Danny Deepak
← All casesLuxury-watch marketplaceGlobal, remotePast role

EveryWatch

New users grew 20× in 17 months. Organic impressions grew 55×. The next lever: bringing them back.

20×

monthly new users, Jan 2024 (9,113) → May 2025 (182,782), GA4 Measured

A luxury wristwatch on a dark display
Stock photo (Unsplash) · illustrative
0.10M → 5.5Morganic impressions a month
10.2%returning users, May 2025
~$1.5kpaid budget a month, held small
1.33Mnew users in 17 months
Level 1

The 60-second story

For the CEO. Read left to right.
  1. 1Problem

    Monthly new users sat at 4.5K-17.7K through May 2024. Returning users stayed near one in ten as the base grew.

  2. 2The leak

    Returning users, and measurement: paid leads inflated, revenue undefined, October 2024 totals broken. Revenue links are missing.

  3. 3How I found it

    Back-solved paid spend from leads and ROAS. Compared GA4 users with Search Console impressions month by month.

  4. 4What I did

    Built reference-level content pages, ran paid at a small fixed budget, cut spend when lead cost rose. All claimed in narrative.

  5. 5What changed

    New users 9,113 to 182,782 (Jan 2024 to May 2025, GA4). Returning share 7.0% to 10.2%. No revenue claimed.

  6. 6Proof

    GA4 users series (17 months) and Search Console impressions (19 months). Paid series is lead counts, not revenue.

  7. 7Lesson

    A 34x ROAS was leads times $15. Define one revenue event before reporting a return. Audit any rate above 100%.

Level 2

The business first

Before any channel: what they sell, to whom, and how money comes in.
Business model
Two-sided marketplace: price and listing data for watch collectors (demand) and dealers (supply).
Industry
Luxury and collectible watches, secondary market
Target customer
Collectors researching fair price, provenance and availability. Dealers listing inventory.
Geography
Global; I worked remotely from Bangalore. Catalogue provenance line says 'India operation'.
Commercial objective
Grow collector audience, then returning users, then dealer supply. Subscription or fee revenue.
How revenue is made
Unclear. Catalogue Folio V lists listing fees, dealer subscriptions or data API as candidates and says the product line in scope is undefined.
How they convert
Sign-up, saved search and alerts, dealer handoff. Lead definition for Google Ads: DATA NOT AVAILABLE (counts inflated by multi-fire conversions).
01

What the customer sees: before and after

The same touchpoint on a phone, before my work and after it. Screens are schematic; every change and number comes from the account, the change log or the client report.

Searching a watch reference numberDoneMeasured
Before
rolex 126610ln price history
other sitesDealers and forums

Few EveryWatch pages for reference-number intent. 0.10M impressions a month (Jan 2024).

After
rolex 126610ln price history
everywatch.com › referenceReference page · prices and history

Reference-level pages. 5.5M impressions a month at peak (Jun 2025).

02

What the history shows

Patterns found by re-reading the full history, not just the months I ran. Each is labelled: measured pattern, correlation, anomaly or seasonality.

Anomaly

The June 2024 step is not in organic impressions

New users rose 7x in one month. Organic impressions rose 26%. New users per 1M impressions went from 4.3K (May) to 24.3K (Jun) to 43.1K (Aug). The step is a change in yield, not in search visibility. The page credits the content engine without a click or channel series.

The numbers

New users 4,538 to 32,014 (x7.05). Impressions 1.05M to 1.32M (x1.26). Users/1M impr: May 4,322; Jun 24,253; Jul 31,944; Aug 43,057. July impressions -42% (1.32M to 0.77M) but users only -23%. May 2024 also shows returning users 857 to 273 (-68%) just before the step.

So whatCannot tell if the June step was content, a new traffic source, or a tracking change. Pull sessions by channel for Apr-Aug 2024 before telling this story.

