Invoice + Project Quote

Quote #QT-NEE-2026-059

What is still writing your tags — and the tag backlog, cleaned afterwards

Priority: Low, with one exception — the $545 first item is recommended and is not contingent on anything; the three cleanup items behind it are deliberately not urgent, and our own tag audit argues against doing them at all on your current settings. Four items. The first, at $545, finds out what is still writing tags out of your headlines — it is recommended, it changes nothing on your live site, and it depends on nothing else. The other three are housekeeping on the 24,328 tags already in your site: remove the import artifacts, decide the 515 generic single words, merge the near-duplicates — 4,634 tags in total, and deliberately not urgent. On your current settings that housekeeping returns nothing in search: all 24,328 tag archives together earned 46 clicks in 90 days, and tag pages are set to noindex. It is only worth buying after QT-NEE-2026-057 stops what is generating the mess. Priced now so the number is known, not because it needs doing now.

Michael,

This document prices four things, and it is honest about which of them we think you should buy.

The first — finding out what is still writing tags out of your article headlines — we do recommend. It is the cheapest item here at $545 and it depends on nothing else. The other three clean up the tag backlog itself, and those we are recommending you do not buy yet.

The three exist here because the cost of a full tag cleanup is a fair question and you should have the number rather than a shrug. What comes with the number is the honest answer about timing: not now, and not until one other thing is done first. Every reason below is a measurement, and each one is checkable.

ItemAmount
Item 1

First, and cheapest: find out what is writing the tagsshow detailhide detail

$545.00

Take this one first, or take only this one. It is the cheapest item in this document and it is the only one here that does not depend on anything else being bought. Something is still writing tags out of your article headlines. It wrote them on an article published in July 2026, and articles published this year still take 69% of their tags from their own headline. Until that is identified, every other item below rebuilds itself from the next published headline — which is the whole reason the rest of this document is not recommended yet. Where it writes from has already been looked for in the obvious places and is not there. No active plugin and no theme file on the live site writes tags. The single match inside the plugin folder is the WordPress function name sitting in a code comment, not a call to it. So it is most likely an import route, a syndication feed, or an outside posting tool, which makes this genuine investigation rather than a known edit — and it is why it is priced as its own small job rather than folded into a cleanup. This item stops at the finding. It names the mechanism, proves it on a copy of your site, and tells you in writing what stopping it would involve and what leaving it running continues to cost you. Nothing on your live site changes, no tag is merged and no tag is removed. It is worth having whether or not a single tag is ever cleaned, because it is the answer to a question you currently cannot answer about your own site — and because every larger decision about your tags, including the three items below, is a different decision once you have it.

  • Write-path forensics — every route on your site able to write a tag enumerated and tested one by one$325
  • Controlled reproduction on a copy of your site — a test article published and its tags captured as they are written$220
  • Every route on your site able to write a tag listed and tested — scheduled tasks, the publishing interface, the import routes, the feed handlers, and the active and inactive plugin and theme files — proof: the list with the result of each test beside it, including the routes ruled out
  • The mechanism named, with the evidence that identifies it — proof: a tag produced by it on demand on a copy of your site, captured as it is written
  • A test article published on that copy and the tags it receives recorded against its own headline — proof: the headline and the resulting tag set, side by side
  • What stopping it would involve, what it would affect and what leaving it running continues to cost, in writing — proof: the recommendation itself, delivered as a document you keep
  • No change of any kind made to your live site under this item — proof: your tag count and your tag assignment count before and after, shown unchanged

This is the first half of the first item on QT-NEE-2026-057, which finds the mechanism and then stops it. If you take that item afterwards, this $545 is credited against it in full, so finding out first costs you nothing extra. And if the mechanism turns out not to be site code at all, the finding and the recommendation are still delivered and this item closes there rather than expanding into work you did not agree to.

Item 2

The import artifacts removed — 2,703 tags that were never subjectsshow detailhide detail

$750.00

These are the tags that are not wrong so much as not tags at all. 2,092 of them are an article's own headline copied whole into the tag field, producing a one-article archive whose only content is the article that generated it. “'Dig Like Kim' volleyball tournament a great success” is a tag on your site, and it is also the headline of the one article carrying it. This particular artifact is historical — it ran hardest in 2023 and stopped in 2024 — which is why it can be cleared without waiting for anything else in it to be true. 400 are bare numbers or fragments of a date: 12, 29, 911, 6043. 239 are two characters or fewer. 38 are hashtag-style names. One is made of whitespace. Some tags fall into more than one of those descriptions, which is why the figure that matters is the 2,703 distinct tags, not the sum of the classes. Between them they carry 7,809 tag assignments across your articles — a small share of the total, because a tag made from one headline reaches one article. Nothing here has a correct destination to merge into. There is no subject that “6043” belongs under. Removal is the only operation that helps, which is what makes this the least contentious of the three cleanup items here.

