live·github ⇄search console

Does your code improve your search ranking? Know at merge.

Most engineering teams ship and hope. Code Results predicts which deploys will move your rankings the moment they merge — then confirms it once your search data lands.

code changes × search rank · 90d
↑ 25 spots
PR merged~14-day lagsearch rank ▲ better
Code changes Search rank
Start for free Case study: ranking 30 → 2
✓ Free to start✓ No credit card✓ 3 min setup
01
4,700+
PRs analyzed
02
11M+
Search impressions
03
Instant
Prediction at merge
04
30 → 2
ganttchart.ai case study
01Capabilities

The full picture,
from merge to rank.

Four data layers. One dashboard. Finally connected.

GitHubGoogle Search Console
01

Developer Velocity

PR throughput, cycle time, and churn — by repo, by sprint, by person. Not a headcount report.

02

Search Performance

Your Search Console data syncs automatically. Impressions, clicks, CTR, rankings — always fresh.

03

Code-to-SEO Correlation

We overlay your deploy history with your search trends. See — not guess — the lag between shipping and ranking.

04

AI Causal Attribution

Statistical changepoint detection isolates when rankings genuinely shifted. Attribution cards link each shift to the exact PRs that caused it.

$ correlation · code changes ⇄ search metrics
real panel · sample data
Sample data: Code Contributions and Search Metrics Correlation panel with daily code addition and deletion bars overlaid on the average position line across three months, and metric toggles for position, impressions, and clicks

Try it on your repo.

Free plan · 3 min setup

Start free
02The Problem

Your team ships every week. But do you know what's working?

Search rankings move. Traffic fluctuates. Engineers ship. Connecting those two things has always required a spreadsheet, a guess, and a prayer.

>Did that content update improve our search rankings?
>Which deploys correlate with traffic increases?
>How long after shipping do we see SEO impact?
>Are we actually driving organic growth?
Get answers
Live · insight feed
PR merged: "Optimize meta tags"
14 days ago · 3 files changed
+23% impressions in the following 2 weeks
Sprint velocity: 47 PRs this month
+2,340 lines added · 23 contributors
Highest output quarter on record
Average ranking position
Google Search Console · Last 90 days
Moved from 21.4 → 14.2 average position
Your feed joins GitHub Search Console

Sample events · your feed builds itself from GitHub + Search Console.

03At-merge Prediction

Know the outcome before Google does.

Rankings take 10–21 days to react to a deploy. Your expectations shouldn't. The moment a PR merges, Code Results matches it against thousands of similar changes shipped by other teams and predicts the ranking impact — then the causal engine confirms it once your search data lands.

>Every merged PR embedded as a 34-dimension feature vector
>Cohort-matched against similar PRs from other teams (pgvector)
>Expected position change with a p25–p75 confidence band
>Confirmed — or corrected — by causal analysis when data lands
At-merge prediction · just mergedcohort-matched · pgvector HNSW
#163feat/product-schemaseo 0.79merged 8 minutes ago
Cohort match
52 PRs · 14 teams
similarity ≥ 0.70
Expected Δ position
↑ 4.1 spots
p25–p75: ↑ 7.9 → ↑ 1.2
Confirmation ETA
~12 days
median across cohort
12 days later · confirmed↑ 3.8 spots actualpredicted ↑ 4.1 · within band
every forecast is graded against what actually happened — accuracy you can audit

Sample prediction · from the same engine the dashboard runs on.

04Causal Analysis

Not just correlation.
Actual causation.

Statistical changepoint detection + LLM attribution turns "something changed around here" into "these 3 PRs moved your ranking by 8.3 positions."

