Key takeaways
- Neutralized, not penalized: most dead links were silently ignored by SpamBrain. Nothing to appeal.
- Check manual actions first: no manual action means a disavow file rarely changes anything.
- Use TRACE: Timeline, Reports, Acquisitions, Context, Equity gap. One afternoon, one verdict.
- Plan for months: budget a 90-day rebuild, then judge at six months.
- Not every drop is links: AI Overviews cut top-result CTR by 58% in Ahrefs' 2026 data.
- 2027 rewards trust: relevant regional links and real brand mentions beat volume.
The monthly report says 42 new links. Search Console says nothing happened. You paid $4,000, £2,500 or A$5,500 for that report, and the week a Google update rolled out, your rankings didn't just stall. They slid.
If you're searching for why your backlinks are not working, you're in good company. Google has run four spam updates in 2026 alone, and the September one took 13 days to roll out, against two days or less for the three before it. A lot of link equity that business owners in Chicago, Manchester, Calgary and Perth paid for quietly stopped counting.
This is the playbook we use heading into 2027: a five-part diagnostic called the TRACE Method, a symptom-to-fix table, an honest take on disavow and a 30/60/90-day recovery plan.
Short answer: If your backlinks are not working in 2027, Google has most likely neutralized them rather than penalized you. Its SpamBrain system ignores paid, networked and irrelevant links, and Google says that lost value "cannot be regained." Check Search Console for a manual action first. If there isn't one, skip the disavow file, fix the target pages and replace the lost equity with relevant, earned links over 90 days.
Why are my backlinks not improving rankings in 2027?
Your backlinks usually aren't improving rankings because Google has stopped counting them. Between August 2025 and October 2026, Google ran five spam updates and three core updates, and SpamBrain targets the sites that sell links, not only those that buy them.
Google's December 2022 link spam update announcement said SpamBrain can "detect both sites buying links, and sites used for the purpose of passing outgoing links." Translation: when a guest-post farm gets flagged, every customer loses that equity at once. You did nothing new; your vendor's inventory went to zero.
Google spam and core updates that affected link value, Aug 2025 to Oct 2026
| Update | Dates | What to look for in your data |
|---|---|---|
| August 2025 spam update | Aug 26 – Sep 22, 2025 | Slow fade; looks like seasonality |
| December 2025 core update | Dec 11 – 29, 2025 | 18-day rollout over the holidays; don't mistake it for a seasonal dip |
| March 2026 spam update | Mar 24 – 25, 2026 | Sharp step; done in under 24 hours |
| March 2026 core update | Mar 27 – Apr 8, 2026 | Overlaps the spam update; compare losing pages with winners |
| May 2026 core update | May 21 – Jun 2, 2026 | Page-level shifts; mostly content |
| June 2026 spam update | Jun 24 – 26, 2026 | Steps on pages relying on bought links |
| August 2026 spam update | Aug 18 – 21, 2026 | Check whether your vendor was hit |
| September 2026 spam update | Sep 24 – Oct 8, 2026 | Two-week rollout; let data settle until late October |
Dates are from Google's Search Status Dashboard via Search Engine Roundtable and Search Engine Land. A step down near any of these dates gives you suspects.
Penalized, neutralized or innocent: which one do you have?
Every case of backlinks not working falls into one of three verdicts: a manual penalty, algorithmic neutralization, or an innocent link profile with a problem elsewhere. They need opposite responses, so the verdict comes first.
Link neutralization is when Google's automated systems, mainly SpamBrain, identify links as spam and simply stop counting them. There's no notification and no penalty to appeal. Google's spam updates documentation says that once this happens, "any potential ranking benefits generated by those links cannot be regained."
The three verdicts compared: penalized vs. neutralized vs. innocent
| Penalized (manual action) | Neutralized (SpamBrain) | Innocent (not a link problem) | |
|---|---|---|---|
| Who decided | A human reviewer at Google | Automated spam systems | Ranking systems, SERP changes or your own site |
| Search Console notice | Yes, under Manual actions | None | None (check Page indexing) |
| Typical graph | Sharp, often site-wide drop | Step down near a spam update, or a slow fade | Step near a core update, or clicks falling while rankings hold |
| Does disavow help? | Yes, with removals and a reconsideration request | Rarely; the links already count for nothing | No |
| Real fix | Clean up, document, request reconsideration | Replace lost equity with earned links; improve pages | Content, intent, technical or GEO work |
Open Search Console, then Security & Manual Actions, then Manual actions. It takes 30 seconds. "No issues detected" means no penalty. Good news: nothing to appeal. Bad news: nothing to undo.
Neutralization is far more common than manual actions, and it breaks the "toxic link cleanup" sales pitch. No cleanup restores neutralized equity. The fix is links that count and pages that deserve them.