  • The removal list built across all five classes and de-duplicated to 2,703 distinct tags$210
  • Checked against the 5,398 hand-applied editorial tags so none is caught up in it$175
  • The list delivered for your approval, and the removal performed only on what you approve$190
  • A before-and-after record per affected article, and a documented way to put it all back$175
  • The removal list covering all 2,703 tags, each with the class that put it there and the number of articles it sits on — proof: the full list, not a sample
  • Every one of the 5,398 tags your editors applied by hand checked against that list — proof: the check shown, including any tag it pulls out of scope
  • Removal performed only on the tags you approved — proof: the approved list and the executed list compared, identical
  • The 7,809 affected article-tag associations recorded before and after — proof: the two states stored and diffable, not summarised
  • A documented way to restore every removed tag and every association — proof: the restore instruction, and it performed once on a test tag and shown reversed
  • No article text, headline, picture, category or publication date altered — proof: those fields compared before and after across the affected articles, shown unchanged

Nothing is removed until you have seen the list. The 5,398 tags your own editors applied by hand are checked against the removal list first and are excluded from it, and that check is shown to you rather than asserted.

Item 3

The 515 single-word tags, decided one at a timeshow detailhide detail

$795.00

This is the item with the most at stake and the least obvious answer, so it is priced on its own rather than folded into the sweep above. 515 of your tags are a single common English word used on 25 articles or more. Between them they carry 41,404 tag assignments — more than half of everything this document touches. They exist because a machine split headlines into words: Announces, Launches, Hosts, Named, Following. There is no merge for these. Nothing is the correct destination for “Announces.” The only operation available is removal, and removal on this scale is the one part of a tag cleanup that can actually cost you something, because a handful of these words are genuine subjects rather than sentence debris. Police is a single common English word. So is Hurricane. So is Election. So each of the 515 is decided individually and the decision is put in front of you as a list before anything happens to any of them. You rule on it. The work is the deciding and the evidence behind each decision, not the deleting. The audit behind this quote recommended against touching these at all while the mechanism that made them is still running, and that recommendation stands. It is the same condition attached to every item here.

  • All 515 words assessed individually against the articles actually carrying them$385
  • The evidence per word — article count and sample articles — so each decision is checkable$175
  • The list delivered for your ruling, and one revision after it$120
  • Removal of the words you rule out, with a before-and-after record and a way back$115
  • All 515 words assessed one at a time — proof: the list, every word carrying a recommendation and the reason for it, none left blank
  • The article count and a sample of the articles behind every word — proof: printed beside each one, including the words the evidence argues for keeping
  • Your ruling applied as given — proof: your marked list and the executed list compared, identical
  • One revision after your ruling, applied as changes rather than returned as a fresh list — proof: the before and after
  • Words you keep left untouched on every article carrying them — proof: the assignment count on each kept word, before and after, unchanged
  • Every removal recorded per article and reversible — proof: the two states stored, and a restore performed once and shown reversed

The list is delivered with the article count and a sample of the articles behind every one of the 515, so a decision to keep a word is as easy to make as a decision to drop it. Any word you keep stays exactly as it is, on every article it currently sits on.

Item 4

The near-duplicates merged — 1,416 tags folded into the spellings you keepshow detailhide detail

$1,150.00

This is the judgement-heavy item, and it runs last because the two items above shrink it. What remains after them is 1,416 tags carrying 27,129 assignments, in three kinds. 316 are case, spacing, hyphenation or punctuation variants of another tag — Miami Beach and miamibeach, city of miami beach and Cityofmiamibeach. 980 are singular and plural pairs. 202 are a broader tag sitting inside a narrower one, so both say the same thing about the same articles. The variants are the ones that actually cost you something today, and not in search. Of the 220 variant pairs the audit found, 196 sit on completely disjoint sets of articles — no article in common. Your coverage of a single subject is split across two archives and neither one is complete. That is a real defect in your own archive whether or not a search engine ever reads it. The merge map is reviewed pair by pair rather than applied by rule, because the rule produces false positives and it is already known to. MIA the airport code and M.I.A. the musician collide under normalisation. So do florida and Flo-Rida. Both were caught by hand in the audit, and both would have been merged by any automatic pass. Every merge keeps one spelling and folds the other into it. No article loses its subject; it gains the articles that were filed under the other spelling.

