Resolution Layer Case Study  ·  Read & Diagnose

Used golf drivers.

A catalog read across a used golf club retailer's drivers — where every used club is one of one, and a club's own page can say two different things about it.

Scope
The retailer's Drivers category: every listing checked for length, flex and shaft names, and 1,999 product pages, drawn at random, read in full
Lead finding
About 1 in 100 sampled club pages name a different model above the club
Issues surfaced
15 reported across 8 themes  ·  36 drivers at a flex no filter matches
Method
Read of the published pages  ·  each club's fields held against each other and against what a driver can be; every count taken again from the pages
Vertical
Used golf equipment — every used listing a single club with its own condition and price
This pass
Read & diagnose — step one of the arc
Prepared by
EKOM
Type
Case study — client anonymized

What this is.

This is a real EKOM catalog analysis, with the client's identity removed. The subject is a used golf club retailer where each used club is listed on its own: one club, one condition grade, one price. EKOM read the retailer's Drivers category. Fifteen findings are reported here. Confirming against source, near the end, accounts for the ones that were not.

A used club online is a record. Each page names the model the club belongs to, gives the club a product name, and carries a spec table — loft, length, shaft, flex, material. Golfers choose a used driver on exactly those specs, and the site's filters and the name the page gives Google are built from them. This read asks whether each club's page agrees with itself.

When every club is a new record, every listing is a chance for one field to say something another does not.

Club names, prices and field values are described rather than quoted here. Counts are as of the day the catalog was read.

What's inside

At a glance.

Nothing was supplied and no internal system was touched. Everything below was read from pages the retailer publishes. The four figures count clubs, not findings.

21
Sampled pages naming
another model — about 1 in 100
33
Drivers under 40 inches,
two at 25 and 26
36
Drivers at a flex
no filter matches
119
Drivers the Driver
subcategory leaves out

Where the fifteen concentrate, and how far they reach

Theme
What it is
Reach
The model named on the page
The label above the club names a different model from the club's own name and Model field
21 of the sampled pages; 18 more labeled with a broader family
Name and spec table
The shaft named one way in the product name and another in the spec table
8 sampled clubs with two shaft weights
Specs a club cannot have
Lengths no driver has; a graphite-only shaft listed as steel; heads with no shaft recorded as graphite
2 at 25–26 inches; 33 under 40, some junior clubs
Values no filter matches
A combined two-flex value; a flex on a shaft maker's own scale; junior age bands offered as colors
36 combined flex; 119 not tagged Driver
Spellings
A shaft maker and a shaft name misspelled in club names
19 and 3 listings
Filters a club falls out of
Ball flight blank on 5 sampled 2025–2026 models; loft blank; a launch filter almost nothing carries
12 sampled with no loft; almost no drivers with a launch value
Reference prices
An original retail price below the used price, or below the listing's own "was" price
4 and 8 of 463 sampled pages with one
What Google is told
The store's own SKU in the barcode field; a platform placeholder as the category
Every sampled page; 2 sampled drivers

How EKOM read this catalog

1
Read
The retailer's public Drivers category, one brand at a time, then 1,999 product pages drawn at random. No file handoff, no credentials, no integration.
Every listing checked; a random sample read in full.
→
2
Analyze
Hold each club's fields against each other and against what a driver can be. Surface the disagreements, rate severity, and tie each to the clubs behind it.
15 findings, 8 themes, 6 mechanisms.
→
3
Check
Every count was taken again from the pages themselves, and cited clubs still for sale were re-opened on the live site weeks later.
Two findings withdrawn, two restated.
Signals from a read of the published pages, checked by an analyst; not a verified defect list. Where two fields disagree, the data cannot say which is right, and some differences may be intentional — which is why each is tied to the values behind it. One category, one day. Used inventory sells daily; the data is a point-in-time read.

The model named on the page.

Every used club's page names a model in the breadcrumb label above the club. On about 1 in 100 sampled pages — 21 of them — that label names a different model from the club's own name and Model field.

Critical  ·  The model named on the page
21 sampled pages name a different model above the club.
On each, the club's name and its Model field agree with each other; the label above them does not. Most name a neighboring model in the same family — a current model labeled with its predecessor, a low-spin head labeled with the standard one — and at least one names another brand's driver. Weeks later, the four cited pages still for sale were re-opened and read the same way. On each, the label links to the club's own model page, and the two checked on the model pages themselves are listed under their own models. The label is the part that disagrees.

A further 18 sampled pages carry a broader family's name — a sub-model labeled with the family it belongs to. That may be deliberate grouping, and it is counted separately.

A golfer comparing used drivers checks one thing before anything else: which club this is. A page that labels one brand's driver with another brand's model contradicts itself on exactly that, for a club that exists once and sells once. Because the club's own name and Model field already agree, the label can be set from them — prepared across every listing and approved as a pattern.

The same club, specified two ways.

A club's product name repeats its loft, shaft, material, flex and length; its spec table holds the same facts as fields. The name is also what the page tells Google the product is called.

