Methodology
How this works
The audits hold your numbers against published benchmarks and answer in four words. This page is the whole method: what each word means, how an area is judged, what we send and store, and every threshold the engine runs on with its source.
The four words
Calculator audits use these four verdicts. SEO reports group individual findings as errors, warnings, notices and passes.
- Dead
- A decisive test failed. Money is going into this area and not coming back, and until it is fixed everything else you spend here lands in the same hole.
- Dying
- The number is near the benchmark but on the wrong side of it. Nothing has collapsed yet. Another quarter like this one and it will be dead.
- Alive
- Above the benchmark. This area does not need fixing. It is the one that deserves more of your money and time.
- Not judged
- The number was missing, or you said you did not know. We do not fill the gap with a guess, so the area is reported as unjudged and the score leaves it out.
How a verdict is reached
The engine is a set of named tests, not a weighted opinion. Four rules govern all of them.
One decisive test can mark a whole area dead
Each area has a few supporting tests and one or two decisive ones. A failed decisive test is fatal on its own: if ROAS is under break-even, paid ads are dead however good the creative is, and however many other tests pass. Averaging would hide exactly the thing you came here to find.
Supporting tests move an area between dying and alive
No supporting test can kill an area by itself, and none can save one that failed a decisive test. They decide the middle ground: three creatives a week instead of one, a share rate that reaches beyond your followers, a canonical tag that exists. Each one names the number it compared and the benchmark it used.
Not knowing is a real answer
Every question in the marketing audit has a Don't know option, and choosing it costs you nothing. We judge what you gave us and list the rest as Not judged. An area left unjudged is left out of the score rather than scored as zero, because a wrong verdict is worse than an honest gap.
Fix-first is ordered by money, then by effort
Every dead or dying area carries an estimate of the rupees exposed each month. The audit ranks by that number and breaks ties with how fast the fix is, so the first item on the list is the one where the delay costs most.
What leaves your device
We keep only the numbers and the score, never who you are on our side. The two tools differ in one respect: the marketing audit never sends anything at all, and the SEO audit sends one public URL because it has to fetch that page.
Marketing Audit — nothing is sent
The whole calculation runs in your browser. There is no submit button and no API call, because there is nowhere for the answers to go.
- Where it runs
- Your own browser, on your own device
- Where answers live
- Packed after the # in the result link, which browsers never send to a server
- What we store
- Nothing. Delete the link and the record is gone
- Who can open it
- Anyone you send the link to, and nobody else
SEO Audit — one public URL is sent
Our own fetcher reads the address you typed, exactly as a browser would. No third-party API is involved, so your site is never handed to anyone else.
- Identifies as
- LapaasAuditBot/1.0
- Refuses
- Private, loopback and link-local addresses, checked again on every redirect hop
- Caps each fetch
- 2 MB, 10 seconds, at most 5 redirects, ports 80 and 443 only
- Rate limit
- 10 scans an hour per IP address, held in memory
- Cache
- The fetch result for 24 hours, keyed by the URL alone — nothing about the visitor
The deep pass: the same page in a real browser
The SEO scan reads the HTML your server sends, which is what a crawler gets on its first pass. Signed in, you can ask for one more thing: the same address loaded in a real Chrome on our server, with Lighthouse run against it. It takes up to a minute and you get ten a day, because a browser costs real memory. The fast scan is the product; this is the second half of the picture.
What the render adds
Nine more checks, scored alongside the rest rather than shown beside them, so a JavaScript gap moves the on-page score the way any other finding does.
- The JavaScript gap
- Words after rendering against words in the raw HTML. Over 40% warns, over 75% fails — at which point every content verdict in the fast scan was reading a shell
- Core Web Vitals, in the lab
- LCP good under 2.5 s, CLS good under 0.1, Total Blocking Time good under 200 ms as the stand-in for INP
- Lighthouse
- The performance, accessibility, best-practices and SEO scores, and the opportunities with the milliseconds each claims
- What broke on the way
- JavaScript errors thrown on load, and every request that failed or 404'd while the page drew
- Proof it is the site
- A few more pages rendered, so the same gap can be shown to be the template and not one page
- A picture
- The finished page photographed on a phone
What it still cannot see
One browser, one load, one moment, on our network. That is a lab measurement, and a lab measurement is not your customers.
- Field data
- Real LCP, INP and CLS at the 75th percentile — the numbers Google actually ranks on — exist only in your own Search Console and in CrUX
- Your visitor's phone
- We render on a throttled profile on our connection, not on a four-year-old handset on 4G in your city
- Anything behind a login
- The browser signs in to nothing, so a members' area or a checkout past the cart stays invisible
- Anything that needs a click
- The page is loaded and measured, not used. Content that appears only on a tap, a scroll or a hover is not counted
- Your rankings and your traffic
- Nothing here reads Search Console, Analytics or a rank tracker. Every number is measured off the page itself
- How often
- 10 renders a day per account. The fast scan stays free, has no limit and needs no account
Every threshold the engine runs on
These tables are rendered from the same constants the engine imports, so this page cannot drift from the verdicts. Rows marked Our call are thresholds we set ourselves where no study gives a usable number; everything else names its study.
Organic click-through rate by position10
| Position | Organic CTR | Clicks per 1,000 impressions | Source |
|---|---|---|---|
| #1 | 39.8% | 398 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #2 | 18.7% | 187 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #3 | 10.2% | 102 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #4 | 7.2% | 72 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #5 | 5.1% | 51 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #6 | 4.4% | 44 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #7 | 3% | 30 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #8 | 2.1% | 21 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #9 | 1.9% | 19 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| #10 | 1.6% | 16 | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
Between two positions the expected rate is interpolated, and past position ten it decays with a floor of 0.2%. That expected rate is what your real CTR is measured against.