Measured pattern

Search can explain only part of new users

From Aug 2024 to Apr 2025 new users ran 43K-58K per 1M impressions. At the report's own 1.9% CTR, organic Google clicks are ~19K per 1M. So at most ~33-44% of new users came via Google organic. Jan-Feb 2024 users (91K-118K per 1M) cannot be mostly organic. Yield then fell 40% Jan to May 2025 (57.9K to 34.9K per 1M) while the page says CTR held.

The numbers

ASSUMPTION: CTR 1.9% (page). At 2.57% (1.06M clicks / 41.3M impr) the ceiling is ~44-60%. Impressions +85% Jan to May 2025 (2.83M to 5.23M); new users +11.5% (163,896 to 182,782). Impressions x4.15 and new users x4.10 from Aug 2024 to Apr 2025.

So what'Organic became the largest contributor' is untested. 35-65% of users have no named source. Direct, referral, social and paid need splitting.

Correlation, not causation

Returning share rose with the base, but return rate per acquired user fell

Pooled returning share was 6.0% for Jun-Oct 2024 and 9.6% for Nov 2024 - May 2025. The step came in Nov 2024 (returning x2.58 vs new x1.24) after cumulative new users passed ~290K. Returning users as a share of all prior new users ran ~3.0% in Nov-Jan, then fell to 1.8% by May 2025. The 14% in the text is not in the series.

The numbers

Returning 3,500 (Oct) to 9,036 (Nov). Returning / cumulative prior new: Nov 3.13%, Jan 3.02%, May 1.82%. Early-2024 shares of 12.6-13.7% (Mar-Apr) sit on 6.8K-7.8K total users. Text: 215,636 returning (14%); series sums to 128,063 (9.0% of 1,427,835; 8.8% with Oct total corrected).

So whatShare rises by arithmetic as the base grows. Judge lifecycle work on return rate per cohort, which no source holds. The 18% to 47% claim is unsupported.

Anomaly

October 2024: the flagged series is not the broken one

The source calls the October returning figure (3,500) an anomalous dip. It is not: it sits between Sep 3,289 and Nov 9,036. The broken number is total users 64,274, below new users 87,490. New + returning = 90,990, which fits the trend. Users per 1M impressions with 87,490 is 48.9K, in line with Sep (47.7K) and Dec (50.0K). Same month: paid leads x2.79 on spend +28%.

The numbers

Total 64,274 vs new 87,490 (impossible). Corrected total 90,990. Returning share 3.8% corrected vs 5.4% published. If new users were wrong, yield would be ~35K per 1M, outside the 43K-58K band. Paid leads 1,380 to 3,850; implied spend $1,183 to $1,520; implied CPL $0.86 to $0.39.

So whatFix the total, not the returning figure. The lead step lands in the month of the claimed tag redeploy: consistent with a conversion counting change, unproven.

Correlation, not causation

Paid leads were decoupled from both spend and organic growth

At a flat implied ~$1,520 from Oct 2024 to May 2025, paid leads moved 2,840-4,700 while organic impressions rose 2.2x (Oct to Apr). May tripled leads on the same spend. June's 3.4x spend returned no extra leads. Implied lead cost ran $0.14-$1.10.

The numbers

Oct-Apr: impressions 1.79M to 3.94M (x2.2); leads 3,850 to 4,700 (x1.22); correlation 0.38 (n=7). Including May: 0.79, driven by one point. Leads: May 10,700, Jun 10,600 (-0.9%) with spend $1,521 to $5,146 (x3.4). Implied CPL: May $0.14 (stated $0.10), Jun $0.49, Jul $0.79, Aug $1.10. Only 8 months at ~$1,520 (page says ten).

So whatLead counts track tag behaviour, not budget or organic demand. Spend decisions made on them are unsafe. Back-solved spend assumes $15 per lead (circular).

