Desk Work, Back and Neck Pain: What Actually Works

By mid-afternoon the right side of your neck has tightened into a band that does not let go until you sleep, or your low back has settled into a dull ache that keeps you shifting in the chair. You already bought the ergonomic chair. You already bought the standing desk. It still aches — and underneath the ache sits a quieter worry: that this is simply what a body at a desk becomes, and that you have been paying for furniture instead of a fix. That suspicion is more right than the product pages are.

Because here is what the evidence actually says, and it is oddly liberating: sitting is a weaker culprit than you have been told. The pooled associations are real but modest and almost entirely cross-sectional — photographs, not films — and a snapshot cannot say whether sitting produced the pain or the pain produced the sitting. So the question this article answers is not which chair to buy; it is which minutes of your day carry randomized-trial support for desk-related neck, shoulder and back pain. The most defensible answer is unglamorous: about ten minutes of high-intensity neck and shoulder resistance work on most days, walking accumulated toward a real weekly total, and enough consistency to hold the pattern for three months.

What this will not do: diagnose you, choose your chair, or serve as a medication guide. Most of these trials are small and short, and where a study did not produce a number, this article describes the direction of the finding instead of inventing precision. It will also name the symptoms that mean you close this tab and get assessed today.

The uncomfortable part: sitting is a weaker culprit than you have been told

A 2021 meta-analysis in Health Promotion Perspectives pooled 27 cross-sectional studies and found that in adults, a sedentary lifestyle was associated with low back pain at an odds ratio of 1.24 (95% CI 1.02 to 1.50). Prolonged sitting time was 1.42 (1.09 to 1.85), and among office workers specifically the association was 1.23. For scale, the same analysis put excess weight at 1.35 and smoking at 1.28. Driving was the strongest sedentary exposure at 2.03 (1.22 to 3.36).

A second 2021 review, in the International Journal of Behavioral Nutrition and Physical Activity, screened 178 studies and included 79. Self-reported workplace sitting was associated with low back pain at 1.47 (1.12 to 1.92) and with neck and shoulder pain at 1.73 (1.46 to 2.03), full-day sedentary behavior with low back pain at 1.19 (1.03 to 1.38), and sitting was not significantly associated with pain in the extremities, at 1.17 (0.65 to 2.11). Then the sentence most summaries skip: evidence of prospective associations was insufficient, and the authors state that reverse causality cannot be ruled out.

Read that carefully, because it changes what you should do. Almost all of the human data linking your chair to your back is cross-sectional. It is a photograph, not a film. People with back pain sit more, and people who sit more report back pain, and a snapshot cannot say which came first. The associations that exist are modest, in the same range as being overweight, and in the one review that measured both, the association was stronger for the neck and shoulder (1.73) than for the low back (1.47).

The consequence: subtraction is the wrong model. You will not solve this by removing hours from a day that structurally requires them. The interventions with trial support are additive.

What actually changes when you train

The most informative mechanism study here is small and specific. Thirty female office workers with chronic neck and shoulder pain were randomized to high-intensity elastic resistance training for two minutes a day for ten weeks, or to weekly emails about general health. Electromyography recorded the splenius and upper trapezius across a normal workday. Training increased isometric strength by 6% and reduced neck and shoulder pain intensity by 40% versus control. The interesting part is the muscle behavior: in the hours right after a session, muscles were more tense with fewer complete relaxation periods, but by ten weeks that reversed, with the average duration of complete motor-unit relaxation periods up 71% and their frequency up 296% during the workday in the splenius (Biomed Research International, 2013).

That is the mechanism worth holding. Your neck does not hurt because a muscle is weak in the abstract. It hurts because a muscle spends an eight-hour workday near a low, continuous, never-quite-releasing level of activation, and training appears to restore its ability to switch off between demands. It also explains why the first week can feel worse. Thirty women, ten weeks, one workplace: a plausible mechanism, not a proof.

A 2018 systematic review of deep cervical flexor training pooled 12 randomized trials and found strong evidence of benefit for neuromuscular coordination, no or only small effects on strength and endurance at higher loads, improved head and cervical posture, and limited or contradictory evidence elsewhere. The reviewers proposed a multimodal program rather than deep cervical flexor work alone (BMC Musculoskeletal Disorders, 2018). Chin nods are an ingredient, not the meal.

