programs cut sports injuries to about a third, pooled across many sports, while stretching showed no benefit (British Journal of Sports Medicine). It doesn’t prove my particular split protects my knee, and heavy leg work costs recovery, but it’s the main reason lifting stays in the plan. Leg day stays out of the two days before a long run, and leg sets get lighter in the half’s race week, the January peak and the taper.

The build

Mileage and long runs

How it builds. Weekly miles rise from 13 to 36, and the long run grows by about one mile a week early on, then 1 to 2 miles at a time later. Steady, small steps are the main protection for my knee. The one study I’d point to followed 874 beginners for a year. Those whose weekly distance jumped more than 30% had more “distance” injuries, the kind that include knee pain and IT band trouble, than those who grew less than 10%. The difference wasn’t statistically certain, and the authors called it exploratory, but they suggested beginners keep increases under 30% over two weeks (Nielsen and others, 2014). I treat 30% as a reason to look closer, not a hard line.

Lighter weeks on purpose. The Nov 14, Jan 2 and Jan 23 long runs are cutbacks. The Dec 5 long run is short because the half is the next Saturday, and the Dec 19 long run is easy so I recover from it. These give my body time to absorb the training before the next step up. The B.A.A. and Nike plans also drop the long run back every few weeks before building again (Nike’s weeks 4, 7 and 13, for example).

The peak and the taper. The longest run, up to 20 miles, is on Jan 30, four weeks before Cowtown. Long runs then step down 14, 10 and 6. That timing comes from the training guides:

PlanBiggest long run2 weeks out
Nike Marathon plan18 to 22 mi, 4 weeks out12 to 14 mi
B.A.A. Level 1 (novice)15 to 18 mi, 3 weeks out13 to 14 mi
B.A.A. Levels 2 and 318 to 20 mi, 3 weeks out13 to 15 mi
My planUp to 20 mi, 4 weeks out10 mi

I went with Nike’s four weeks rather than the B.A.A.’s three because, at my pace, a 20-miler is over four hours on my feet and needs more time to recover from.

The 20 has to be earned. It happens only if the Jan 16 18-miler passes five checks: the effort stayed easy, fueling worked, my knee never changed my stride, walking and stairs were normal within 48 hours, and the next easy runs felt normal. If not, Jan 30 is the longest distance I recovered well from, not automatically another 18. Either way, the Jan 30 run ends at 4:00 elapsed, turning around at 2:00. At 12:00 a mile or faster that’s the full 20. On a slow or hot day the cap stops me around 18, which the B.A.A. novice plan treats as a full peak, and that’s fine.

Effort

Mostly easy, with a little faster running

Easy really means easy. Most of my runs are at a conversational effort. Both big guides define easy the same way: the B.A.A. calls an easy run “conversational,” and Nike rates efforts out of 10, with recovery runs at the bottom. For me that means:

Honestly, this is where I most need to improve. In the four weeks before Nov 12, only about half my easy runs stayed under my heart rate line.

Run/walk. Most runs use intervals on a timer (run, then walk, with the run never shorter than the walk): 2:00 / 1:00 on cool days, 1:30 / 1:00 on hot days, and 2:00 / 1:00 on race day. This is Jeff Galloway’s run-walk-run method, and his starting ratio for runners at 10 to 12 minutes a mile is 2:00 run, 1:00 walk. He says to change the ratio only in 2 to 3 week blocks, and only once long runs feel strong. His runDisney plans also suggest shorter pieces at the same ratio, like 1:00 run / 0:30 walk, which he says cause less fatigue. I tested that on the Nov 21 long run and didn’t like it. It felt like I was pushing a lot harder than at 2:00 / 1:00, so I’m sticking with 2:00 / 1:00.

Long-run pace and heat. Galloway says long runs should be at least 2 minutes a mile slower than the marathon pace his Magic Mile formula predicts. With my 7:45 Magic Mile and his runDisney marathon formula, that works out to about 12:50 a mile at 60°F. In heat, his runDisney plans say to slow about 1 minute a mile at 70°F and another minute at 80°F. I adopted that rule on Nov 12. My recent long runs have been around 11:00 a mile, so by his standard they’re on the quick side. That’s one more reason effort, not pace, is the guide on long runs.

Hills. Short hill repeats (30 to 60 seconds uphill at about 7 out of 10 effort, easy back down) fold into the Wednesday run every two to three weeks. San Antonio was hilly and I walked a lot of it, and Cowtown is rolling, so hills got a regular spot. The B.A.A. novice plan uses hill repeats the same way, as a regular Tuesday session. There are no hills the week after the half, the week of Jan 9, or the week of the 20.