Picture a Leeds homewares store that bought 30 guest posts from one vendor in 2025. Rankings held until the March 2026 spam update, then slid. No manual action. A disavow file would change nothing; those 30 links already count for zero. The useful question is how to earn ten links a Yorkshire homewares buyer might actually read.
How do I diagnose a link profile Google doesn't trust? Use the TRACE Method
The TRACE Method is a five-part link profile audit that reaches a verdict in an afternoon: Timeline, Reports, Acquisitions, Context and Equity gap. Work in order; each step can end the investigation early.
We built it at Elite Information Tech, a digital growth agency founded in 2008 with 300+ specialists, after seeing too many link problems "fixed" with a 9,000-row export and a needless disavow file.
T: Timeline
Overlay weekly clicks and impressions on the update table above. A step within a week of a spam update points to neutralization; a change inside a core update window points to content; a drop with no update nearby points to something technical.
R: Reports
Check three Search Console reports. Manual actions shows if a human was involved. Page indexing catches the noindex tag that went live in a redesign. Links shows what Google itself sees.
A: Acquisitions
Pull every invoice and report from every link vendor you've used, and mark each domain you paid for. Everyone skips this step, and it matters most. Tools can't tell you which links were bought. Your accounts payable folder can.
C: Context
Score each paid domain on four signals: real organic traffic, topic fit, your market, and no open "write for us, $150" page. Then group your anchors into branded, URL, generic, partial-match and exact-match, and compare the mix with three competitors that rank where you want to be.
E: Equity gap
Count the relevant, trusted referring domains your top competitors have that you don't: regional press, trade bodies, suppliers, partners. Then look honestly at your target page. If it's 400 words saying what every competitor says, the gap isn't only links.
TRACE Context scorecard: what to do with each paid domain
| Score (out of 4) | Verdict | What to do |
|---|---|---|
| 4: real traffic, on-topic, in your market, doesn't advertise link sales | Keep | Leave it alone. If you paid for the link itself, ask for rel="sponsored". |
| 2–3 | Watch | Stop buying there. Don't disavow; it likely counts for little already. |
| 0–1 | Write off | Assume it's neutralized. Disavow only with a manual action or if it came from a PBN or bulk package. |
Vendor rule: if more than half of one vendor's domains score 0–1, treat everything it sold you as lost equity and budget to replace it, not to chase removals.
The TRACE audit checklist
- Weekly clicks and impressions overlaid on the 2025–2026 update dates
- Manual actions report checked and screenshotted with the date
- Links exported from Search Console plus one tool (Ahrefs, Semrush or Moz)
- Every paid domain flagged from invoices and vendor reports
- Each paid domain scored for traffic, topic, market and open link selling
- Anchor mix grouped and compared with three ranking competitors
- Equity gap written down as a number of relevant domains, not a DR target
- One-line verdict: penalized, neutralized or innocent
Symptom → cause → fix: the backlinks not working diagnostic table
Match what you see to the closest row. If two fit, rule out the cheaper cause first.
Diagnostic table: why backlinks aren't improving rankings, and what to do
| Symptom | Likely cause | Fix |
|---|---|---|
| Manual action notice mentioning unnatural links | Human penalty for paid or manipulative links | Remove what you can, disavow the rest at domain level, file a documented reconsideration request |
| Step drop within days of a spam update, no manual action | A vendor or network you used was flagged; equity neutralized | Stop buying from that source, list what was lost, rebuild with earned links |
| New links land, nothing moves for 4–6 months | Placements on sites Google already discounts | Audit host sites for real traffic and topic fit; buy fewer, better placements |
| Fast climb, then a slide that stays down | PBN or bulk links that were later caught | Exit the network. Don't "top up." Treat as neutralized |
| One page tanked; most anchors to it are the money keyword | Over-optimized anchors | Shift future anchors to brand and natural phrases; improve the page |
| Your sponsored posts on big publishers vanished or stopped ranking | Site reputation abuse enforcement on the host section | Stop paying for "parasite" placements; invest in your own domain |
| Drop lines up with a core update; links look fine | Content quality or intent mismatch | Content and UX work, not link work |
| Rankings steady, clicks down | AI Overviews or SERP features absorbing clicks | Track impressions and CTR; work on AI citations |
| Sudden site-wide drop, no update, no notice | Technical failure: noindex, canonicals, robots.txt, migration | Crawl the site and fix before touching links |
| Ranks in the US but not in .co.uk, .ca or .com.au results | No local relevance; all links from one country | Earn links from regional publications and associations in that market |
If there's a manual action, fix that and nothing else first. If there's no manual action but indexing changed, it's technical. If neither, and the drop sits inside a spam update window or you bought links in bulk, assume neutralization.
