The founding story
I bought a $45 case based on a review published 11 days after the phone launched. The hinge failed at week nine.
The review I read was thorough in every way that didn’t matter for longevity. It covered the cutout alignment, the raised lip clearance, the texture grip, the exact gram weight. What it didn’t cover — couldn’t have covered, published at day eleven — was what happens to the hinge mechanism on a folio-style case after it’s been opened and closed roughly six times a day for two months. Mine developed a lateral flex by week seven and separated cleanly at the spine by week nine. The review still showed up at the top of search results for that case. Nothing in it was wrong. None of it was useful for what I actually needed to know.
I started testing differently. What “differently” meant from the beginning: a hard minimum carry period of eight weeks before anything gets written, all units purchased at retail from consumer channels with the receipt stored alongside the notes, and environments logged separately — front denim pocket, daily bag, car mount, desk. The first review I published under those rules was a Bluetooth earbud. At six weeks the left bud started producing an intermittent crackling during high-frequency audio, consistent across multiple source devices. The problem wasn’t in the drivers or the codec. It was moisture ingress at the charging contact. That failure mode doesn’t exist at week two. It barely exists at week four. A standard review cycle wouldn’t have found it.
PocketSpec now covers phones and eleven accessory categories — cases, cables, chargers, car mounts, earbuds, screen protectors, power banks, stands, wallets, cleaning gear, and styluses. I’ve completed carry testing on 214 products since launch. The team has grown to three regular testers for accessory categories, though phones still go through one pair of hands. What hasn’t changed: the 8-week floor, retail-only sourcing, and the failure index — a public log of every product that degraded or failed after its original verdict. That index currently has 47 entries. It’s the part of this site I’d want to exist if I were a reader.
Mission and Vision
Mission
Test things long enough to get bored of them, then report what didn’t hold up.
In practice, that means no review ships before the product has been carried daily through at least eight weeks and at least one documented stress event. It means distinguishing between a product that passed initial tests and a product that performed consistently over a realistic ownership window. It means the word “durable” only appears in a review when we can attach a timeframe to it.
Every review in the database names at least one specific failure mode or compromise, even for products that earned a recommendation. Products that perform well across every documented test still ship with conditional verdicts — the conditions under which the recommendation holds. “Works well for most people” is not a useful sentence. We don’t write it.
Vision
Every carry-tested failure mode documented publicly, permanently, and linked from the product that caused it.
Five years from now, a $28 charging cable’s connector durability under daily-carry conditions should be as findable as its wattage rating. The manufacturer publishes the wattage. Nobody publishes the failure timeline. That’s the gap. The failure index is an attempt to start filling it from the consumer side, one tested product at a time.
Getting there requires that every verdict stay accurate after publication — not just on the day it went live. It requires that failure reports get indexed alongside the original reviews, not buried in comments. It requires that readers who made a purchase decision based on a review from eight months ago can check today whether that verdict has changed. We maintain the infrastructure to do that. It’s not glamorous work. It’s the only work that matters here.
Operating Principles
How we run this.
Policy 01
“The 8-Week Floor”
Every review has a hard minimum carry period of eight weeks before it publishes. Not a target, not a guideline — a cutoff. The policy exists because most observable failure modes don’t surface until week five or six: coating fatigue, connector loosening, hinge creep, seal compression. A review published at day fourteen is reporting on a product that hasn’t been stressed yet. We don’t know what those reviews don’t know.
Policy 02
“Retail Only”
Every product in the database was purchased at full retail price from a consumer channel — no manufacturer samples, no seeded units. Samples sent for review purposes arrive configured for short-window reviewer use; they are not the product that ships in volume six weeks into a product cycle. Retail sourcing also decouples our review calendar from launch windows entirely. We review what’s currently on the shelf, not what’s being announced this week.
Policy 03
“The Failure Index”
A public, timestamped log of every product that degraded or failed after its original verdict. Every review links to its current failure log status. When something that passed testing fails at month seven, the log entry is updated, the verdict badge changes, and anyone who bookmarked the review sees the current state. The failure index is the part of this site that nobody else publishes, and the part that’s most likely to actually protect someone from a bad purchase.