For the low back, the Cochrane review of motor control exercise in chronic non-specific pain covered 29 trials and 2,431 participants, with 76.6% at low risk of bias. Motor control exercise beat minimal intervention for pain at every follow-up with medium effect sizes (long term, mean difference -12.97; 95% CI -18.51 to -7.42), was not clinically better than other forms of exercise at any point, and gave outcomes similar to manual therapy at moderate-to-high quality evidence. The reviewers concluded the choice should come down to preference, cost and safety (Cochrane, 2016).

So the active ingredient is not a special muscle or a magic cue. It is dose. A 2024 Bayesian network meta-analysis of 82 trials and 5,033 participants with chronic low back pain found a U-shaped dose-response, with the maximum significant response at 920 MET-minutes (standardized mean difference -1.74; 95% credible interval -2.43 to -1.04) and a minimal clinically important difference reached at 520 MET-minutes per week, at very low to moderate certainty; Pilates showed the most pronounced effect (JOSPT, 2024). MET-minutes multiply intensity by time. The point is the shape of the curve: too little does nothing, there is a middle range where the effect is largest, and beyond it more is not better.

Posture: what the forward-head-posture review can and cannot say

Worth being precise rather than smug. A 2019 systematic review and meta-analysis found 15 eligible cross-sectional studies. Ten compared forward head posture in asymptomatic people versus people with neck pain and found a significant between-group difference (mean difference 4.84; 95% CI 0.14 to 9.54). Eight found significant correlations between the head-posture measure and neck pain intensity (r = -0.55; 95% CI -0.69 to -0.36) and disability (r = -0.42; -0.54 to -0.28) in adults and older adults. In adolescents the between-group difference was not significant (-1.05; -4.23 to 2.12), and no association was found with most neck pain measures. The reviewers' own summary: adults with neck pain show more forward head posture than asymptomatic adults, and age was an important confounder (Current Reviews in Musculoskeletal Medicine, 2019).

Posture is not nothing. But note what that review is made of: fifteen snapshots. None can say whether the posture produced the pain, the pain produced the posture, or age produced both. The adolescent null is the tell: if forward head posture caused neck pain by geometry alone, teenagers with phones would be the loudest signal in the data, and they were not.

Practically: stop treating posture as a position to hold and start treating it as a variable to vary. Nothing here supports holding a corrected posture for eight hours as treatment. Deep cervical flexor training did improve head and cervical posture, which is a fine byproduct, but what moved pain in the office-worker trials was loading and frequency.

Your standing desk: what it actually bought you

The 2019 Cochrane review of workplace interventions to increase standing or walking covered ten studies and 955 participants, all from high-income countries. Sit-stand workstations versus no intervention produced no significant difference in low back symptom intensity in the short term (SMD -0.35; 95% CI -0.80 to 0.10), and no significant reduction in pain-related disability at medium term (mean difference -0.4; -2.70 to 1.90; one trial). Activity trackers produced no significant difference in low back, upper back, neck, shoulder, or hand and wrist symptoms. Multi-component programs produced risk ratios sitting on top of 1.00, with one trial showing reduced pain-related disability at medium term (mean difference -8.80; -17.46 to -0.14). The conclusion: the available limited evidence does not show these interventions reduced musculoskeletal symptoms, at low or very low quality.

The honest counterweight: the 2021 IJBNPA review reported that experimental and intervention evidence, synthesized narratively, did indicate reduced low back, neck and shoulder and general musculoskeletal pain when workplace sitting was reduced. The picture is not uniformly null. It is underpowered and inconsistent, which is not the same thing.

The individual trial shows why. Forty-seven office workers were randomized to a sit-stand workstation for eight weeks or nothing. Versus controls, the intervention reduced sitting by 80.2 minutes per 8-hour workday (95% CI -129.0 to -31.4; p = 0.002), increased standing by 72.9 minutes (p = 0.007) and lowered total cholesterol by 0.40 mmol/L (p = 0.049), with beneficial changes in flow-mediated dilation and diastolic blood pressure and no harmful change in musculoskeletal discomfort (BMC Public Health, 2015). The participants were asymptomatic. A trial recruiting people without pain cannot demonstrate pain relief; the most it can show is that the desk caused no discomfort, which is what it showed.