  • The 316 case, spacing and punctuation variants reviewed pair by pair, false positives excluded by hand$395
  • The 980 singular and plural pairs resolved, with the surviving spelling chosen per pair$305
  • The 202 broader-inside-narrower redundancies resolved$175
  • The whole map dry-run on a copy of the site and reconciled before anything is applied live$165
  • Applied, reconciled against the map, and reversible$110
  • A merge map covering all 1,416 tags, each row naming the surviving spelling and the one folded into it — proof: the map, delivered before anything is applied
  • All 316 variant pairs reviewed individually and the false positives excluded — proof: the excluded pairs listed with the reason, including MIA and M.I.A. and florida and Flo-Rida
  • The surviving spelling recorded per pair with the reason it was chosen — proof: the map column, filled on every row
  • The map dry-run on a copy of your site and reconciled before any live change — proof: the dry-run output and its reconciliation against the map
  • The 196 split pairs shown reunited — proof: the article count on each surviving tag, before and after, with the two former counts adding up to it
  • All 27,129 affected associations reconciled after the live run, with any that failed listed rather than dropped — proof: the reconciliation table
  • A documented way to unwind every merge — proof: the instruction, and it performed once on a test pair and shown reversed

The map is dry-run against a copy of your site and reconciled before a single change reaches the live one. If the import-artifact item and the single-word item above are not taken, this item covers 1,796 tags instead of 1,416 at the same price — the extra 380 are tags those items would otherwise have removed first.

Total$3,240.00

Approve & pay

Tick the items you're approving. You can approve any one of them on its own, or all of them together — you are only charged for what you tick.

Total$3,240.00
THE ONE ITEM WE DO RECOMMENDshowhide

Something is still writing tags from your headlines. It wrote them on an article published in July 2026, and articles published this year still take 69% of their tags from their own headline. Where it writes from has been looked for in the obvious places and is not there: no active plugin and no theme file on the live site writes tags.

Finding it is the cheapest move available, it changes nothing on your live site, and it is the one item in this document that is not contingent on anything else. It names the mechanism, proves it on a copy of your site, and puts the answer and the recommendation in writing.

It is worth having whether or not you ever clean a single tag. It is also the first half of the first item on QT-NEE-2026-057 — the one that finds the mechanism and then stops it — so if you take that afterwards, this is credited against it in full and finding out first costs you nothing extra.

WHAT A FULL CLEANUP WOULD ACTUALLY TOUCHshowhide

Your site carries 24,328 tags. Every one of them was examined on 23 September 2026. 5,271 of them — 21.7% — are near-duplicates of each other or are import artifacts rather than subjects, and between them those 5,271 carry 63% of every tag assignment on your site.

637 of those 5,271 are already priced on QT-NEE-2026-057, which covers the ones a reader can see today: the tags rendering an HTML code instead of a character, the misspelled columnist names, the empty tags and the orphaned rows. They are not charged again here. The three cleanup items below price the 4,634 that remain, and nothing else.

WHY THE CLEANUP IS NOT WORTH BUYING TODAYshowhide

Your tag archive pages are set to noindex, and that instruction was confirmed as genuinely served on the live pages rather than merely switched on in a settings panel. There is no tag sitemap. Search engines are being told, successfully, not to rank these pages.

Across the full 90 days to 21 September 2026, all 24,328 of your tag archive pages together earned 46 clicks from Google, against 9,170 across your site. Half of one per cent.

So merging 4,634 tags would be consolidating authority onto pages that are noindexed and kept out of the sitemap. There is no route by which that produces a search result. This is the opposite of the category work, where merging demonstrably concentrated real ranking signal — because your categories are indexed and your tags are not.

AND IT WOULD NOT HOLDshowhide

The mechanism that generated most of this is still running. It wrote the tags on an article published in July 2026, and articles published this year still take 69% of their tags from their own headline.

Clean 4,634 tags today and the next published headline begins rebuilding the pile. That is the whole argument. It is not that the work is difficult or that the price is high. It is that the result would start decaying the day it was delivered. It is also exactly why the first item in this document exists and why it is the one we recommend.

WHAT CHANGES THISshowhide

QT-NEE-2026-057 finds and stops that mechanism, and replaces it with tagging that reads the article and draws only from a vocabulary you have approved. Once that is delivered, the tag set stops growing on its own.

At that point this cleanup becomes genuinely worth doing, because what it cleans stays clean. Not before.

That is the condition attached to the three cleanup items below and it is printed on each one. If QT-NEE-2026-057 is not taken, our recommendation is that those three are not taken either. We would rather say that here than sell you 4,634 tag operations that quietly undo themselves. The first item is not subject to that condition — it is the diagnosis, not the rewrite.

WHAT IS EXPLICITLY NOT WORTH DOINGshowhide

The audit behind this document ends with a list of what it says is not worth doing to your tags. It is reproduced here in full, with what this document does about each line, so you are looking at the whole picture rather than only at the part that is for sale.

