Your search data, turned into a ranked fix list your AI agent can ship

ScaleSleek reads Google Search Console and Bing Webmaster Tools, ranks what to fix by the clicks at stake, and hands each fix to your agent over MCP. Then it measures every change at day 14 and day 28.

Free during the beta · no card

scalesleek.com/app/sites

Sites

Sep 6 – Oct 3 · changes vs the 28 days before

Synced · Google 6m · Bing 12m
Google clicks16,455+4.5%
Bing clicks4,485+12.3%
Indexed (sampled)90%
Open fixes13ranked by clicks
Google Bingdaily clicks
  • brightloaf.example4,812+12.4%7open fixes · Bing 1,106
  • invoicekit.example9,407+3.1%2open fixes · Bing 2,961
  • hikeplanner.example2,236−4.2%4open fixes · Bing 418
Top fix/sourdough-starternear-win at position 9.4est. +220 clicks / 28 days

Example data, the same site the whole way down.

From search data to a measured change, in four steps

One example site follows the whole way: brightloaf.example and its sourdough starter page.

Connect Google and Bing once

Sign in with Google for read-only Search Console access, then paste a Bing Webmaster API key. ScaleSleek can't change anything in Search Console, and it sends URLs to Bing only when you ask your agent to.

Every property you can see shows up. Pick the sites you want, and the first sync reads Google’s last 180 days and Bing’s weekly data.

scalesleek.com/app/onboarding

Google Search Console

Google sign-in · read-only

sc-domain:brightloaf.exampleConnected

Bing Webmaster Tools

Your Bing API key

https://www.brightloaf.example/Connected
First sync · Google’s last 180 days, Bing by weekdone

Get one fix list, ranked by the clicks at stake

Pages just off page one, low click-through, pages Google won't index, pages losing clicks, two pages splitting one search, and pages one engine ranks while the other doesn't.

Estimates come from your own data with brand searches left out, and from a published click-through curve where your data is thin.

scalesleek.com/app/sites/brightloaf.example/fixes
  1. 1

    /sourdough-starterNear-win

    Position 9.4 for “sourdough starter ratio”, 4,860 impressions

    +220
  2. 2

    /banana-breadEngine gap

    Bing position 3.1, Google 22.6 for “easy banana bread”

    +127
  3. 3

    /focacciaLow CTR

    Position 2.3 but 1.9% CTR, where 8% is expected

    +43
  4. 4

    /rye-breadNot indexed

    Google: Crawled, currently not indexed (since Sep 28)

    +41
  5. 5

    /bagelsDecaying

    Clicks down 31% on the 28 days before while the site rose

    +22

est. extra clicks over 28 days if fixed

Your agent makes the change in your repo

The fix packet holds the evidence, the live page, the searches to target, the words to keep and the length limits. Your agent edits the page; ScaleSleek never writes to your site.

Then it logs the change and checks it went live: status 200, indexable, the right canonical.

~/brightloaf · claude

> fix the top finding on brightloaf.example

scalesleek · get_findings(site: "brightloaf.example")

1. /sourdough-starter · near-win · pos 9.4 · est. +220 clicks

scalesleek · get_fix_packet(site: "brightloaf.example", id: "f_k3m7q2xw4p")

target “sourdough starter ratio” · keep: “starter”, “feeding” · title ≤ 60 chars

app/sourdough-starter/page.tsx

- title: "Our Sourdough Starter Story"

+ title: "Sourdough Starter Ratio and Feeding Guide, by Weight"

scalesleek · log_change(site: "brightloaf.example", url: "https://www.brightloaf.example/sourdough-starter", what: "Title targets the ratio search", findingId: "f_k3m7q2xw4p")

scalesleek · verify_fix(site: "brightloaf.example", fixId: "fx_3f9a1c7e5b2d4086")

✓ 200 · indexable · canonical · target search in the title

See what each change earned at day 14 and day 28

Each change is measured against the rest of your site over the same weeks, so a sitewide rise doesn't count as a win.

Losses show up as plainly as wins.

scalesleek.com/app/sites/brightloaf.example/tracked

/sourdough-starter

Day 28 · measured Nov 4
Clicks
171
Expected
64
Result
+107
28 days beforechange shipped Oct 428 days after

The rest of brightloaf.example rose 13.7% in the same weeks, so the page alone would have got about 64 clicks. Day 14 was +49.

Built for the agent that already edits your site

ScaleSleek is an MCP server. Add one address and your agent can read both engines, pick the top fix, and check its own work.

Claude Code

claude mcp add --transport http scalesleek \
  https://www.scalesleek.com/mcp

Cursor · mcp.json

{
  "mcpServers": {
    "scalesleek": {
      "url": "https://www.scalesleek.com/mcp"
    }
  }
}

You sign in once in the browser, and your agent only sees the sites you've added. The app has the steps for Claude, Claude Code and Cursor.

agent chat

How did the sourdough change do?

scalesleek · get_tracked_fixes(site: "brightloaf.example")

It beat the expected number. In the 28 days after you changed the title on /sourdough-starter, it got 171 clicks.