Verdict: a sit-stand desk is a defensible behavioral purchase whose cardiometabolic case rests on one eight-week trial in 47 asymptomatic workers, with cholesterol at p = 0.049 and authors who framed even that as conditional on the habit continuing. It is a poor bet as a treatment for an aching neck or back. If cardiometabolic risk is your real reason for standing, measure it rather than assume it, with the Biomarker Tracker. Keep the desk. Stop expecting it to be the intervention.

The protocol

Here is what has trial support in exactly your population. Four findings set the design.

One hour a week works, however you slice it. A cluster-randomized trial allocated 447 office workers at the cluster level to 1 x 60, 3 x 20 or 9 x 7 minutes a week of supervised high-intensity neck and shoulder strength training for 20 weeks, or to a reference group. Among the 256 with baseline pain of at least 3 on a 0-to-9 scale, all three schedules significantly reduced neck pain versus reference (p < 0.01). From a baseline of 3.2 (SD 2.3), reductions were 1.14 (95% CI 0.17 to 2.10) for the single hour, 1.88 (0.90 to 2.87) for three sessions and 1.35 (0.24 to 2.46) for nine short ones. Work disability improved only in the 1 x 60 and 3 x 20 groups (BJSM, 2012).

Ten minutes daily is as good as twenty. Thirty people with mild-to-moderate neck and shoulder pain did an eight-week control period then eight weeks of either 10 or 2 x 10 minutes of high-intensity neck and shoulder resistance training, four exercises, five days a week. No differences between doses. Merged, mean pain fell 25% and worst pain 43% (p = 0.05 and p < 0.01), with quality of life up 10.6%; measured strength did not change (BMC Sports Science, Medicine and Rehabilitation, 2020).

You do not need supervision. A 20-week trial of 351 office workers compared training three times weekly with ongoing supervision, the same program with minimal initial supervision, and a reference group. Neck pain over the last seven days fell in the minimally supervised group versus reference (-0.5 +/- 0.2; p < 0.02), with a trend in the supervised group (-0.4 +/- 0.2; p < 0.07). Headache intensity fell by 1.1 +/- 0.2 in both training groups (p < 0.001). There were no differences between the training groups (Biomed Research International, 2014).

Strengthening is the part with the evidence; stretching is an add-on. A 2024 meta-analysis of eight randomized trials in office workers with chronic neck pain found significant reductions in pain intensity and disability with strengthening of the neck, shoulder and scapular muscles versus control (p < 0.01), while stating that all eight trials were at high risk of bias and overall certainty was low (Applied Ergonomics, 2024). Separately, 96 people with neck pain of at least 5 out of 10 for three months or more got an ergonomics brochure alone or the brochure plus stretching twice a day, five days a week for four weeks. The stretching group improved more: -1.4 points on the visual analogue scale (95% CI -2.2 to -0.7), -4.8 on the Northwick Park questionnaire and +14.0 on the SF-36 physical dimension. Those exercising at least three times a week improved more than those doing less on neck function and on the physical dimension of quality of life, though not on pain (Clinical Rehabilitation, 2016).

One honesty note. These abstracts report intensity, session length and weekly frequency, not set-and-rep schemes. The sets-and-reps column below is a practical translation of "four to five exercises, high intensity, ten minutes," and is labeled as such. The frequency and minutes columns are anchored to trials.

The low back half of the plan

The American College of Physicians guideline is the anchor, and it is unusually blunt. For acute and subacute low back pain, because most people improve over time regardless of treatment, non-drug treatment comes first: superficial heat at moderate-quality evidence, massage, acupuncture and spinal manipulation at low-quality. For chronic pain the initial selection is again non-drug: exercise, multidisciplinary rehabilitation, acupuncture and mindfulness-based stress reduction at moderate quality, with tai chi, yoga, motor control exercise, biofeedback, cognitive behavioral therapy and others at low quality. Both are strong recommendations. Medication sits third in the ordering, after an inadequate response to non-drug care, with anti-inflammatories first-line and opioids reserved for people who have failed everything else (Annals of Internal Medicine, 2017). That last part is a conversation with your own clinician, not a step in anyone's protocol.

A 2025 umbrella review synthesized 70 systematic reviews. For secondary prevention, six of ten reviews with meta-analyses indicated a small benefit from general exercise and leisure-time physical activity at low-to-moderate certainty. For management, 35 of 36 reviews reported small beneficial effects on pain and disability versus minimal intervention, mostly short-term, across Pilates, motor control, mixed exercise, tai chi, water-based exercise and yoga. Adverse events were reported in fewer than 31% of reviews, were predominantly post-exercise soreness and temporary pain increases, and none were serious (Journal of Sport and Health Science, 2025). That is the realistic frame: small effects, moderate certainty at best, very low risk.