Fartlek. Two sessions, Dec 30 and Jan 27, of 7 rounds of 1 minute hard and 2 minutes easy. This is Nike’s “One Hard Two Easy” workout. Both parts are running, with no walking for those 21 minutes.

Race rehearsals. The last 3 miles of the Jan 9 long run and the last 3 to 4 on Jan 16 are run at Cowtown race pace (13:30 to 13:45 a mile) with my race-day run/walk and fueling. That’s slower than my usual long runs, so these are practice at holding back, timing walk breaks and eating on the move, not speed work. It’s a beginner version of the B.A.A.’s “marathon simulations,” which put several miles at marathon pace inside the long run.

Why so little speed work compared with the guides? Even the B.A.A. novice plan has 2 or 3 faster sessions most weeks, and Nike’s plans have speed runs from week one. Mine has about one a week at most. That’s deliberate: my knee, my weight and a finish-first goal all point toward keeping most of the stress in the long run.

Fueling. Before the January long runs, I’m writing a fueling plan in grams of carbs per hour, starting around 30 to 45 and working toward 45 to 60 if my stomach handles it, with bottle sizes and refill spots. “Every 30 to 45 minutes” tells me when, not how much.

Check-ins

Magic Miles

Three timed miles on a local track check whether the training is working. Results live on the Road to 26.2 page.

Magic MileDateEffortResult
#1Nov 9A controlled baseline, a little faster than easy7:45, in rain, two days after being sick
#2JanuaryHarder: OK to huff and puff in the second halfComing
#3FebruaryStrong but not all-out, since it’s in the taperComing

Where it comes from: the Magic Mile is Jeff Galloway’s test. You warm up, run one timed mile, and use the time to estimate race pace. His runDisney plans multiply the mile time by 1.3 for half marathon pace and 1.4 for marathon pace. An older article of his uses 1.3 for the marathon, which comes out faster, so on Nov 12 the site switched to the runDisney numbers.

How I use it: as a progress check, and for a sanity check on goals. Galloway’s numbers from #1 (a 2:12 half, a 4:44 marathon) are a perfect-day ceiling for a first marathon, not a goal. Each Magic Mile is a different effort on purpose, so a change in time isn’t only fitness, and I set Cowtown pace mostly from the December half and my January long runs, not from one fast mile.

A race inside the build

The Dec 12 half

The half lands in week 3 of the marathon block. It’s a real race with a short taper, and a light week after it. My race-day plan has the lap-by-lap details.

WeekLong runWhy
Nov 27 (Fri)11 miLast long run before the half, 2 weeks out
Dec 57 miRace-morning rehearsal: same breakfast, gear and fuel, with the last 2 miles at half goal pace (about 12:20 a mile)
Dec 1213.1, the raceGoal 2:40 to 2:45 at 2:00 run / 1:00 walk. Hold the 2:45 schedule through lap 6, then pick up a little only if everything feels good. No lifting the Friday before
Dec 1912 to 14 mi12 easy, or up to 14 if walking, stairs, both group runs and my knee feel normal by Thu Dec 17. No race-pace finish, no hills that week
Dec 2615 to 16 mi15 easy, or 16 if Dec 19 went fine

Why Dec 5 is only 7 miles: every guide goes short the week before a race. Nike’s half plan goes no longer than 7 miles in its last two weeks, and the B.A.A. novice plan’s last long run, 8 days before the marathon, is 5 to 7 miles.

The week after: a PR attempt takes real recovery, even with run/walk keeping the pounding down. The decision about Dec 19 is made by Thursday Dec 17, not in the middle of the run.

Changes

What changed, and why

The plan has been checked three times so far, and the way weeks are counted changed once. Here’s every change, with the before and after.

Nov 7: checked against the B.A.A. and Nike plans

Most of the plan already matched. Three things didn’t, and I changed them:

ChangeBeforeAfterWhy
Marathon peak20 mi on Jan 16, 6 weeks out, then 14, 12, 10, 6, 020 mi on Jan 30, 4 weeks out, then 14, 10, 6Nike peaks 4 weeks out and the B.A.A. 3 weeks out, and both keep 12+ mile long runs until about 2 weeks out
Dec 5 long run10 mi7 miEvery guide goes short the week before a race
Faster runningHills and race-pace finishesPlus two fartleks (Dec 30, Jan 27)Every guide has at least one faster session a week, even for beginners

Was the old plan wrong? Not wrong, just cautious. Its taper started early and ran long, which meant my longest run would have been well over a month old by race day. Every guide I read does it the other way.