Changepoint detection
CUSUM + sliding-window Welch t-test identifies when your search position genuinely shifted — not just noise or seasonal variance.
Counterfactual chart
OLS regression fits a baseline to the pre-shift period and extrapolates it forward. The gap between that line and reality is your attributed gain.
AI SEO scoring
Every merged PR is scored 0–1 for search relevance nightly by DeepSeek. Content changes, meta tags, structured data, page speed — each rated and explained.
PR attribution cards
Each rank shift is matched to the PRs that caused it, weighted by a 34-dimension feature vector. One card per changepoint. Confidence score included.
$ changepoint_spotlight · avg_position
real panel · sample data
Sample data: changepoint spotlight panel showing a ranking improved 8.3 spots, position 20.5 to 12.2, with the actual position timeline against the dashed no-PR counterfactual trend and a strong-confidence badge
Net Position Δ
↑ 8.3
attributed
SEO-Relevant PRs
64%
score > 0.5
Avg Time to Impact
11d
PR → rank shift
ATTRIBUTED PRsconfidence 91%
feat/optimize-meta-tags
seo score 0.87
62%
feat/add-structured-data
seo score 0.74
28%
fix/canonical-urls
seo score 0.61
10%
05Demo

See your dashboard.

A snapshot of what Code Results looks like once you connect a repo and a Search Console site. Real layout, sample data.

▶ Launch interactive demo
$ dashboard · acme-co/marketing · sc-domain:acme.com
last 90 days
The Code Results dashboard rendering sample data: data-source banner, KPI bar, and changepoint spotlight

Sample data shown above · the real dashboard renders your live GitHub + GSC.

06AI Agents

Bring your AI agent. It gets the full picture.

Code Results ships an MCP server. Claude — or any MCP-capable agent — gets read-only access to your search performance, your PR timeline, and the cross-company benchmarks. Grounded answers instead of generic advice.

  • ›Five read-only tools: properties, daily performance, top queries, PR timeline, benchmarks
  • ›Three installable skills: weekly SEO report, regression triage, ROI narrative
  • ›OAuth consent scoped to your account — revoke anytime in Settings
One command
$ claude mcp add --transport http code-results https://code-results.com/mcp
Setup guide and skills
> why did our average position drop last week?
✓ get_search_performance · position 8.4 → 13.1, shift on Jul 9
✓ get_top_queries · loss concentrated in 6 product queries
✓ get_pr_timeline · 4 PRs merged in the lag window
✓ get_benchmarks · median merge→movement lag fits
Looks like a ranking change, not a demand change. Two candidate PRs merged inside the window — #241 template refactor is the strongest suspect. Timing is evidence, not proof: inspect that diff first.

Sample exchange · your agent runs against your live data.

07What We Track

One platform. No stitching required.

Pull Requests
Commit Frequency
Cycle Time
Contributor Activity
Code Additions
Merge Frequency
Search Impressions
Keyword Rankings
Click-Through Rate
Avg. Position
Traffic Correlation
Impact Lag Window
Causal Changepoints
Counterfactual Trend
PR SEO Relevance Score
Attribution Confidence
$ striking_distance · ranking opportunities
sample data
Sample data: striking distance panel listing terms ranked 5 to 25 with position, trend, impressions, CTR, and estimated click upside
$ keyword_heatmap · impressions per week
sample data
Sample data: keyword heatmap of weekly impression counts for the top six keywords across three months, shaded by volume
08Roles

Built for every seat at the table

CTOs get the org view. Managers get the team view. Everyone gets the answers they've been asking for.

CTO / VP Engineering
  • ›Org-wide velocity at a glance
  • ›Engineering ROI by initiative
  • ›Headcount vs output trends
  • ›Board-ready reporting
Engineering Manager
  • ›Team cycle time and throughput
  • ›PR review bottlenecks
  • ›Sprint-over-sprint comparison
  • ›Contributor health signals
SEO / Growth Lead
  • ›Search impact of recent deploys
  • ›Causal attribution per changepoint
  • ›AI SEO score for every merged PR
  • ›Content experiment results
Product Manager
  • ›Feature ship-to-impact timeline
  • ›Avg days PR → rank shift
  • ›Cross-team velocity comparison
  • ›Release cadence tracking

Ready to connect the dots?

Set up takes 3 minutes. No engineering required.

Get started free
09Pricing

Start free. Go deeper with Pro.

The free plan is the product, not a trial — connect, contribute to the benchmark pool, and see your last 30 days. Pro unlocks everything.