Walking earns its own paragraph because it is the thing people sustain. A 2022 systematic review of 19 trials and 2,362 participants with chronic or recurrent low back pain found high-certainty evidence of a small effect favoring walking or running over minimal or no intervention for pain in the short term (SMD -0.23; 95% CI -0.35 to -0.10) and medium term (-0.26; -0.40 to -0.13), and also found walking less effective than alternative active interventions. No trial addressed prevention or acute pain (JOSPT, 2022). A 2026 dose-response meta-analysis of 13 randomized trials then quantified the target: pain fell (SMD -0.59; -0.95 to -0.24, low certainty) and disability fell more (SMD -0.69; -1.14 to -0.23, moderate certainty), along a U-shaped curve peaking at 1,000 total minutes of walking spread over a mean of eight weeks for pain, and 900 minutes for disability. Effects looked larger outdoors and with motivational support, though those subgroup differences were not statistically significant (PM&R, 2026).

If running appeals, there is one small trial. ASTEROID randomized 40 adults with non-specific chronic low back pain, mean age 33 (SD 6), half female, to a 12-week progressive run-walk program of three 30-minute sessions a week or a waitlist. Average pain intensity improved by 15.30 points on a 100-point scale (95% CI -25.33 to -5.27; p = 0.003) and Oswestry disability by 5.20 points (p = 0.038). There was no attrition, mean adherence was 70% (2.1 of 3 sessions a week), and nine non-serious adverse events were judged likely study-related, seven of them lower-limb injury or pain (BJSM, 2025). Forty people, aged 18 to 45, unblinded against a waitlist. If you are 55 with a twenty-year desk history, that trial does not describe you. The walking evidence does.

Progressions and regressions

The most useful implementation data here comes from 269 participants keeping training diaries during a 12-week workplace strength program of five neck and shoulder exercises. Reaching a clinically relevant pain reduction, defined as at least 30%, was associated with how well people adhered and progressed, though nobody was randomized to adhere, so the arrow could point either way. At a 70% adherence threshold, a cumulative training volume of about 11,000 kg (a threshold that emerged only in the women) and progressions of one to two times baseline values were significant for clinically relevant pain reduction. Thirty percent of participants were absent for at least two consecutive weeks, with median quitting time around weeks six to eight (Physical Therapy, 2023).

Three rules fall out. Seven sessions in ten is the standard, not ten in ten. The load should climb across the three months: in that diary analysis, the people who reached a clinically relevant improvement had progressed to somewhere between one and two times their starting resistance, and a band that felt hard in week one and still feels the same in week ten has stopped training you. And weeks six to eight are when people quit, so plan for that week rather than being ambushed by it.

When it gets easy. Add resistance before you add time. The frequency and duration comparisons found no advantage to longer sessions, so increase band tension or weight and keep the ten minutes.

When it hurts more. A temporary increase in pain is a documented minor adverse event in the exercise reviews, not evidence of damage. Reduce load rather than stopping, and protect frequency above everything else. In the run-walk trial the adverse events were mostly lower-limb, the predictable result of adding impact to a body with no recent impact history; if that is where your trouble shows up, drop back to walking. If pain is sharply rather than mildly worse, if it travels below the knee or into the hand, or if anything on the red-flag list appears, stop adjusting the program and get assessed.

Is anything in the supplement aisle worth buying for this?

From this evidence base, no. Not one of the 23 studies behind this article tested a supplement for desk-related neck or back pain. That is not a hedge, it is an absence, and on the tiers described on our evidence grading page there is nothing here to grade. The best-supported things in this entire pool are ten minutes of resistance training five days a week and a walking dose you can hit on a lunch break. Both are free.

Topicals, labeled honestly. A menthol and methyl salicylate patch such as the Salonpas Pain Relieving Patch is a symptomatic adjunct, and that is the whole claim. It is a counter-irritant on the skin over a sore area. It does not change muscle activation, it does not build capacity, and nothing in this article's evidence pool tested it, so it is ungraded here rather than graded well. We handle that literature separately in what actually works in topical pain relief and in our menthol gel review. If a patch makes the last stretch of the workday tolerable enough that you still do your ten minutes, it has earned its place. If it becomes the plan, it has replaced the thing that works. Superficial heat is the one comfort measure carrying a moderate-quality rating in the ACP guideline for acute low back pain, and it costs almost nothing.

