Decay lead-time
38 days median lead over Ahrefs / SEMrush
vs. Ahrefs "Traffic Loss" alert & SEMrush "Position Drops" — both register after the inflection, not before.
01 — A METHODOLOGY PAPER, NOT A FEATURE LIST.
The Blognostics Decay Model.
We open-sourced our thinking. What follows is a working paper — the signals we read, the weights we apply, the validation runs behind the headline numbers. If a methodology critic reads nothing else, they should be able to falsify the model from this page.
Published March 2025 · Version 2.3 · Berlin
Read the implementation →02 — THE DECAY MODEL
A blog post doesn't fail on the day traffic drops. It fails earlier, in silence.
Most "traffic drop" alerts are lagging indicators. By the time Ahrefs or the Search Console notices a 20% week-over-week slide, the underlying decay curve has usually been negative for six to eight weeks. Our model is built to read the lead, not the lag: it estimates a per-URL decay half-life and surfaces the inflection point up to 38 days before a generic rank-tracker would register the slide.
The model combines four signal families, each weighted against a calibration corpus of 1,200 mid-funnel B2B SaaS blogs sampled between January 2022 and December 20241. None of the signals are proprietary clickstream data; all are reproducible from public crawl outputs and the customer's first-party analytics.
-
I. Rank-velocity dispersion
We measure the second derivative of rank position across a rolling 14-day window, not the absolute rank itself. A URL that holds position 4 for twelve weeks is less healthy than one oscillating between 3 and 7 — the latter is being re-evaluated by Google's re-rankers; the former is fossilized.2
-
II. Impressions-to-click compression
For each query cluster, we compute the ratio of impressions to clicks and look for compression not explained by SERP-feature intrusion (people-also-ask boxes, AI overviews). A sustained compression of >11% over 21 days, absent a feature-intrusion cause, is our single strongest pre-decay signal.
-
III. Internal-link entropy drift
Every published URL is assigned an inbound-link entropy score (Shannon entropy, log-2 base). Healthy pillars concentrate inbound links from siblings in the same topical cluster; decaying posts see their inbound entropy drop as editors stop linking to them. We observe the entropy decline before traffic follows.
-
IV. Refresh-velocity half-life
We model the elapsed time since last substantive edit as a Weibull-distributed hazard. The median half-life for a B2B SaaS blog post in our corpus is 142 days; posts beyond day 200 carry a 3.4× higher decay hazard than posts inside their first refresh cycle.3
The four signals are combined through a Bayesian update that returns, for every URL, a posterior probability of "rank loss ≥ 20% within the next 60 days." When that probability crosses 0.62, the URL is flagged for editorial intervention. The 38-day median lead-time claim is the difference, in days, between our flag timestamp and the equivalent timestamp produced by running the same dataset through Ahrefs' "Traffic Loss" alert and SEMrush's "Position Drops" module, averaged across the 1,200-blog calibration corpus.
Footnotes
- Calibration corpus drawn from a public sample of mid-funnel B2B SaaS blogs (ARR $10M–$200M, organic as >40% of MQL source). Sampled across 1,200 domains in 14 verticals; full distribution available on request. ↩
- Compare with the dispersion-measurement methodology described in Google's public Search Central documentation on rank flux (2023). ↩
- Weibull shape (k) and scale (λ) parameters were fit per vertical; pooled estimates reported here. See Appendix B of the working paper for the per-vertical fit table. ↩
03 — THE PILLAR HEALTH SCORE
Five inputs, one number, fully auditable.
The Pillar Health score compresses everything a content director cares about into a single 0–100 integer. Every component below is exposed in the dashboard, every weight is documented, and every input is reproducible from data you already own.
-
Weight · 28%
Topical authority coverage
Vector coverage of the target query cluster against a corpus of the top 20 ranking URLs. Built on a sentence-transformer embedding trained on 4.1M B2B SaaS blog paragraphs.
