Case 51. Crypto portal: 1.24 million ₽ in links, −96% crash and ×15 return (Tier1.shop)

Case from the Tier1.shop collection · 2026 · ← all cases

This is one of the most expensive cases in our selection — second by budget after case #42 — and, perhaps, the most inconvenient for a promotional retelling. The client invested more than a million rubles in links over three years, and during this time the site managed to grow, crash by 96%, lie at the bottom for a year and a half, soar fifteen-fold in two months, and lose half again. We show the entire trajectory because that is what the life of a real site between Google core updates looks like — and it is here that you can see what links do, and what they do not do.

Immediate important context, so as not to give a false impression: the 2024 collapse is not a story about "something being wrong with this site." All comparable crypto portals in the market fell, synchronously and equally deep: out of every 100 visitors that were there before the update, sites were left with 5 to 15 — we checked competitors and show this on the graph in a separate section. The difference is what happened after: the client's site returned two-thirds of its former traffic and continues to grow, one competitor — less than half, the second — one seventh.

Result in brief: 📈 an information portal about cryptocurrencies; the campaign started on a falling site — in a year, traffic grew from 10.7 to 20.6 thousand visits/month (×1,9), keys in the top-3 became 492 versus 183 (×2,7). 📉 March 2024 Core wiped out minus 96% — with continuous purchasing. 📈 Then a recovery that no one expected: 887 → 13 077 visits in two months (×15) — immediately after the campaign's largest batch and June 2025 Core. 📉 December 2025 Core cut it in half again. 📈 Now the second recovery is underway: from March to July 2026, traffic rose to 8.7 thousand and continues to grow. And one result that the client did not order, but which we consider the main lesson of the case: the pages to which the links were pointed barely grew — the entire domain grew.

Starting point

Cryptocurrency portal: reviews of exchanges, exchangers, wallets, faucets — hundreds of pages for informational and near-commercial queries. By the start of the campaign, the site was several years old and was falling: a year before the first link — 19.4 thousand visits (September 2021), by October 2022 — 10.7 thousand, minus 45% per year. This pre-link year is shown on the graphs as a separate gray period — so that the "before" is visible, rather than having to take our word for it. By December 2022, 183 keywords remained in the top-3 compared to 315 in the summer.

Demand in the niche, meanwhile, did not disappear and remains alive: "криптокошелек" — 42 749 impressions/month, "криптобиржи" — 24 465, "криптообменник" — 15 179 (Yandex Wordstat, Russia, July 2026). A disclaimer that is more honest to make immediately: crypto demand is cyclical and follows the market. 2023–2024 saw a bull run — audience interest in cryptocurrencies grew on its own, and part of the traffic fluctuations in this case is explained by the market, not by positions. We take this into account in the conclusions: where growth could have been market-driven, we state it as such.

What was done

The client purchased links from PBN homepages for thirty-eight months almost without interruption: the first months — 5–12 links, from autumn 2023 — batches of 15–26, in May 2025 — the largest batch of the campaign, 41 links. A total of 346 placements on 66 target pages: review sections for exchanges, exchangers, brokers, wallets, bots. 95% of anchor texts are unique — intent + entity, without repeating formulations (source: Tier1.shop placement records).

Three years on one graph

Read from top to bottom: domain traffic, monthly link batches, accumulated mass. Between the graph and the batches is a context track: the gray "year before links" period and the golden "crypto market bull run" (from the cycle reversal to growth up to the price peak of late 2024) — the market background, which must not be confused with either the link effect or Google updates. Five phases:

• Start during the decline (Oct. 2022). The first batches were placed during a downward trend — a rare case for this selection where "before" is not a plateau, but a fall.
• Year of growth (2023). By September 2023 — 20 563 visits, ×1.9 from the starting point. Keywords in top-3 — 492 versus 183, ×2.7. Growth occurred against the backdrop of a starting bull run, so we will not attempt to divide the credit between links and the market — but positions (not just traffic) grew 2.7 times, and positions depend less on market sentiment than clicks do.
• Crash and a year and a half of bottoming out (March 2024 — June 2025). March 2024 Core began the crash: 13.1 thousand in February → 3.8 thousand by April → bottom of 887 in June 2025, minus 96% from the peak. All this time the client continued to buy — 15–26 links in placement months. Links did not protect against the core update — we say this plainly.
• Explosion (July–August 2025). 887 → 6 466 → 13 077 visits, ×15 in two months. Two events coincided: the largest batch of the campaign (41 links in May, another 41 in June–July) and June 2025 Core (30.06–17.07). It is impossible to separate their contribution, and we will not pretend that it is. One thing is certain: there was something to return to — during the "bottom" period, another 177 links were placed on the site, and the recovery brought the site not back to the bottom level, but straight to 13 thousand.
• Second cut and second recovery (December 2025 → now). December 2025 Core: 12.1 → 6.3 thousand. From March 2026 — growth again, 8.7 thousand in July, and this is not yet a closed month.