Should I disavow links in 2027?
Only if you have a manual action for unnatural links, or you knowingly bought links at scale and a manual review is realistically likely. For everyone else, Google already ignores the spammy links, and a disavow file adds risk without adding value.
Disavow is the most oversold fix in link SEO, and Google says as much. Its disavow help page reserves the tool for "a considerable number of spammy, artificial, or low-quality links" that have caused a manual action or likely will, and warns that incorrect use "can potentially harm your site's performance."
John Mueller went further. At Search Central Live NYC he said "internally we don't have a notion of toxic backlinks" and that disavowing is "not a part of normal site maintenance," as Search Engine Journal reported. In September 2026 he added the other side: the tool "has its place & time" and "does have effects," per Search Engine Roundtable.
"Toxic score" is a vendor metric. Google has said it has no notion of toxic backlinks. Disavow for a manual action, not for a red number in a dashboard.
- Disavow if there's a manual action, or you bought a PBN package or exact-match guest-post blast.
- Don't disavow scraper sites and auto-generated stats pages. Google already ignores them.
- Never upload a disavow file you haven't read. We've seen files that disavowed a business's own regional press coverage.
A "toxic backlink audit" that ends with a disavow file and no plan for new links cleans up things Google already ignores. A real audit says what to stop, what was lost and what to build.
Which kinds of links did Google stop counting?
The most common reason for backlinks not working after 2025–2026 is simple: the links exist because someone paid for them, not because a reader would care. Google's spam policies name "buying or selling links for ranking purposes" and "links with optimized anchor text in articles, guest posts, or press releases distributed on other sites" as link spam. Five patterns explain most dead profiles.
- Link farms with a "DA 60" sticker. Forty guest posts a week on crypto, roofing and wedding dresses. No real search traffic of its own usually means Google has decided it exists to sell links.
- Private blog networks. PBNs still produce short bumps, which is why they're sold. Once the network is identified, every connected site loses the value at once.
- Anchors that read like a keyword list. There's no official safe anchor text ratio, but if most links to your "emergency plumber Manchester" page say exactly that, it's a pattern. Algorithms catch patterns.
- Velocity with no story. A viral study can earn 300 links in a week. Sixty guest posts in a month for a 12-person Calgary HVAC firm that's never been in the news can't.
- Site reputation abuse placements. Posts in big publishers' coupon sections or third-party hubs, content "published on a host site mainly because of that host's already-established ranking signals." When enforcement hits, the links go too.
"High authority backlinks" are worth paying for only when the authority is real: an audience, an editor, a reason to link. A DA-band price list isn't that.
How do I recover from a Google spam update? The 30/60/90-day plan
Google spam update recovery takes months, not weeks. Google's documentation says changes help once "our automated systems learn over a period of months" that a site complies, and neutralized links never come back. The plan isn't to reverse the update; it's to replace fake equity with real equity.
- Freeze all link buying (days 1–3)Pause every vendor and "monthly link drip." You can't diagnose a profile that's still changing.
- Run the TRACE audit (days 1–14)End with a written verdict, a list of paid domains and a yes, no or scoped decision on disavow.
- Fix blockers and manual actions (days 15–30)Resolve technical problems first. If there's a manual action, request removals, disavow what's left at domain level and file a documented reconsideration request.
- Repair the pages that lost the most (days 31–60)Rewrite the money pages with real expertise and current data, and fix internal linking so authority flows to them. Links amplify good pages; they rarely rescue thin ones.
- Earn your first relevant links (days 31–60)Pursue placements a customer might actually read: regional press, trade publications, associations, partner pages and digital PR built on your own data.
- Build a steady cadence and measure (days 61–90)Settle into a believable monthly pace with natural anchors. Report impressions, CTR, in-market referring domains and AI citations, not only positions.
What "done" looks like at each checkpoint
| Checkpoint | Done means | Warning sign |
|---|---|---|
| Day 30 | Written verdict, risk list of paid domains, technical issues fixed, reconsideration filed if needed | Still arguing about whether to disavow everything |
| Day 60 | Improved target pages live; first handful of relevant earned links | Links landing on pages nobody has touched |
| Day 90 | Sustainable monthly cadence; impressions stable or rising; six-month roadmap | Pressure to "speed it up" with a bulk package |
On budget: expect to spend more per link and less in total. In niches like personal injury in Texas or solar in Queensland, 90 days gets you back on track, not to page one. Anyone promising page one probably sold you the last batch.
If you want to test a new provider before committing, start small. Our small backlink packages let you judge the sites and reporting on a small order before you commit.
What if backlinks aren't the problem at all?
Often they aren't. Plenty of audits end with the verdict "innocent": the links are fine, and something else moved. Check these before you spend another dollar, pound or loonie.