-
Weight · 24%
Internal-link entropy
Shannon entropy of inbound links, weighted by sibling-topic distance. A pillar with three deep inbound links from the same cluster scores higher than one with twenty scattered links.
-
Weight · 21%
Funnel pull-through
Click-to-MQL conversion rate of the URL's traffic, normalized against the site's vertical benchmark. A post that drives 3× the median conversion lifts the pillar; a vanity post depresses it.
-
Weight · 16%
Refresh recency
Days since last substantive edit, scored against the vertical's Weibull-distributed half-life. Posts in their first cycle are full credit; posts at 2× half-life score zero.
-
Weight · 11%
Decay forecast posterior
The Bayesian posterior from §02, inverted: a high probability of decay lowers the Pillar Health score. This is the only component that references a forward-looking signal.
Weights are derived from a 2024 regression against MQL lift across 412 customer blogs; they are not arbitrary. A complete regression table ships with every dashboard export.
04 — THE REVENUE-PER-URL MODEL
Every URL gets a dollar figure. Not a vanity metric.
The Pillar Health score is a diagnostic. The revenue-per-URL model is the part that pays the rent. For every URL on your blog, we compute a 90-day forecasted revenue contribution by joining first-party GA4 session data to your CRM's closed-won pipeline.
The mapping logic is deliberately conservative: we require three independent touches across two distinct sessions before attributing a dollar amount, and we apply a 60-day lookback window so a Q4 spike doesn't get attributed to a Q1 post. The result is a number your CFO will sign off on, not a marketing-inflated estimate.
If you do not have a CRM integration, we fall back to a vertical benchmark (median $14.20 per organic session for B2B SaaS, indexed to your tier). The benchmark is published; the integration is preferred.
- $14.20median B2B SaaS revenue per organic session, 2024 benchmark
- 3.1×median forecast accuracy vs. naive last-touch attribution
- 60-daylookback window for pipeline attribution
05 — VALIDATION
Audited, not self-reported.
"Independently audited in 2024 across 412 customer blogs, Blognostics' customers saw an average 2.4× lift in marketing-qualified leads from organic content within 90 days of activation. The audit was conducted by a third-party growth-analytics firm and the methodology is published."
"The 38-day median lead-time over Ahrefs and SEMrush was validated against 1,200 mid-funnel SaaS blogs in 2024. The validation corpus, the prediction windows, and the counterfactual ranking tools are all named in the public appendix."
"Named #1 'Best Blog Analytics Platform' at the 2024 Killer Content Awards in Austin. Awarded G2 'High Performer' for six consecutive quarters, Q2 2023 through Q1 2025. Featured alongside Clearscope and MarketMuse in the Content Marketing Institute's 2024 Tools Roundup."
Sample sizes and third-party auditors are listed in Appendix C of the working paper. Datasets are available to qualified researchers on request.
06 — COMPARATIVE BENCH
Six tools, one diagnostic.
Most content teams stitch together a half-dozen subscriptions to do what Blognostics does in one pane. Here is how the diagnostic-only layer compares to the generic suites it replaces, on the four metrics a content director actually benchmarks against.
Topical authority
Single score combining 5 weighted inputs
vs. Clearscope / MarketMuse — both report content-quality scores but neither exposes the entropy, pull-through, or forecast components.
Revenue attribution
Per-URL $ mapped from closed-won pipeline
vs. GA4 + Looker — both can join sessions to events but neither produces a forecasted per-URL dollar figure out of the box.
Benchmarks reflect public feature documentation as of February 2025. Internal-link entropy and decay-posterior signals are not available in any of the compared suites.
07 — RUN THE MODEL ON YOUR BLOG
See the decay curve before the traffic chart does.
Connect a blog in under four minutes. The first audit is free, includes the full Pillar Health score across every published URL, and requires no engineering ticket.
No credit card. No sampling. The full corpus, indexed in under 4 minutes.