Free
$0/ forever
  • Last 30 days of search + PR history
  • Core dashboard: metrics, correlation chart, changepoint spotlight
  • Top search terms + 3 recommended actions
  • Benchmarks on the global cohort
  • Unlimited repos and sites — your data joins the anonymized benchmark pool
Start free
Pro
$29/ month
  • Full history — everything ever synced, unlocked instantly
  • All modules: striking distance + cannibalization, keyword heatmap, per-term breakouts, DORA, review analytics, live site health, at-merge predictions
  • Benchmark filters by industry and service type
  • Opt your data out of the benchmark pool
  • Unlimited saved views
Start free, upgrade anytime
10Setup

Up and running in minutes

No complex setup. No engineering required. Just connect and go.

01

Connect GitHub

Install the Code Results GitHub App and select which repositories to track. We handle everything else — including automatic repo-to-site matching.

02

Link Search Console

Authorize Google Search Console access with one click. We sync your search data and automatically pair your sites to the right repos.

03

See the impact

Your unified dashboard is live. Correlations, causal changepoints, and AI attribution emerge automatically as your team keeps shipping.

11FAQ

Questions, answered.

Do you need write access to my code?+
No. The Code Results GitHub App is read-only — we read pull-request and commit metadata to map your code activity over time, and we never write to your repositories. You choose exactly which repos to share, and you can revoke access anytime from GitHub.
Is my Search Console data safe?+
Yes. We request read-only access to Google Search Console so we can sync your impressions, clicks, and rankings. We never post, change, or share your data — it is used only to build your dashboard.
How long until I see results?+
Two timelines. Your dashboard populates within minutes once GitHub and Search Console are connected. Ranking impact, though, lags the deploy — most teams see movement 10–21 days after a PR merges, because Google has to crawl, index, and re-rank. We surface that lag instead of hiding it, and predict the expected outcome at merge while you wait.
What if my site doesn't get much search traffic yet?+
It still works. Changepoint detection runs on ranking position, not just clicks, so it can pick up movement on low-traffic pages. We also benchmark each PR against similar changes from other teams, so you get an expected outcome even before your own data is conclusive.
How is this different from Ahrefs or Semrush?+
Those tools tell you what your rankings are doing. They don't connect it to what your engineers shipped. Code Results joins your GitHub history to your Search Console data and runs statistical attribution to tell you which specific PRs moved your rankings — the cause, not just the metric. And it's built for engineers: no keyword research or manual tagging, every merged PR is scored for search relevance automatically. And it works forward, not just backward — each PR gets a predicted ranking impact the moment it merges, so you are not waiting three weeks to find out.
How much does it cost?+
The free plan is free forever: your last 30 days of data, the core dashboard, and benchmarks on the global cohort — no credit card. Pro is $29/month and unlocks your full history, every module, benchmark filters by industry and service type, and the option to opt your data out of the benchmark pool.
12From the blog

Field notes on code & rankings

How engineering work actually moves organic search — crawl lag, causal attribution, and the deploy-level mistakes that quietly tank rankings.

2026-07-19 · 10 min read

SEO attribution tools: a methods guide

Your engineering team merged forty pull requests last month. Organic clicks are up 12%. Most teams cannot say which changes did it. A taxonomy of the five methods that connect what you shipped to what Google did — including where each one honestly breaks down.

Read post
2026-07-02 · 10 min read

Discovered – currently not indexed: what it means and how to fix it

Google knows your URL exists and has chosen not to fetch it. It is not an error, and on a small site it is almost never the "crawl budget" problem the guides tell you to fix — it is crawl demand. What the status actually means, why our own posts sat in it, and the structural fixes that ship in a pull request.

Read post
2026-06-18 · 6 min read

Next.js SEO: the deploy-level changes that move rankings

A client-only SPA fails SEO loudly — the crawler gets an empty shell. Next.js fails it quietly: it renders on the server by default, so a one-line diff can flip a whole route subtree to dynamic, delete a page's metadata, or push its tags out of Google's first crawl — with no error and no failing test. The deploy-level regressions to check by sight.

Read post
All posts

Stop guessing.
Start knowing.

Join engineering leaders who've connected their code to their growth metrics.

Start for free

Free to start · No credit card required · GitHub + GSC included