Article headlines stored as tags — 2,277 terms. The audit: ugly but inert. They are noindexed, they sit on one article each, and the mechanism that made them stopped in 2024. In this document: 2,092 of them are the removal priced in the import-artifact item, and the other 185 are already covered on QT-NEE-2026-057. They are sold as tidying your own archive. No search return is claimed for them, because there is none.

Generic single-word tags — 6,977 terms. The audit: not a merge problem. There is no correct destination to merge a word like "Announces" into, so only deletion helps, and only after whatever is writing them has been stopped. In this document: only the 515 that sit on 25 articles or more are priced, and they are decided one at a time rather than deleted by rule, because a handful of them are real subjects. The other 6,462 are in no item here at any price.

Singular and plural pairs — 1,259 terms. The audit: merging "Host" into "Hosts" yields one meaningless tag instead of two. In this document: 275 of them are generic sentence words and are routed into the single-word item to be decided rather than merged, for exactly that reason. The 980 that name real subjects are merged in the near-duplicate item. The remaining four are covered by the other items here or on QT-NEE-2026-057.

Tags duplicating a category — 95 terms. The audit: this would be the top priority if tags were indexed. They are not, so your category page is already winning uncontested. In this document: nothing is scoped to this and nothing is priced against it.

Containment pairs, a broader tag sitting inside a narrower one — 281 terms. The audit: real redundancy, and no consequence at all while tags are noindexed. In this document: 202 are resolved inside the near-duplicate item because they already sit in its map, 67 are generic words decided in the single-word item, and the remaining 12 are covered by the import-artifact item or on QT-NEE-2026-057.

That list is the reason this document is shaped the way it is. It prices a defined 4,634 tags as housekeeping, on archive-quality grounds, with no traffic claimed — and it does not offer a full taxonomy rebuild at any price, because your own numbers do not support one.

WHAT IS NOT BEING CLAIMEDshowhide

No traffic estimate is offered, for either outcome. None was measured, and a number invented for a quote is worth nothing to either of us. What was measured is stated: 46 clicks, half of one per cent of your site traffic, noindex confirmed served on the live pages, and a generator still running.

If your tags are ever opened to search, that is a separate decision and one worth taking only after the tagging itself is fixed. It would change the value of this cleanup materially, and the work would be repriced honestly at that point. On today's settings it does not.

THE ORDER, AND WHYshowhide

The first item stands on its own and can be taken by itself, before or without anything else in this document or any other.

The three cleanup items run in the order printed, and each one shrinks the next. Removing the import artifacts takes them out of the merge map. Deciding the 515 generic words removes a further 311 tags the merge would otherwise have processed and then discarded anyway.

That is why the three counts — 2,703, 515 and 1,416 — add to exactly 4,634. The sets are drawn so that no tag is handled twice, and so that nothing is paid for twice.

WHERE THIS SITS AGAINST THE OTHER TWOshowhide

This is the lowest-priority document of the three in front of you, and it is meant to read that way. QT-NEE-2026-057 fixes what is writing the tags. QT-NEE-2026-058 adds location data to articles that already rank. Both change what your site does.

The cleanup half of this one tidies a part of your site that is currently invisible to search by design, and it is contingent on the first of them. The exception is the first item here, which is diagnosis rather than tidying, and which is worth taking ahead of the cleanup it would inform.

Four items, invoiced in the order printed. Each item is invoiced when it is delivered and is payable on that invoice. Nothing is billed ahead of the work it covers.

Stage 1 — $545. What is writing the tags, identified and reproduced, with the answer in writing. Not contingent on anything, and nothing on your live site changes. Stage 2 — $750. The 2,703 import artifacts removed, under a list you approved. Stage 3 — $795. The 515 single-word tags decided and your ruling applied. Stage 4 — $1,150. The near-duplicate merge, dry-run and then applied.

Each stage is a separate decision and can be taken on its own. They are not a package and there is no discount attached to taking all four.

Stage 1 is recommended on its own. Stages 2, 3 and 4 are each contingent on QT-NEE-2026-057 being delivered first — until the mechanism that writes these tags is stopped, that work rebuilds itself from the next published headline, and we do not recommend buying it.

If you take the first item on QT-NEE-2026-057 after buying Stage 1 here, Stage 1 is credited against it in full.

These prices are held through Wednesday, March 31, 2027 — a deliberately long hold, so that taking QT-NEE-2026-057 first and the cleanup later costs you nothing in price terms.

Nothing in this document is removed or merged before you have seen and approved the list it appears on. Every stage is reversible and the way back is written down as part of the delivery.

Payment methods: Stripe (card) OR Zelle → EDILBERTO MORILLO · 786-426-4902.