Positions: the same story, but cleaner

Traffic in crypto is noisy along with the market, so positions are a more honest ruler. The picture is the same, five-phase: growth of top-3/top-4–10 baskets throughout 2023 (top-3: 183 → 492), crash down to 25 keywords in top-3 by October 2024, July 2025 explosion (top-4–10: 133 → 697 over four months), December cut and current recovery: 189 in top-3, 590 in top-4–10 as of July 2026 — approximately the level of the campaign start (220 and 895 in October 2022), after the bottom of 25 keywords this is a distance of ×7. Total visibility in top-50: peak 6 883 (November 2023), bottom 1 791 (June 2025), currently 4 435.

Where the effect went: the main finding of the case

Now the most interesting part. We summed up the traffic of ten pages that received the most links — 157 out of 346, almost half the budget — and placed it next to the domain traffic. The best total result of these ten pages for the entire observation window was 851 visits per month, and it was even before the links, in autumn 2021. The maximum within the campaign — 731 (beginning of 2023). Now — 18. With the domain at 8.7 thousand.

An extreme example — the crypto exchange review section: 27 links over three years, and the page never once entered the top-10 for a single query. It hovers at positions 11–100 with a handful of keywords — with a demand for "crypto exchanges" at 24 465 impressions/month. The reason is visible in the search results: for such queries, the exchanges themselves, aggregators, and media giants are positioned — an information portal failed to break through this with link weight, neither in 27 links nor in three years.

Meanwhile, the domain traffic is collected by completely different pages — reviews of specific services: crypto faucets, electronic currency exchangers, individual payment services. Pages with positions 1–4 for their queries, to which almost no links were sent. The only target page from the top-10 by links that gave a noticeable result was the review of exchanges without verification: peak 303 visits in 2025, a narrow query with low competition.

Full breakdown of the ten target pages. An important honest detail: for most sections, the best values were pre-link (autumn 2021 — 2022), links did not return these peaks:

We have already seen this mechanism in the case #50, where duplicate sections without a single link collected traffic while linked pages fell. It is the same here on a larger scale: link weight works at the domain level, and Google decides which pages to lift, based on the competitiveness of the search results. In hyper-competitive results, the goal is not achieved, but the weight does not disappear — it pushes up those pages that have a chance (source: Tier1.shop observations, cases #50–51).

Comparison with competitors: everyone had a crash, but not everyone had a recovery

To understand what in this story is the fate of a specific site and what is the fate of the niche, we took comparable sites of the same profile (Russian-language information portals about cryptocurrencies, selected by the intersection of search semantics with our site in Semrush) and overlaid their trajectories on one scale: the traffic of each site as a percentage of its own February 2024 level — the last month before the crash. We are not disclosing the names of competitors for the same reason the client's domain is hidden.

• March 2024 Core hit the entire niche. Both comparable portals crashed synchronously with our site, in the same months: by summer 2024, each of the three had 5 to 15 visitors left out of every hundred from February. Competitor B, like our site, began to decline even before the update. This is not "our site was punished" — Google reshuffled the entire class of medium-sized crypto-info sites.
• A large industry media outlet did not experience a crash — at its worst moment it lost about half of its traffic, and now it stands above the February 2024 level (121%). The update redistributed the niche in favor of large brands.
• Recovery was not universal. All three medium sites jumped equally on June 2025 Core — but only ours held on: it returned two-thirds of the pre-update traffic (66%), competitor A — less than half (47%), competitor B — one seventh (15%), and only our site has a 2026 trend directed upwards.

Honest disclaimer: we do not know the link strategies of competitors — perhaps they were also increasing mass, or perhaps not. Therefore, the comparison proves one thing and suggests another. Proves: the crash was niche-specific, and it cannot be held "against" links. Suggests: of the three equally fallen sites, the one that recovered best was the one that received continuous placements — 346 links over 38 months (source: placement tracking and Semrush comparison, Tier1.shop).

Dynamics by Ahrefs: a second source confirms the shape

Everything above was measured by Semrush. So that the story does not depend on a single measurer, we took the same five years via Ahrefs — this is a different system with a different database (world, daily points, estimation of visits per day; absolute numbers are not directly compared with Semrush). Key events coincided:

• The bottom — in the same month: 321 visits/day on June 27, 2025, against the Semrush bottom in the same June. Two independent services date the lowest point identically.
• The same explosion after the bottom: July–August 2025, growth several times over in a matter of weeks, and the same December pullback.
• The same current recovery: from spring 2026, the curve is going up again, the peak of recovery is 4.4k/day in June 2026.

An honest discrepancy that we are not hiding: according to Semrush, the campaign peak (September 2023) is nearly double the pre-link level, while according to Ahrefs, the campaign peak — 15.4k/day — never reached the maximum of early 2022 (17.2k). Different databases evaluate the site's "best times" differently; they fully agree on what happened next — the crash, the bottom, the explosion, the second slice.