Policy 04
“Spec vs. Field”
Published specifications describe controlled conditions. Daily carry is not a controlled condition. Every review explicitly documents where the two diverge. A charger rated at 67W that only sustains that rate for the first 12 minutes of a charge cycle is not delivering 67W charging in practice — it’s delivering 67W for 12 minutes and something slower for the remaining 40. That distinction belongs in the review. We put it there.
Policy 05
“Conditional Verdicts”
No unconditional recommendations. Every verdict ships with the specific conditions under which it holds — user profile, use case, durability threshold, price sensitivity. Two readers with different needs get different answers from the same review. A cable that earns a recommendation for someone who replaces cables annually might be wrong for someone running USB 3.x peripherals. Those are different verdicts. They go in the same review.
Policy 06
“Version Control”
When a product changes after publication — hardware revision, firmware update, supply chain substitution, price change — the review is updated with a dated changelog entry. The original verdict is preserved as a historical record. Readers see both the current assessment and the full delta: what changed, when it changed, and whether it affected the verdict. A review from 14 months ago that’s still sitting at the top of search results should say so clearly. Ours do.
The Process
From the retail shelf to the published review.
Sourcing
Three things get something into the review queue: reader requests tracked in a public queue (the queue is live and sortable by category), category gaps — when a segment has no currently reviewed option under a certain price point, it moves up the priority list — and personal need, which I’ll be direct about: it’s the most honest trigger. If I need to buy a car mount and there’s nothing in the database, that product gets tested. All three triggers result in the same action: a retail purchase logged with the sourcing date, purchase channel, and price paid.
The purchase price is listed in the final review — not the current price, not an affiliate price, the price on the receipt. The sourcing decision log is public. If you want to see what’s in the queue and why something hasn’t been reviewed yet, that’s where you look. Products don’t skip the queue based on margin, release timing, or brand familiarity.
Carry Phase
Minimum eight weeks from first carry date. Environments are documented separately and require coverage before testing is considered complete: front denim pocket (key contact, body heat, compression), daily bag (varied movement, dust accumulation across different bag materials), car mount (vibration, direct sun exposure, thermal cycling), desk use (connector cycling, surface scratch accumulation). Usage is logged daily in a private notes file — specific timestamped observations, not general impressions. “Felt loose by today” is not a note entry. “Connector-to-port engagement requires more lateral pressure than week one — approximately 15% more force to seat fully” is.
“Completed testing” requires satisfying all four of the following: minimum eight weeks elapsed, all four environments covered with documented observations, at least one documented stress event (rain exposure, drop from desk height, heavy bag compression, extended heat exposure), and a final physical inspection — connector pins, hinge mechanisms, coating surfaces, port covers. Products can fail to complete testing requirements and get held without publication. That happens.
Writing
The review structure follows a fixed sequence: manufacturer specifications (labeled as such, taken from official product pages with version date), observed behavior (what the daily log actually showed), divergence table (where observed behavior differed from published specs, with magnitude noted), failure log status (current entry or a clean entry if no issues found at publication), and verdict with explicit conditions. The verdict section answers three questions: who should buy this, who shouldn’t, and what would change this recommendation.
A review draft clears for publication when all four criteria are met: the failure log entry exists, the verdict has explicit use-case conditions rather than general praise, at least one specific failure mode or limitation is named by name and timeline (not “may show wear over time” — “paint separation at the volume button recess appeared at week five under front-pocket carry”), and the testing documentation file is complete and archived. Drafts that don’t meet criteria don’t get pushed to meet a schedule.
Maintenance
Reviews are revisited at scheduled six-month intervals regardless of reader activity. Any product with a reported failure in the comment queue triggers an immediate out-of-schedule review. The update sequence runs in order: failure log entry updated first, then verdict badge, then body copy, then subscriber notification. The sequence matters — the failure log is public-facing and indexed, and it should reflect the current state before any other part of the review does.
A product that received a recommendation and later failed gets a “VERDICT CHANGED” banner at the top of the review — not an edit note below the fold, not a parenthetical appended to the original conclusion. Readers who reach that review from a search result see the current verdict before they see the original one. That’s the intent. When a product changes — hardware revision, component substitution, firmware that affects measured behavior — the changelog is dated and preserved. The original verdict stays in the record. You can see what we said, when we said it, and whether it still holds.
The failure log is public. The testing is ongoing. Pull up a chair.
If you want to know what’s in your pocket before you buy it, you’re looking at the right place.