03

From digital to revenue: the strongest verified chain

Verified for 4 of 6 links. The chain breaks at “Subscriptions or dealer revenue”, so revenue attribution is unavailable beyond that point.

Organic impressions (May 2025)5.23M a monthMeasuredindex.html ORGANIC
Organic clicks1.06M lifetime stated. No monthly series.Measuredindex.html Pillar 01 (stated only)
New users (May 2025)182,782 (GA4)Measuredindex.html USERS
Paid leads (May 2025)10,700, inflated; CVR 214.9% statedMeasuredindex.html PAID, Lot 02
Subscriptions or dealer revenueChain breaks hereData not availablenone
Return on paid spendChain breaks hereData not availableclient monthly history caveats
04

Where the money leaks: the journey

This business's own path from first search to revenue. Tap a stage for the five-step leak card: what was wrong, how I found it, the evidence, what changed, and after.

01Search visibilityImproved02Landing on reference pageImproved03Sign-up / leadFlagged04Saved search and alertsFlagged05Returning userFlagged06Subscription / revenueWorking
01Search visibilityImprovedMarketplace near invisible in search through early 2024.
Stage metricOrganic impressions 0.10M (Jan 2024) to 5.23M (May 2025) a month Measuredindex.html ORGANIC array (Search Console)
  1. 1 · What was wrong

    Marketplace near invisible in search through early 2024.

  2. 2 · How I identified it

    Search Console impressions trend.

  3. 3 · Evidence

    0.10M impressions Jan 2024 Measuredindex.html ORGANIC

  4. 4 · What changed

    Programmatic reference and model pages (claimed). Page counts conflict: 280 pages at ~3 a week (catalogue) vs 'thousands a month' (index).

  5. 5 · After

    5.23M impressions May 2025 Measuredindex.html ORGANIC

02Landing on reference pageImprovedUsers flat at 4.5K-17.7K until June 2024. Plateau at ~155-183K from Jan 2025.
Stage metricNew users 9,113 (Jan 2024) to 182,782 (May 2025); 1,326,488 total Measuredindex.html USERS array (GA4); sum calculated
  1. 1 · What was wrong

    Users flat at 4.5K-17.7K until June 2024. Plateau at ~155-183K from Jan 2025.

  2. 2 · How I identified it

    Monthly GA4 curve.

  3. 3 · Evidence

    May 2024 4,538 to Jun 2024 32,014 (x7.05) Measuredindex.html Lot 01

  4. 4 · What changed

    Content engine scaled from pilot (claimed).

  5. 5 · After

    Jun 2024 new users 32,014 Measuredindex.html USERS

03Sign-up / leadFlaggedLead counts multi-fire. Reported ROAS 34x and revenue $787.9K were leads x ~$15.
Stage metricPaid leads 52,560 (Aug 2024 - Aug 2025) Measuredindex.html PAID array; counts inflated, CVR 214.9% impossible
  1. 1 · What was wrong

    Lead counts multi-fire. Reported ROAS 34x and revenue $787.9K were leads x ~$15.

  2. 2 · How I identified it

    52,560 x $15 = $788,400, matches 'attributed' $787.9K.

  3. 3 · Evidence

    52,560 x $15 = $788,400 vs $787.9K stated Calculatedclient monthly history caveats; index.html Figure 04

  4. 4 · What changed

    Figure retired. Not used here.

  5. 5 · After

    DATA NOT AVAILABLE (no clean conversion count) Data not availablenone

04Saved search and alertsFlaggedLifecycle layer described but unmeasured. Catalogue claims returning share 18% to 47%; series maxes at 13.7%.
Stage metricData not availableno saved-search or alert counts in any source
  1. 1 · What was wrong

    Lifecycle layer described but unmeasured. Catalogue claims returning share 18% to 47%; series maxes at 13.7%.

  2. 2 · How I identified it

    Series vs claim comparison.