High  ·  Name and spec table
On 8 sampled clubs, the name and the spec table give different shaft weights.
The weight number in a shaft's name is how golfers choose between versions of the same shaft. On these clubs, the name gives one version and the table another — a 5 in one place and a 6 in the other, a 50 in one and a 65 in the other. Separately, one shaft maker appears under two names on several clubs: its full name in the table, an abbreviation in the product name.
Critical  ·  Specs a club cannot have
Drivers listed at 25 and 26 inches, and a graphite-only shaft listed as steel.
A standard driver runs about 44 to 46 inches. Across every driver listing, two were listed under 27 inches, 33 under 40 — some of them genuine junior clubs — and four over 48. One club pairs a shaft line made only in graphite with a material of steel, repeated in its name. And all three heads sold without a shaft in the sample still record their shaft material as graphite.

Length, material and shaft are where a golfer narrows hundreds of drivers to a handful, and where a fitting turns into a club. A value outside what a driver can be cannot be caught by the golfer browsing a filter — only by a check on the record as each club is entered. Where the shaft line itself settles a disagreement, the correction can be prepared; where only the club in hand can, it goes to the retailer.

Filters, reference prices and what Google is told.

High  ·  Values no filter matches
36 drivers carry a combined two-flex value that matches neither flex filter.
The site's flex filter offers a fixed list of values. A value outside it matches no filter — so these clubs are missing from both of the filters a golfer looking at them is most likely to set. One sampled club carries a number from a shaft maker's own rating scale in a field whose other values are words. Junior sizing bands appear as colors in the drivers' color filter, and 119 drivers are not tagged with the Driver subcategory.

The smaller items

A golfer who sets a flex and a loft is trusting that every matching driver in stock is in front of them. A value the filter does not know drops the club from that list without anyone seeing it go. Holding each field to its own list, and spelling each maker one way, can be prepared across every listing and applied as each new club is listed.

Confirming against source.

The automated analysis raised seventeen findings on the sampled pages. Fifteen are reported. Both of the others are accounted for here.

Withdrawn  ·  two findings
One flagged new-club listings as having no specs; on those pages the specs sit on the options a golfer chooses, not on the listing itself. The other compared a reference price against values EKOM had worked out from the page, not values the retailer publishes. Both reflected how the pages were gathered, not the retailer's data.
Restated  ·  two findings
Two were real but named the wrong field, and were restated to the field actually at fault: the reference price that sits below the used price, and the shaft material recorded on heads sold with no shaft.
Counted from the pages, and re-opened live
Every count was taken again from the pages themselves, and the case study uses those counts. Weeks after the read, cited clubs still for sale were re-opened on the live site. All but one read as cited; one of the two placeholder categories now names the club's category.
What the data cannot tell
Where a club's name and spec table disagree, the page shows the disagreement but not which side is right — only the club in hand can. Some family-level labels may be deliberate grouping, and some short drivers are junior clubs. Page-level rates come from a random sample and are stated as such.
A finding that names the disagreement without choosing a side is not a weaker finding. It is the one the retailer can act on, because only the retailer has the club.

Where it traces back.

Read together, most of the findings collapse into six likely mechanisms — and a mechanism can be corrected once, where each club's record is made. Each tell below is a pattern in the data; the mechanism is EKOM's reading of it.

A new record for every club, entered one at a time
The tell: small slips — a letter added, letters out of order, a digit off — in fields typed per club.
The model label set apart from the club's model
The tell: the club's name and Model field agree; only the label above them differs, usually a neighbor in the same family.
The same fact kept in the name and in the table
The tell: shaft, flex and length appear twice on every page, and the two copies drift.
Fields without a fixed list of values
The tell: values that do not fit the field — a combined flex, a number where words belong, age bands as colors.
New models added without every attribute
The tell: ball flight filled on older models; where it is blank in the sample, the models are 2025–2026.
Template defaults published as values
The tell: the same value on every page, or the platform's own placeholder.

What this means, and what's next.

A used club exists once. It reaches a golfer through its own page — the model named above it, its name, its spec table — and that page is all a golfer online has in place of the club in hand. In this category, the page sometimes disagrees with itself: one brand's driver labeled with another's model, two shaft weights for one shaft, a driver at 25 inches, a flex no filter knows.
None of this needed access to the retailer's systems. Every finding here came from pages the store already publishes to anyone who visits.

This pass read and diagnosed. The same structural understanding powers what follows — turning a diagnosed catalog into one where every club's page agrees with itself.

1  ·  Prepare the corrections the record already supports
Set each model label from the club's own Model field. Spell each shaft maker one way. Flag lengths no driver can have. Fill ball flight once per model. Approved as patterns, not club by club.
2  ·  Settle the questions that gate the rest
Which field is the record of truth when the name and the table disagree, what a combined flex means, which short drivers are junior clubs, and when a reference price should show. Each is a judgment the data can show but cannot make.
3  ·  Check every club as it is listed
Keep ongoing intelligence where each new club's record is made, so the next week's listings agree with themselves before a golfer sees them.
This is how EKOM moves a catalog from insight to impact —
and keeps every club's page true as the inventory turns.
EKOM
The Resolution Layer
ekom.ai
Case Study  —  Client anonymized  ·  The Resolution Layer
EKOM