What to skip

Buying your way out. The Cochrane review covered sit-stand workstations, treadmill workstations, activity trackers and multi-component programs and found no significant reduction in musculoskeletal symptoms at short, medium or long term, at low or very low quality, the one exception being the single trial's reduction in pain-related disability noted above (Cochrane, 2019). Equipment changes the environment; it does not load the tissue. If you are going to buy anyway, buy with clear eyes, which is what our buyer's guides are for.

Trackers as a pain intervention. In that same review they produced no significant difference in any body region. Fine for measuring your walking dose. Not treatment.

Posture perfectionism. Fifteen cross-sectional studies, an age confounder, and a null result in adolescents do not add up to a treatment. Holding a corrected position all day is effortful, unpleasant and unsupported by anything in this pool.

Believing you found the one correct exercise. Motor control exercise was not clinically superior to other exercise at any follow-up in the chronic review. In acute low back pain, Cochrane found only three small trials totaling 197 participants and drew no firm conclusions, with no benefit over spinal manipulation, other exercise or medical management. One very low-quality finding suggested adding motor control exercise to medical management reduced one-year recurrence by 64%, and the reviewers explicitly said whether it prevents recurrence remains uncertain (Cochrane, 2016). Treat that 64% as a hypothesis, not a reason to pick a program.

Passive modalities instead of loading. The chronic review found motor control exercise clinically more effective than exercise combined with electrophysical agents for pain, disability and quality of life with medium to large effect sizes, though at very low to low quality. Machines applied to you are not a substitute for work done by you.

Waiting until it stops hurting to start. The ACP framing for acute pain is that most people improve over time regardless of treatment. Resting until you feel ready mostly means taking credit for time passing.

Imaging: why an early scan usually makes things worse

A 2021 systematic review asked whether early MRI, within the first four to six weeks, for acute low back pain without red flags is associated with longer disability. Seven studies met inclusion, all of good methodological quality, and all consistently reported longer disability in the early-MRI groups. Three retrospective cohorts reported a higher mean length of disability ranging from 9.4 days (95% CI 8.5 to 10.2) to 13.7 days (13.0 to 14.5) at one year. The others reported hazard ratios of work disability from 1.75 (1.23 to 2.50) to 3.57 (2.33 to 5.56) (BMC Musculoskeletal Disorders, 2021).

A 2013 retrospective cohort in Spine is the most detailed. From a nationally representative sample of workers with acute, disabling, occupational low back pain, cases where early MRI might have been indicated were deliberately excluded, leaving 555 people. Even so, 37% of the non-specific cases and 79.9% of the radiculopathy cases had an MRI within 30 days. The early-MRI groups had much lower rates of coming off disability and, on average, $12,948 to $13,816 higher medical costs. Even in a subgroup with minimal disability impact, costs ran $7,643 to $8,584 higher. The authors concluded that non-indicated early MRI has a strong iatrogenic effect regardless of radiculopathy status.

Two caveats, because this gets oversold. These are retrospective cohorts in workers' compensation populations, graded level 3 evidence, and association is not causation even after adjustment. But the direction is consistent across every included study, and the mechanism is not mysterious: a scan of an ordinary adult spine reliably finds disc bulges and degenerative changes that are also present in people with no pain, and once those words are in your file they are in your head.

When imaging is indicated. Red flags, always and immediately. Beyond that, the guidelines cited in this literature indicate MRI for persistent radicular pain after about a month of conservative management, in people who are candidates for surgery or epidural steroid injection. Imaging is for answering a question that changes what happens next. If nobody will act differently on the result, the scan is not information. If you want the framework for telling nerve pain from joint pain before deciding anything, that is what Pain Care, Explained is for.

Red flags: stop and get assessed

Everything above assumes ordinary mechanical neck or back pain in someone otherwise well. The following are not that. There are three urgency tiers below, and a non-urgent one at the end, and the tier is the instruction: emergency now, same day, or days rather than weeks. If something of yours appears in two tiers, act on the faster one.