The rest of brightloaf.example rose 13.7% in the same weeks, so the page alone would have got about 64. It beat that by 107 clicks, about half the +220 the finding estimated.

Next on the list is /banana-bread: Bing ranks it 3rd, Google 23rd. Want the packet?

11 tools. None of them change your site, and your agent runs submit_indexing only when you ask it to.

  • list_sitesYour sites with Google and Bing clicks for the last 28 days.
  • site_summaryTotals for 7, 28 or 90 days against the period before, with the top gaining and losing pages.
  • page_reportOne page on both engines: clicks over time, top queries, index status, on-page facts.
  • search_dataA live Search Console or Bing report by query, page, date, device or country.
  • inspect_urlsIndex status for up to 25 URLs: Google's verdict and last crawl, Bing's status.
  • get_findingsThe fix list, ranked by estimated click gain.
  • get_fix_packetEverything needed to fix one finding, with the site's rules.
  • log_changeRecords a change so it's measured at day 14 and day 28.
  • verify_fixFetches the page live and checks the change is there.
  • get_tracked_fixesLogged changes and their results, wins and losses both.
  • submit_indexingOnly when you ask: sends URLs to Bing and gives Google inspection links.

Google and Bing, side by side

Most tools read Google only. Bing's data sits in a separate dashboard few people open, and it often disagrees with Google in ways worth fixing.

One dashboard
Clicks and impressions from both engines side by side, with exact dates, plus Google's index status and Bing's crawl data.
Engine gaps
Pages one engine ranks and the other buries become findings, because the fix is often the same.
Copilot reads Bing
Microsoft Copilot answers from Bing's index, so a page Bing can't find is missing there too.
brightloaf.example · clicks by engine · Oct 5 – Nov 1
Google Bing
  • /sourdough-starter17133
  • /banana-breadEngine gap531
  • /focaccia1510
  • /bagels4740
  • /rye-bread012

Tested on my own sites first

These are real results from the sites ScaleSleek was built on, not example data.

I run a handful of my own sites: pedscalc.com, allthecalculations.com and swiftschema.com. I built ScaleSleek because I wanted my agent to pick fixes from real search data, and then tell me honestly whether they worked.

The first measured result was a change on pedscalc's ondansetron page. My agent answered from the ledger, and every number it gave was in a tool result: 141 clicks, about 92 expected, +49. The day-28 results come in the same way, wins and losses both.

Austin WitherowBuilds ScaleSleek and the sites it was tested on
+49
clicks at day 14 on pedscalc.com/ondansetron, beyond what the rest of the site's rise would explain
141 vs 92
clicks the page got, against the about 92 it would have got as pedscalc.com rose 13.5% in the same weeks
41
earlier pedscalc changes loaded into the ledger, each measured as its day-14 and day-28 dates come due

Free while it's in beta

beta

$0while the beta runs

We'll email everyone on the waitlist when it opens.

  • Up to 5 sites, Google and Bing
  • All 11 MCP tools
  • Every fix's day-14 and day-28 result
  • No card
Join the waitlist

after the beta

Free
$0
1 site, the fix list and your first 3 results
Pro
$29/mo
5 sites and every result
Portfolio
$99/mo
25 sites and every result

There's no AI running on our side; your own agent does the writing. That keeps the price low. Exact limits are set at launch.

Want us to run the fix loop for you? Concierge is $999 a month: we pick the fixes, open each one as a pull request in your repo for you to merge, and report every result. Email austin@buildleansaas.com

Questions about your data

ScaleSleek connects to your Google Search Console with read-only access, so it can't change anything there. It finds the pages losing clicks or sitting just off page one, and checks at day 14 and day 28 whether each fix worked. How we use your Google data is in our Privacy Policy.

What can ScaleSleek do with my Google account?

Read your Search Console data, plus the name, email and profile picture you sign in with. It asks Google for the read-only Search Console scope, so it can't change settings, remove URLs or add users.

And with Bing?

It reads your Bing Webmaster data with the API key you paste. It sends URLs to Bing for indexing only when you ask your agent to, within Bing's daily quota.

Does it change my site?

No. Your agent edits your site in your repo, the way it edits anything else. ScaleSleek gives it the fix packet, then fetches the live page to check the change landed.

Do you use my data to train AI?

No. We don't sell it, use it for ads or use it to train AI models, and the app runs no AI on your data.

How do I delete my data?

Revoke Google access in your Google Account and remove your Bing key in the app to stop all reading. To delete everything, email austin@buildleansaas.com and we delete your account's data within 30 days. The details are in the Privacy Policy.

Which agents work with it?

The app has the steps for Claude, Claude Code and Cursor. Other MCP clients that support sign-in over OAuth should work, but we haven't tested them yet.

Get on the list, and we'll email you when ScaleSleek opens

Join the waitlist