  3. 3 · Evidence

    Max monthly returning share 13.7% (Mar 2024); 10.2% May 2025 Measuredindex.html RETURNING array; calculated share

  4. 4 · What changed

    Saved-search alerts specced and described as built (claimed, no data).

  5. 5 · After

    Data not availablenone

05Returning userFlaggedShare stayed ~5-10% while base grew 20x.
Stage metric10.2% of users returning, May 2025 (20,867 of 203,649) Calculatedindex.html RETURNING array
  1. 1 · What was wrong

    Share stayed ~5-10% while base grew 20x.

  2. 2 · How I identified it

    Returning / total per month.

  3. 3 · Evidence

    Pooled share 6.0% (Jun-Oct 2024) vs 9.6% (Nov 2024 - May 2025) Calculatedindex.html RETURNING array

  4. 4 · What changed

    Nothing proven. Lifecycle build was the next-step recommendation.

  5. 5 · After

    Data not availablenone

06Subscription / revenueWorking
Stage metricData not availableno subscription, dealer or GMV data
05

Before, what I found, action, after

Verified numbers only. Where an after number does not exist yet, it says so.

Monthly new users (GA4)
Before9,113 (Jan 2024) Measuredindex.html USERS
What I found

Flat 4.5K-17.7K until June 2024. Marketplace near invisible in search.

Action

Reference-level content pages, scaled from pilot (claimed).

After182,782 (May 2025) Measuredindex.html USERS
Organic impressions a month
Before0.10M (Jan 2024) Measuredindex.html ORGANIC
What I found

Few indexable pages for reference-number intent.

Action

Programmatic model and reference pages.

After5.51M peak (Jun 2025); 5.44M Jul 2025 Measuredindex.html ORGANIC
Returning share of users
Before7.0% (Jan 2024) Calculatedindex.html RETURNING
What I found

Stayed 5-10% while acquisition grew 20x.

Action

None proven. Lifecycle layer recommended.

After10.2% (May 2025) Calculatedindex.html RETURNING
Paid spend response (Jun 2025 test)
Before$1,520 spend, 10,700 leads (May 2025) Measuredindex.html Lot 02, PAID
What I found

Leads did not follow spend.

Action

Raised spend to $5,140, then cut to $1,140 by Aug.

After$5,140 spend, 10,600 leads (Jun); $1,140, 1,040 leads (Aug) Measuredindex.html Lot 02-03

Not a marketing problem

What the data shows ads cannot fix. Saying so early saves budget.

  • Revenue definition is a product and finance issue. Source cannot say if 'revenue' is GMV, lead value or pipeline.
  • Dealer supply grew slower than collector demand (catalogue Trade-off 04). No dealer counts or listings data in any source.
  • Conversion double-counting (CVR 214.9%) is a tracking setup issue, partly marketing-owned.

Data not available

What would prove more, and does not exist in the sources yet.

  • Sessions and users by channel (organic, direct, referral, social, paid).
  • Search Console clicks and CTR by month. Only lifetime 1.06M clicks stated.
  • A single deduplicated conversion event; sign-ups, saved searches, accounts.
  • Subscription, dealer or listing-fee revenue. Any money received.
  • Cohort retention tables by acquisition source and month.
  • Raw GA4, Search Console and Google Ads exports. All series are as published in an earlier case study.
  • Oct 2024 GA4 total users and any tag change log for Oct 2024.
  • Meta and PMax spend. Catalogue names them; no series exists.
  • My exact start date, end date and what he built versus what the team built.
  • Spend for Jul 2025 (not stated; implied $2,421 only).
Level 3

Implementation and data

For the team that has to run it: every action, test, date and dataset.
06

What I actually did

6 actions found in the project evidence, 5 done or live. Grouped by type, each with its source.