AI Overviews are taking the clicks
If rankings hold but traffic falls, look at the SERP. Ahrefs' 2026 study of 300,000 keywords found the top result's click-through rate is 58% lower when an AI Overview appears. More links won't fix that. Getting cited inside the answer might: that's the job of AI search optimization (GEO).
A core update re-scored your content
If the drop started between December 11 and 29, 2025, March 27 and April 8, 2026, or May 21 and June 2, 2026, compare your pages with the ones that replaced you. If theirs are more specific, current and expert, that's content, not links.
Search intent moved
The query that showed service pages now shows comparisons, Reddit threads or videos. Your page didn't get worse; it stopped being the right format.
Something technical broke
A migration, a staging noindex that went live, a plugin rewriting canonicals: all look exactly like a penalty. Check Page indexing first.
The money page is thin
Fifty links pointed at a 300-word page won't make it rank. Link work pairs best with proper SEO services that fix what you're linking to.
What does a trustworthy link look like in the US, UK, Canada and Australia?
A trustworthy link comes from a publication or organization your real customers read or belong to. In 2027, relevance means topic and geography: a Sydney accountant gains more from one state business publication than a dozen generic US blogs.
Trusted link sources and watch-outs by market
| Market | Links that tend to hold value | Watch out for |
|---|---|---|
| USA (USD) | Metro business journals, state trade associations, chamber of commerce pages | Scholarship-page link schemes; FTC rules on disclosing paid endorsements |
| UK (GBP) | Regional dailies, trade bodies, edited .co.uk industry directories | Unlabelled advertorials; ASA/CAP rules require paid content to be obviously identifiable |
| Canada (CAD) | Provincial business press, outlets like BetaKit for tech, .ca partner pages | Links only from US sites; Competition Bureau scrutiny of undisclosed paid promotion |
| Australia (AUD) | State business press, SmartCompany-style SME outlets, .com.au supplier pages | Offshore guest-post farms with a .com.au veneer; ACCC guidance on disclosing paid endorsements |
Geography is an overlooked cause of backlinks not working in .co.uk, .ca or .com.au results. A few relevant US links are fine; a profile of mostly US blogs is a weak signal for UK, New Zealand or Irish results. Comparing link building services in the USA with UK guest posting providers? Ask each for live examples in your country. For businesses that sell in one city, those regional links also do double duty for local SEO.
How do I vet my next link building provider?
Ask for live examples, vetting criteria, anchor control and reporting, and walk away from anyone who won't show sites before placement. This script takes five minutes.
Subject: A few questions before we sign
Hi [Name],
Before we sign, could you answer these?
1. Share 5 live placements from the last 90 days in our industry.
2. Which signals do you check before placing, besides third-party
authority scores? (organic traffic, topic fit, audience country)
3. Who picks the anchor text, and what mix do you usually aim for?
4. Do you or your partners own any of the sites you place on?
5. Will we approve each site before the content goes live?
6. Which markets can you earn placements in? We need
[US / UK / Canada / Australia] publications specifically.
7. What do you report monthly besides a list of links?
WALK AWAY IF: "our network is proprietary", pricing by DA band,
50+ links in two weeks, exact-match anchors by default, or the
only KPI is "links delivered".
Thanks,
[Your name]
Ask us the same questions. To start with a few placements instead of a retainer, our guest post packages are a smaller first step, so you can judge the sites and reporting before committing to more.
Ten placements on relevant, real sites will outlast a hundred generic guest posts, and they'll still be counting after the next spam update.
That's our standard. Elite Information Tech's link building services are built on earned placements on relevant, real sites in your market; see how that plays out in our case studies.
What should I expect from Google and link building in 2027?
Expect fewer working shortcuts, longer spam rollouts and more weight on brand signals AI systems can read. These are our forecasts, based on changes through October 2026, not 2027 data.
- Spam updates will keep coming, some slower. Google shipped four in 2026, and September's took 13 days. Plan on two to four in 2027.
- Vendor-level risk will grow. Because SpamBrain targets sellers, one flagged marketplace can neutralize hundreds of clients' links at once.
- Brand mentions will matter more for AI visibility. Ahrefs found branded web mentions correlate with AI Overview visibility at 0.664, against 0.218 for raw backlink count. Expect digital PR to take budget from bulk guest posts.
- Reporting will shift from rankings to visibility. Impressions, AI citations and branded search will sit next to positions in serious reports.
The bottom line on backlinks not working in 2027
Fixing backlinks not working in 2027 starts with the verdict. Check for a manual action, line your graph up against the update dates and audit what you paid for. Disavow only when Google's guidance says to. Then spend on links a real editor would give you, pointing at pages that deserve them.
It's slower than buying another batch. It's also the only approach that survives the next update.