Nakodo has a glossary: 27 plain definitions of the words that come up in creator sponsorships. It exists because the rest of the site uses those words, and a reader who does not know what a rate card is should not have to leave to find out.
That is a claim with two halves, and both are measurable. Are the words we define actually used? Are the words we use actually defined? I fetched the 76 public pages and counted.
Method, and the two traps in it
Take the visible text of each page with html.parser, then strip header, footer and nav before counting anything. The first version of this measurement did not, and the result was a list of our own navigation links: the top phrases across the site were "guides blog", "rate calculator" and "start free", on 76 pages each, because every page carries the same footer.
The second trap is substring counting. The word "reach" appears 251 times if you count it with str.count, and 71 times if you count it with a word boundary, because "outreach" contains it. A jargon census is a list of words that look alike, so every count below uses (?<![a-z])term(?![a-z]).
Half the glossary carries the site, the other half is barely used
Occurrences of each defined term, and how many of the 76 pages use it:
Engagement rate 362 76 pages
Disclosure 171 41
CPM 99 18
Median views 96 37
Business email 89 38
Discount code 67 18
Shorts 63 15
Affiliate link 61 14
Link-in-bio page 55 11
Product seeding 51 24
Usage rights 50 22
Exclusivity 48 20
Fit score 39 26
Media kit 32 8
Dedicated video 31 9
Flat fee 23 15
Micro creator 22 6
Rate card 21 13
Views per subscriber 20 5
Sponsorship brief 18 12
UTM parameters 14 5
Cost per acquisition (CPA) 13 2
Brand safety 6 4
Multi-channel network (MCN) 5 1
Paid promotion label 5 4
Talent manager 4 3
Brand integration 3 1
Nothing in that list is wrong, but the bottom five are definitions written for completeness rather than for a reader who got stuck. "Brand integration" is defined once and used three times, all on one page. A glossary is allowed to be an encyclopaedia, and ours was clearly written as one.
The links only go one way
pages linking to /glossary (the footer does it): 76
pages linking to a specific definition, /glossary#x: 23
total in-text links to definitions: 59
distinct anchors linked: 22 of 27
anchors nothing links to: business-email, engagement-rate, fit-score,
multi-channel-network, talent-manager
Every page can reach the glossary and a third of them point at a definition. The five anchors nothing links to are interesting as a group: three of them ("business email", "engagement rate", "fit score") are the three most-used concepts on the whole site. They have no inbound links because they each have something better: a guide, a calculator, or a page explaining it. The glossary entry is the fallback for a term with no home of its own, and the measurement shows exactly which terms those are.
The words we use and do not define
Counting a hand-built candidate list of industry vocabulary against the glossary:
term uses pages defined
outreach 147 38 no
reels 76 18 no
reach 71 36 no
commission 62 22 no (it is inside the affiliate-link definition)
spark ads 60 22 no
median plays 44 20 no
paid partnership 44 14 no (we define "paid promotion label")
follower count 40 21 no
partnership ads 37 13 no
ugc 36 9 no
fake followers 23 9 no
subscriber count 21 10 no
pitch 19 13 no
deliverables 18 8 no
gifting 11 8 no
Two of those stand out because of what we do define.
We define Shorts, with a link to YouTube's help page, and we do not define Reels, which our pages use 76 times across 18 pages. We define Median views, which is the number our whole pricing discussion rests on, and we do not define median plays, which is the same idea on TikTok and appears 44 times on 20 pages. The glossary even says "keep long-form videos and Shorts apart", so the asymmetry is not an opinion about platforms.
The explanation is a date. This product started as a YouTube tool and the glossary was written then. Instagram and TikTok pages arrived later, with their own vocabulary, and nobody went back. You cannot see that from reading the glossary, which looks complete. You can see it instantly from a count of what the other 75 pages say.
The acronyms nobody expands
For each acronym, how many pages use it, and how many of those spell it out at least once:
used on expanded on
FTC 23 1
ASA 18 0
CMA 18 0
UGC 9 1
CPM 18 10
UTM 7 5
CPA 1 1
MCN 1 1
The bottom four are in the glossary, and the glossary entries carry the expansion, which is why those numbers look healthy. The top four are not, and two of them are regulators: 18 pages of ours mention the ASA and the CMA, and not one of them says Advertising Standards Authority or Competition and Markets Authority.
These appear on our disclosure rules pages, which exist precisely so a reader who does not know the rules can follow them. Telling that reader to check "the ASA" is telling them to search for an acronym. This is the clearest finding of the whole exercise and it is six words of copy per page.
What I am changing
- Four new definitions: Reels, median plays, Spark Ads and partnership ads, UGC. All four are already used on a dozen pages or more.
- Expand ASA, CMA and FTC on first use, on every page that names them. Then a test: a list of acronyms that must be expanded somewhere on any page that uses them, checked over the rendered text rather than the source, because the rendering is where a reader meets it.
- The glossary keeps its five unlinked entries. Three of them are unlinked because the concept outgrew the glossary, and that is the system working.
- The jargon census becomes a script in the repo, run before publishing a new batch of pages. Not a test: a list of words a human should look at, because "a word that appears on twenty pages and is defined nowhere" is a judgement, not a failure.
Top comments (0)