Paid Link Acquisition

An internal tool that runs the backlink workflow end to end: find sites worth buying a link from, find who to email, email them, chase the ones who go quiet, collect the rates into one list, put a shortlist to the three of you each month, then keep checking — forever — that the links we paid for are still there.

It buys link insertions only — a link added into an article a site has already published. Guest posts are deliberately out of scope: those need somebody to write an article, which is a whole separate pipeline, and because we’d be the author, the full HCSA advertising rules would land on us rather than on one sentence.

Five brands: slac · slms · tcl · pww · plmc. euph is out of scope for now.

It never touches money. It tracks where an invoice is and stores it. Finance pays, exactly as today.

1. 01 Keywords and target pages we want to rank for Stays yours. You maintain a Targets tab; the system reads it. It may suggest additions from Rank Tracker — keywords stuck on page 2 where competitors have links we don’t — but never adds them itself. Irma keeps
2. 02 Competitor research on Ahrefs Runs nightly off your Targets tab. Competitors → their backlinks → candidate sites. Automated
3. 03 Filter on category and industry relevance Automated and upgraded — see section 04. Category alone lets link farms through. Automated
4. 04 Find email addresses We crawl the site’s own /contact, /write-for-us and /advertise pages first — free, more accurate than any database, and finding a “write for us” page is the strongest signal they’ll say yes. Only then do we fall back to a paid lookup, and we verify before sending. Automated
5. 05 Send outreach emails Sent from one mailbox, as a real conversation. Two follow-ups at +4 and +7 working days, then it goes quiet for 180 days. Any reply stops the chasing immediately. Automated
6. 06 Compile into a list, tag to brands The list is the system. A reply quoting a rate gets read, the numbers pulled out with the sentence they came from, and you confirm it in one cell. Automated
7. 07 Map links to target pages, decide anchors, finalise costs Proposed for you, not decided for you. We also propose which article on their site the link should go in — see section 05. Anchors are suggested against each brand’s existing anchor spread so the profile stays natural. You override anything. Proposed
8. 08 Team alignment and approval A monthly Buy List in the sheet. Clara, Jolene and Vance veto rows with a reason; one sign-off approves the rest. See section 06. Team keeps
9. 09 Reply with details, get the invoice for finance The reply is drafted for you. Invoice chasing and finance stay human — but the system holds the invoice and knows whether it’s been paid. Half
10. 10 Track backlinks once they’re up, keep the record Automated, and much stricter than today — it keeps checking forever, not just once. This is what we’re building first. Automated

Plus one step that doesn’t exist today at all: a permanent watch on every link we’ve bought. That’s next.

This was originally the last phase. We’ve moved it to first, because it has no dependency on anything else and it’s the highest value for the least work in the whole plan.

For every link on the existing sheet, we fetch the page and check four things:

1. aIs our link still on the page at all?
2. bDoes the anchor text still say what we agreed?
3. cIs the page still live, and still indexable?
4. dIs the link still dofollow?

And we check our end too. A link can be perfectly healthy on their side and worth nothing because the page it points at moved, redirected, or 404s. Worth checking both ends.

What Irma’s sheet needs to have: at minimum the linking site and which of our pages it points to. Anything more — the exact URL of the article, the anchor, what we paid, when — makes the audit better, but we can work without it. Where only the domain is recorded, we can usually recover the specific article from the Ahrefs data we already hold.

Expect this report to contain surprises. That’s the point of running it.

Domain Rating is gameable, and the sites keenest to sell us links are frequently the ones gaming it. A site can show DR 60 while having almost no real visitors and an enormous number of outbound links — which is the signature of a link farm, not a publication.

So a site is rejected outright, before it ever reaches the sheet, if any of these are true:

1. —Organic traffic below a floor DR without readers is the tell.
2. —Traffic down more than half year on year The site is dying; our link dies with it.
3. —Far more outbound links than inbound They sell to everyone.
4. —No overlap with health, beauty, wellness or lifestyle Relevance.
5. —Domain Rating below a floor A floor, not the criterion.

Everything that survives is scored and ranked rather than simply passed — on authority, real traffic, topical fit, share of Singapore traffic, and value for money against the rate quoted. A pass/fail filter can’t answer the question you actually have each month, which is “which six of these forty do we buy?”. A ranking can.

The exact thresholds in the list above are opening guesses. We should set them properly after the first cycle, once we can see what they let through and what they block.

A link is bought on a page, not on a website. And the gap between two pages on the same site is routinely wider than the gap between two sites — a strong publication’s dead 2019 archive post is worth less to us than a mid-sized site’s live, on-topic, actually-ranking article.