Search4
| What the engine checks | Threshold | Source |
|---|---|---|
| Searches that end without a click on anything | 68% | SparkToro via Search Engine Land, 2026 |
| The same figure when an AI Overview is shown | 83% | SparkToro via Search Engine Land, 2026 |
| Your CTR against the CTR your average position should earn — dead below | 50% | Our callLapaas threshold — our own judgment, not a study. |
| The same comparison — dying below | 80% | Our callLapaas threshold — our own judgment, not a study. |
Paid ads5
| What the engine checks | Threshold | Source |
|---|---|---|
| India Facebook CPM, top of the usual range | ₹400 | Sahil Khanna, 'Your Marketing Is Officially Dead' (2026) · Lapaas threshold |
| Sales-campaign CPM where the alarm starts | ₹700 | Sahil Khanna, 'Your Marketing Is Officially Dead' (2026) |
| New creatives tested every week, minimum | 3 a week | Our callLapaas threshold — our own judgment, not a study. |
| ROAS of the top Indian D2C advertisers, average | 3.2× | Meta ROAS benchmarks, Indian D2C, 2026 |
| LTV:CAC a business needs to stay viable | 3.9× | Indian D2C unit-economics reports, 2026 |
Social10
| What the engine checks | Threshold | Source |
|---|---|---|
| Instagram reach, share of followers a post reaches at 10–50K | 5.5% | Socialinsider, Social Media Reach Statistics, August 2026 · 872,075 posts |
| Instagram interactions, share of followers at 10–50K | 0.55% | Socialinsider, Social Media Video Statistics 2026 · 111,000 business-page videos |
| Instagram interactions, share of the people reached | 5.5% | Buffer, State of Social Media Engagement 2026 · 52M+ posts across 200,000+ accounts |
| Facebook reach, share of followers a post reaches at 10–50K | 2.7% | Socialinsider, Social Media Reach Statistics, August 2026 · 872,075 posts |
| Facebook interactions, share of followers at 10–50K | 0.2% | Socialinsider, Social Media Video Statistics 2026 · 111,000 business-page videos |
| Facebook interactions, share of the people reached | 5.6% | Buffer, State of Social Media Engagement 2026 · 52M+ posts across 200,000+ accounts |
| Instagram shares, share of followers per post | 0.1% | Socialinsider, Instagram Engagement Report, May 2026 · 15M posts across 417,130 pages |
| LinkedIn impressions a post takes on a 10–100K page | 2,321 | Metricool, LinkedIn Study 2026 · 673,658 posts across 63,108 accounts |
| YouTube interactions, share of subscribers | 0.21% | Quid (formerly Rival IQ), 2026 Social Media Industry Benchmark Report |
| Posts about the product or an offer, no more than | 60% | Our callLapaas threshold — our own judgment, not a study. |
Execution1
| What the engine checks | Threshold | Source |
|---|---|---|
| Agencies already running on generative AI | 9 in 10 | Forrester Research, agency use of generative AI |
SEO Audit, read off your page13
| What the engine checks | Threshold | Source |
|---|---|---|
| A noindex tag, or robots.txt blocking the site — dead on its own | Present | Google Search Central documentation |
| Time to first byte — dying above | 0.8 s | Core Web Vitals thresholds, web.dev |
| Time to first byte — dead above | 1.5 s | Our callLapaas threshold — our own judgment, not a study. |
| HTML document — dying above | 400 KB | Our callLapaas threshold — our own judgment, not a study. |
| HTML document — dead above | 1 MB | Our callLapaas threshold — our own judgment, not a study. |
| Scripts on the page — dying above | 15 | Our callLapaas threshold — our own judgment, not a study. |
| Sampled images, total weight — dying above | 2 MB | Our callLapaas threshold — our own judgment, not a study. |
| Title length | 30–60 characters | Our callLapaas threshold — our own judgment, not a study. |
| Meta description length | 70–160 characters | Our callLapaas threshold — our own judgment, not a study. |
| Words on the page — dead below | 100 | Our callLapaas threshold — our own judgment, not a study. |
| Words on the page — dying below | 300 | Our callLapaas threshold — our own judgment, not a study. |
| Images with no alt text — dying above | 30% | Our callLapaas threshold — our own judgment, not a study. |
| Words between headings, for a passage an AI can quote — dying above | 300 | Our callLapaas threshold — our own judgment, not a study. |
Sources in full8
| Used for | Source |
|---|---|
| Organic click-through rate at each position | First Page Sage, Google Organic CTR by Ranking Position, updated 28 May 2025 |
| The share of searches that end without a click | SparkToro via Search Engine Land, 2026 |
| CPM ranges for Indian Facebook and Instagram ads | Sahil Khanna, 'Your Marketing Is Officially Dead' (2026) · Lapaas threshold |
| The LTV:CAC ratio a business needs to survive | Indian D2C unit-economics reports, 2026 |
| What good ROAS looks like for Indian D2C | Meta ROAS benchmarks, Indian D2C, 2026 |
| How many agencies already run on generative AI | Forrester Research, agency use of generative AI |
| Our own thresholds, used where no study gives a number | Our callLapaas threshold — our own judgment, not a study. |
| The sales-campaign CPM the talk treats as the alarm point | Sahil Khanna, 'Your Marketing Is Officially Dead' (2026) |