Tier one, emergency now. Go to an emergency department today, without waiting for a callback, if you have any of these:

• New numbness or altered sensation in the saddle area: inner thighs, buttocks, genitals or perineum. People often describe it as feeling wrong when wiping or when sitting on a hard chair.
• New difficulty starting or stopping urination, inability to tell when your bladder is full, urinary retention, or leaking without knowing.
• New bowel incontinence, or loss of the sensation that warns you a bowel movement is coming.
• New loss of genital sensation.
• Weakness or numbness that is progressing, particularly in both legs.

That cluster is cauda equina syndrome, compression of the nerve roots at the bottom of the spinal canal. It is a surgical emergency. Decompression is time-critical, and delay carries the risk of permanent loss of bladder, bowel and sexual function and permanent leg weakness. Do not wait to see whether it settles overnight. Do not book physiotherapy, a chiropractic appointment, or a routine slot next week. Do not accept a few days of rest and see. Go to an emergency department and say the words "I have possible cauda equina symptoms," because those words trigger a specific pathway and an urgent scan. If you are wrong, you lost an evening. If you are right and you waited, that loss is not recoverable.

Still tier one, emergency now, for different reasons. Call emergency services or go straight to an emergency department for any of these:

• Severe, sudden, tearing or ripping pain in the back, chest or abdomen, especially with faintness, sweating or collapse. Call emergency services. That is a vascular emergency, not a musculoskeletal one, and it is the one item on this page where minutes rather than hours decide the outcome. Do not drive yourself.
• Weakness, numbness or unsteadiness that is getting worse over hours rather than over weeks, anywhere in the body. A deficit that is visibly advancing while you watch it is an emergency regardless of which limb it is in.
• Fever, chills or night sweats with spinal pain plus any weakness, numbness, or bladder or bowel change. Treat that combination as a spinal infection compressing the cord or the nerve roots until somebody proves otherwise.
• New spinal pain in someone with a history of cancer, together with leg weakness, numbness, unsteadiness, or a bladder or bowel change. That combination is treated as spinal cord compression until proven otherwise, and it is time-critical in the same way cauda equina is.
• Significant trauma: a fall from height, a collision. That is an emergency department assessment for a possible spinal fracture, not a same-day appointment.

Tier two, same day. Not an ambulance, but not an appointment next week either. Contact a clinician today and expect to be seen today:

• Weakness that is getting worse rather than staying put: a foot that catches on stairs or slaps the floor, a leg that gives way, a failing or clumsy grip, numbness spreading upward or outward. Today, not an appointment next week. If it is advancing over hours rather than days, that is tier one.
• Fever, chills or night sweats with spinal pain, with no weakness, numbness or bladder change alongside it. Higher risk again with a recent infection, a recent spinal injection or operation, injection drug use, or a suppressed immune system. Spinal infection looks exactly like mechanical back pain until it does not, so go the same day rather than watching it for a week.
• A history of cancer with new spinal pain, or unexplained weight loss with new spinal pain, with no neurological symptoms alongside. Contact your cancer team or your family doctor the same day. This is one of the few situations where early imaging is the right move rather than the wrong one.
• A minor fall or an awkward landing in someone with known osteoporosis or on long-term corticosteroids, where a fracture takes very little force.

Tier three, days rather than weeks. Get an assessment booked and keep the appointment:

• Pain that is worse at night, wakes you consistently and is not relieved by changing position. Ordinary mechanical pain almost always has a position that helps.
• New spinal pain in an older adult who has never had it before.

And the non-urgent tier: book a proper assessment, not urgently but do not sit on it: pain radiating below the knee or below the elbow with numbness or pins and needles in a defined strip, morning stiffness that lasts a long time and eases with movement rather than rest, or pain that has not shifted at all after several weeks of doing the work above. None of these are emergencies. All of them change the plan.

How to know whether it is working

Pick your measures before you start, because retrospective memory of pain is unreliable and you will otherwise argue with yourself in week seven.

For function that is not a questionnaire, the five norm-referenced tests in our on-device Movement Lab give you numbers you can repeat at a fixed interval and compare against your age group, which beats asking yourself whether things feel better. Retest on the same day of the week, at the same time, before a training session rather than after.

One last framing. The effect sizes in this literature are small to medium, certainty is mostly low to moderate, and the trials are mostly short. That is not a reason to skip the work. It is a reason to be realistic about what success looks like: a neck that is a nuisance rather than a preoccupation by Friday afternoon, a back that tolerates a long meeting, and a body that stops organizing your week around it. Not the disappearance of every symptom. Anyone selling you that is not reading the same studies. Educational, not medical advice.