The role of anchor text

📈 329 unique formulations out of 346 — 95%, the second result of the selection after the case #42 (96%). There are 16 repetitions, almost all of them twice with a gap from six months to two years, one anchor occurred three times in ten months. This is a safe mode: the clustered repetitions that in the case #44 coincided with the page zeroing out are not present here. Our position remains unchanged: it is better to avoid anchor duplicates altogether — in our experience, the probability of page pessimization with a repeated anchor exceeds 50%.

Conclusions from the case

About updates. Links do not protect against a core update: a minus 96% happened during continuous purchasing and exemplary anchor hygiene. But an update does not cancel the accumulated mass: a recovery of ×15 came to a site that had 300+ links by that time. The update decides "when", the mass affects "whether there is anything to return to". There are no guarantees here — in our own selection, there is a site with a similar history (the same March 2024 Core, also continuous purchasing) that has not returned yet; this analysis will be released separately.

About the choice of target pages. Do not put the entire budget on pages with ultra-competitive search results. 27 links did not bring the page into the top-10 over three years — for queries where exchanges and aggregators themselves are positioned, it is difficult for an informational site regardless of the budget. The rule from cases #43, #44, #47, #49 works in reverse: narrow pages for narrow queries respond to links, wide listings in someone else's search results do not.

About the level of effect. Weight comes to the domain. If the search results for the target query are impenetrable, the effect does not disappear — it manifests on other pages of the domain. For the client, this means: the result of the campaign should be measured by the domain and by visibility, and not only by the traffic of a specific target page.

About the horizon. This is a case about three years. Any measurement within it on a quarterly horizon would have given a false conclusion: in 2023 — "everything is great", at the end of 2024 — "everything is lost", in August 2025 — "a miracle". Long campaigns are evaluated over long windows.

All conclusions are guidelines from a single case, not guarantees: another site, niche, and geo will have its own numbers.

Verifiability

The client's domain is hidden due to the privacy policy, so instead of a link to the site — screenshots from analytics services (estimates in the interface and on our graphs may differ: different databases and regions; the curve shape matches).

Semrush, Domain Overview (daily points for two years): the bottom of the first half of 2025, the vertical takeoff of July 2025, the December slice, and the current recovery — including the fresh jump of late July 2026.

Ahrefs, Site Explorer — organic traffic for 5 years (weekly points): the pre-link maximum of early 2022, the campaign peak of autumn 2023, the crash, the long bottom, and the recovery of 2025–2026.

Ahrefs, Organic positions — top-1–3, top-4–10, and top-11–20 baskets: the same five-phase trajectory as on our graphs.

How we measured

Placement source — Tier1.shop accounting: 346 links with "completed" status, dates, anchors, and target pages; for links from 2022–2024, placement dates were restored from original order tables and verified by donor-acceptor pairs. Traffic and positions — Semrush (RU database), monthly points; separate sweeps by domain and by ten target pages with the maximum number of links. The second independent source — Ahrefs (world, daily points for five years). Competitors for comparison were selected by the intersection of search semantics with the client's site in Semrush and brought to a common scale of "February 2024 = 100%". Query frequency — Yandex Wordstat for Russia. Dates of core and spam updates — according to the confirmed Google calendar; bars on the graphs are built from our directory of confirmed updates for 2022–2026. Two dozen of them accumulated over the case window — in 2023 alone there were four, and the campaign growth continued through them. All graphs were built by us from exported series.

Honestly

• The crypto niche follows the market, and we cannot separate "seasonality" from a bull run. The growth of 2023 coincided with the crypto market rally — part of this growth is market-driven. A counter-fact in favor of the link-based version: the 2024 crash happened while the bull run was still continuing, meaning the site's drop went against market interest — this is an update, not demand. Everywhere traffic could be misleading, we rely on position baskets.
• The contribution of links and June 2025 Core to the ×15 recovery is inseparable. A mega-batch and an update coincided; any percentages of "how much links gave" would be fiction. We show facts and the timing coincidence.
• The site did not reach the pre-link level in terms of either traffic or positions. Campaign peak — 20.6 thousand, currently — 8.7 (−19% from the starting point); keys in top-3 — 189 versus 220 at the start. The case is not about "it became better than it was," but about what happened between these points and how the site returned from the bottom twice.
• Half of the budget went to pages that did not grow themselves. We are not hiding this — it is the main section of the case and its main lesson.
• 31 annual links out of 73 have exceeded their paid term at the time of measurement; eternal (273) are all still active. Expired links are checked for physical viability by our monitoring.
• The campaign was stopped in December 2025 — the second recovery of 2026 is already running on accumulated mass, without new links. This is another argument for the case conclusion #49 regarding the inertia of link mass.
• Parallel work on the site cannot be excluded — we do not have access to the client's internal analytics. What was checked: all ten target pages from the analysis are live and respond at the same addresses.

* Costs are calculated based on current Tier1.shop prices. Links are placed from the home pages of PBN networks — more on the format: links from home pages.