So for every site that passes qualification, we look at its individual pages and rank them on four things:

1. —How much traffic that page gets Not the site — the page.
2. —That page’s own authority Ahrefs URL Rating. A homepage-linked guide carries far more than a buried post.
3. —How many outbound links are already on it A page already pointing at forty other sites dilutes ours to almost nothing.
4. —Whether that page is genuinely on our topic And ideally already ranks for something near it.

Then we name that article in the outreach email. “Could you add a link in [this specific guide]?” is a much stronger and more professional ask than “do you sell links”. It also means we can price the page before agreeing to anything — and it stops the site quietly defaulting to whichever page costs them least to touch.

One consequence worth knowing: a site may quote different prices for different pages. That’s treated as several separate prices, not one site rate — because they’re genuinely different products.

Each cycle you get a ranked shortlist in the sheet: the best-scoring options currently worth buying, each with the brand, their article, our target page and the anchor already filled in. You veto rows with a reason; one sign-off approves everything not vetoed.

The intent is that three people look at one view, once a month — rather than forty rows arriving continuously.

Two more things worth flagging about that view:

Quote age is shown, and there’s no automatic chasing. A rate quoted six months ago isn’t really a rate. We chose to show you how old each price is and let you request a re-confirm on the rows you care about, rather than have the system automatically email everyone every quarter. The trade-off is that prices in the pool will drift out of date over time — if that becomes annoying, we turn the automatic version on.

For SLAC and PLMC there’s an extra gate. Because those two are licensed healthcare providers, a paid placement whose surrounding sentence makes a treatment claim is healthcare advertising that we placed. So before payment is released, the anchor and the sentence around it go through the HCSA checker. A fail blocks payment and we ask the site to reword. The other three brands skip this.

1. The existing backlinks sheet — and what’s actually in it Unblocks the audit entirely, which is the first thing being built. The columns you have determine how much we can check straight away: do you have the exact article URL, the anchor we agreed, what we paid, when it went up? Or only the site name? Owner Irma · blocks the audit
2. How many sites do we want to approach in the first push? This is the one number with a legal consequence. More than about 800 emails inside any 30-day window and we cross into “bulk” and legally have to prefix every subject line with <ADV>. If the opening list is large, we spread it over a few months instead. Easier to plan for now than to discover mid-burst. Owner Irma + Vance · blocks outreach build
3. Is “US$100 a link” literally US dollars? Answered already: no weekly or monthly cap, and we aim for US$100 per link. The remaining wrinkle is that most Singapore sites will quote us in SGD. US$100 is roughly S$130 — so a site quoting S$135 is either inside the guide or outside it depending on which currency the rule is really written in, and on the day’s exchange rate. Cleanest is to fix a policy figure in both currencies and stop it moving with FX. Which do we set? Owner Clara · small but it decides what gets flagged
4. Do we buy rel="sponsored" links at a lower price, or refuse them? A sponsored or nofollow link passes no ranking value. Right now the system treats any link becoming sponsored as a fault. If we’d knowingly buy some at a discount — for referral traffic or brand visibility rather than ranking — we need to record that per link, or the audit will keep flagging them as broken when they’re working as intended. Owner Irma + Vance
5. If a site offers a different article than the one we asked for, who decides? Now that we propose a specific page and price it on that page’s numbers, sites will sometimes come back with “we’d rather put it here instead”. Sometimes that’s fine, sometimes it’s a much weaker page at the same price. Should the system accept it automatically when the alternative scores close to the original, or should it always come to you? Owner Irma
6. When two brands both want the same good site, who wins? Today it would just be whichever scores higher. If there’s a real priority order between brands — or a rule like “SLAC first this quarter” — better to say it than to have the ranking decide by accident. Owner Irma
7. Sanity-check the reject thresholds The floors in section 04 are opening guesses. Not blocking anything — but after the first cycle we should look at what got rejected and confirm we agree with it. Owner Irma · after cycle 1

1. Done Spec and decisionsWritten up, decisions recorded. 1 Sep.
2. Next Audit the links we already ownRuns against Irma’s sheet. Delivers the report in section 03, then keeps watching those links permanently. Waiting only on the sheet.
3. Then Discovery and scoringA ranked pool of qualified sites appears in the sheet — each with the best article on it already identified and scored. Outreach is still by hand at this point, but steps 1–3 and 7 of the manual workflow are gone.
4. Then Finding contactsEmail addresses appear against each site. Sending is still by hand.
5. Then The outreach engineSending, chasing, reading replies, pulling out rates. This is the phase the project exists for — the other work is groundwork, not the goal.
6. Then Buy List, approvals, invoice trackingCloses the loop through payment, with the HCSA gate for SLAC and PLMC.