Nov 12: checked against Jeff Galloway’s runDisney plans

This review didn’t change the weekly mileage or long runs. These did:

ChangeBeforeAfterWhy
Magic Mile estimatesHalf = mile time × 1.2, marathon × 1.3Half × 1.3, marathon × 1.4The multipliers in Galloway’s runDisney plans. My #1 estimates went from 2:02 / 4:24 to 2:12 / 4:44
Running in heatHeart rate as the guideAlso about 1 min a mile slower at 70°F, another minute at 80°FGalloway’s runDisney heat rule. Houston is warm even in winter
Nov 21 long run2:00 / 1:00Tested 1:00 / 0:30, back to 2:00 / 1:00Same ratio in shorter pieces. It felt like I was pushing a lot harder, so I’m sticking with 2:00 / 1:00
Cowtown 5KEasy 1:1 run/walkWalk-heavy is fine, even all walkingGalloway’s advice for back-to-back race days is to mostly walk the first one
Leg dayAnywhere in the lifting splitNot in the two days before a long runFresh legs for the key run of the week

What I didn’t change: the runDisney plans use three run days a week. I’m keeping five. Running is how I clear my head and work through a hard day, and the short weekday runs are easy on my knee, so they stay.

Nov 12: an outside audit

I had ChatGPT audit this site and my planning documents. It found real problems with the weeks after the half and with the 20-miler. I checked each point against the plan and the sources, and agreed to these changes:

ChangeBeforeAfterWhy
Dec 19 long run15 mi with a race-pace finish12 easy, up to 14 if recovered by Thu Dec 17A PR attempt needs real recovery
Dec 236 mi with hills6 easy, no hillsSame
Dec 26 long run16 mi with a race-pace finish15 easy, up to 16 if Dec 19 went fineSame
Jan 30 peak20 if Jan 16 “goes well,” else another 1820 only if Jan 16 passes five checks, else the longest distance I recovered well from. Ends at 4:00 elapsed (changed from 4:30 the same night)A clear rule beats a gut call at mile 15
Feb 36 mi with hills6 easy, no hillsIt sits between the 20 and the 14
Jan 137 mi with hills7 easy, no hillsThe Jan 9 week already had the biggest long-run jump, a race rehearsal and +23% weekly miles
Easy pace“10:00 a mile or slower” on one page, 11:30 to 13:00 on anotherOne rule: talk in full sentences, heart rate under about 155, usually 11:30 or slower10:00 was only seconds off my 10K race pace
Cowtown goalEight different times across the site, and a 6:00 “stretch” inside a 5:54 to 6:07 planThe goal is to finish. Aiming for under 6:00 at 13:30 to 13:45 a mile. Reach: about 5:00One clear set of numbers
Knee“Back off” if it talksMild and steady: walk and reassess. Builds or changes my stride: end the run. Swelling, locking or pain into normal walking: stop building and get it checkedEasier to follow when I’m tired
Extra activityCounted as bonus, never replaced a runCan replace a planned run, and skipped runs come off the totalTotal load matters, not the label
Fueling“Every 30 to 45 minutes”A written plan in grams of carbs per hour, ready before Jan 9Timing alone doesn’t say how much
Dec 12 pacingPick up toward 2:40 after lap 4Hold the 2:45 schedule through lap 6, then a small pickup if everything feels goodSpeeding up at halfway would mean running faster than even 2:40 pace for 6.9 miles

The audit also caught wording I fixed: my “race-pace finishes” are really race rehearsals, since Cowtown pace is slower than my long runs, and the site’s marathon projection leans optimistic because it only counts training runs that point faster.

Nov 21: weeks now run Monday to Sunday

No runs changed. Only the way weeks are counted did:

ChangeBeforeAfterWhy
Training weeksSaturday to Friday, opening with the long runMonday to Sunday, with the long run near the endStrava, Apple Health and the published plans I compared all count Monday to Sunday, so my weekly miles now match what Strava shows
Weekly totalsPeak week 35 miPeak week 36 miEach week now holds a different set of days, so some totals moved by a few miles. The long runs, their dates and the plan’s total miles are the same
Phase datesBase from Oct 17, marathon block from Nov 28, taper from Feb 6Base from Oct 12, marathon block from Nov 23, taper from Feb 1Each phase starts on the Monday of the week holding its first long run

Being honest

Where judgment decided, and known limits

The published plans are written for typical runners. Some choices here are fitted to me rather than taken from a source:

And the honest limits:

Receipts

Sources