When to rewrite the page instead of buying links
Four pages on one of my sites covered the same subject. I wrote all four, years apart, and each had picked up a little of whatever that site has to give.
None of them had all of it.
Merging them took an afternoon and cost nothing. I had a link budget open at the same time, and it never occurred to me that the afternoon should come first, because editing my own old work does not feel like doing something and wiring money to a stranger does.
I am the interested party here
I sell dofollow placements on my own properties at $250. That is income, and it is the reason to read the rest of this with a hand on your wallet. I buy as well. $387.11 across three sites last year, which I have gone through line by line elsewhere.
Everything below costs me money if anyone acts on it. That seems worth saying out loud before the argument rather than in a footnote after it.
Start with where the page actually sits
One check decides the rest. Find the position your page currently holds for the query it is supposed to answer. The query a person types, not the phrase that ended up in the brief. Nowhere at all means links will not rescue it. I mean absent for its own topic, not sitting at 40, not a scatter of impressions on some adjacent phrase.
Just outside the top results, with pages above you cannot honestly call better, is the other case. That one is a link gap and buying is a reasonable way to close it.
Almost nobody runs the check before approving the spend, and I think the reason is that the answer is unpleasant in one direction and boring in the other.
Absence is a different condition from a low position
A page at eleven has been accepted as a candidate answer and then ordered below other candidates. That ordering is an argument you can join. A page nowhere was never in the argument. External agreement about how important it is changes nothing about what happens when somebody reads it and finds the question still open.
I have never watched bought links move a page from absent to the top ten for a query it does not answer. I have watched several people spend four figures finding that out.
Signal one: it left things out
Open the results above you and list what each one covers. Coverage, not a word count target.
If every one of them prices the thing and yours never mentions money, that is your finding. If every one of them has a section on the failure mode and yours only describes the happy path, same.
Short pages beat long ones constantly when they say the specific thing nobody else says.
Length is a symptom of omission and nothing more interesting than that.
Signal two: nobody types your phrase
Written from inside a company, a page gets arranged around the words that company uses in its own meetings. Customers use different words.
The tell is comfortable, which is why it survives for years. You rank first for your phrase, the page feels healthy, and the arriving traffic is a rounding error.
I have done this to myself more than once. One page of mine spent about two years arranged around a term I only ever use when talking to suppliers, while the shorter and blunter thing an actual buyer types appeared nowhere in the title, the opening or any heading on the page.
The repair is a title, an opening paragraph and two headings. Twenty minutes, no invoice, and the highest return edit on this list. Nobody sells it to you, so it stays undone.
Signal three: it was written for a crawler
Read the page out loud. The rhythm answers this on its own.
A keyword repeated because somebody set a target for how often. A definition paragraph aimed at a reader who obviously knows the definition. A heading with nothing under it worth the heading.
The last two years added a specific version: a page assembled by a model in ninety seconds and never edited by anybody who cared. Even paragraphs, hedged throughout, nothing precise enough to be wrong. No volume of links repairs the experience of reading it.
Signal four: you are competing with yourself
Publish for a few years and the same topic comes round more than once. Somebody forgot, or the angle looked different that quarter, or two writers never compared notes.
Four pages, one subject. Each holds a fraction of the standing. They compete with each other before they compete with anybody outside, and every link you buy for one of them makes the split worse.
This is a different fault from one article answering at four separate addresses, which is a delivery problem with its own checklist. These are four genuinely different pages that should have been one.
Finding it takes one search: your own domain, plus the topic, then count how many of your own results come back arguing with each other.
Merging is the cheapest fix in SEO. No outreach and no invoice.
How the merge goes
Keep the one with the strongest external links pointing at it, even when it is not the best written of the four. Links are the part you cannot rebuild in an afternoon. Move everything the other three say that the survivor does not into the survivor, in the place it belongs. Bolting the leftovers onto the bottom under a heading called additional information does not count.
Then redirect the other three at the survivor, and repair the internal links that used to point at the pages you retired. That last step gets skipped, and skipping it leaves a site pointing at its own redirects for years. I have written up internal linking on its own and will not rebuild it here.
Keep those redirects permanent and write them down somewhere you will find again, because a merge quietly undone by a hosting change two years later puts the split straight back and nobody notices for another year.
Fix, wait, then decide
Rewrite. Wait a month. Then look again and decide about links.
The waiting is load bearing. A rewritten page has to be fetched again and reassessed against everything it lost to, and if you rewrite on Monday and buy placements on Thursday you have paid for an outcome you can no longer attribute to anything.
You also destroy your own read on the result.
Two changes at once and all you know afterwards is that a number moved. Writing that instruction is easier than sitting through it. A month of nothing while a competitor publishes twice a week feels like standing still.
Why the budget goes to outreach anyway
Outreach produces artefacts.
A prospect list, replies arriving, a spreadsheet with statuses in it, something to show on a Friday.
Rewriting produces an admission that the first version was weak. If you wrote it, or approved it, that admission costs something internally that has nothing to do with search.
As far as I can tell that is the whole mechanism. The money goes where the discomfort is not. I have done it myself, and bought a placement rather than reopen a page I knew was bad.
The bad page you successfully promoted
Assume the links work. The page climbs and people arrive on the page you already knew was thin. You have paid to route strangers to your weakest work. A position is the start of the transaction. A page that does not answer the question now fails to answer it in front of a larger audience.
What a link is worth when the page underneath it has no readers is its own question, and I have answered it separately. This one is about the page your links point at.
When buying is the right call
I am not writing off my own business, so here is the other half.
A page that is thorough, current, better than at least two things above it, and parked at eleven for six months does not have a writing problem. The pages above it have links it has not got. Buy them.
Same for a genuinely new page in a category where everything ranking is a decade old. Age and accumulated links are the difference, and one of those two cannot be purchased.
So the question is whether anything has ever agreed your page belongs in the conversation. If something has, links argue about the order. If nothing has, you are arguing with a closed door.
What I cannot prove
There is no clean before and after here. My reporting is in a state I do not trust, so any chart would be decoration.
When I merged those four pages I changed other things in the same fortnight, which is exactly the mistake I just told you to avoid.
And I cannot separate the value of the rewrite from the value of the page simply being touched, recrawled and dated more recently. I suspect a real share of it is the second thing. Nobody selling a course on this can separate them either.
Before you approve a budget
Take the pages the money is meant for and run the four signals over each one.
The pages that fail are not link candidates yet. What the money buys on those is a delay, and at the end of the delay you reopen the page anyway. The way I quote placements, and what I ask a buyer before taking money, is here.