Reporting 2

  • Live

    GA4 + GTM + Looker Studio weekly brief (claimed).

    catalogue.html Folio II; index.html Regulator (narrative only)
  • Done

    Retired the 34x ROAS and $787.9K revenue claim after audit.

    earlier portfolio notes everywatch fails; client monthly history caveats

SEO 1

  • Live

    Programmatic reference and model deep pages, scaled from pilot. Page count conflicts: 280 reference pages (catalogue) vs thousands a month (index).

    index.html Lot 01, Going Train; catalogue.html Lot 27 (narrative claim, no page counts in data)

Paid media 1

  • Done

    Google Ads at small fixed budget (~$1.5k a month implied, Oct 2024 - May 2025). Spend raised Jun 2025, cut Jul-Aug.

    index.html Lot 02-03; implied spend back-solved

Campaign restructure 1

  • Done

    Paid account rebuilt by reference family; broad terms defunded (claimed).

    catalogue.html Lot 11 (narrative only)

Automation 1

  • Specced

    Saved-search and alert lifecycle, described as built in catalogue; index calls it the next build.

    catalogue.html Lot 43 vs index.html Lifecycle map; sources conflict
07

Tests and what they taught

Not every test worked. Each one changed the next decision.

  1. Hypothesis

    Reference-level pages win long-tail search demand.

  2. Action

    Piloted programmatic pages for specific reference numbers and comparisons.

  3. Result

    Claimed: indexed pages 10x prior coverage in four weeks, CTR 1.8-2.4%. No page or click series in sources.

  4. Learning

    Impressions grew 0.10M to 5.5M. The June user step is not explained by impressions.

  5. Next decision

    Channel split for Apr-Aug 2024.

  1. Hypothesis

    More paid budget captures the May 2025 peak.

  2. Action

    Raised spend $1,520 to $5,140 in June 2025.

  3. Result

    Leads 10,700 to 10,600 (-0.9%). Implied CPL $0.14 to $0.49.

  4. Learning

    Leads did not follow spend. May count was likely inflated.

  5. Next decision

    Cut to $1,140 by August. Done.

  1. Hypothesis

    Lead cost drift in Jul-Aug 2025 is creative decay or audience saturation.

  2. Action

    Isolated creative refresh without a budget cut.

  3. Result

    Lift 'marginal' (narrative). No data.

  4. Learning

    Source concludes audience exhaustion. Untested.

  5. Next decision

    Hold spend at floor; build a new audience layer.

  1. Hypothesis

    Saved-search alerts raise returning users.

  2. Action

    A/B test of save-search CTA position (claimed).

  3. Result

    DATA NOT AVAILABLE

  4. Learning

    Series shows 10.2% returning in May 2025, not the claimed 47%.

  5. Next decision

    Cohort return rates by acquisition source.

08

Timeline: interventions and events

My actions sit alongside what the client, the previous team and the market did. Nothing here claims a cause.

  1. Client

    Series starts: 9,113 new users, 0.10M organic impressions. My start date is inferred.

    index.html USERS, ORGANIC
  2. Market

    One-month spike to 17,660 users (x1.94 vs Jan), then 6,716 in March. Not explained.

    index.html USERS
  3. Client

    Trough: 4,538 new users, returning users 273 (-68% vs Apr).

    index.html USERS, RETURNING
  4. DannyI join

    New users jump to 32,014 (x7.05). Reference-page content engine credited in narrative.

    index.html Lot 01 (claimed); USERS
  5. Danny

    Paid series begins: 1,420 leads, implied spend ~$1,164.

    index.html PAID; implied spend back-solved
  6. Danny

    Paid leads x2.79 on spend +28%. Published total users (64,274) falls below new users (87,490). Source claims a tag manager redeploy in early October (candidate cause, unverified).

    index.html PAID, RETURNING; catalogue.html Folio V caveat 01
  7. Market

    Returning users step 3,500 to 9,036 (x2.58).

    index.html RETURNING
  8. Market

    New users reach 163,896. Growth flattens (Jan to May +11.5%) while impressions +85%.

    index.html USERS, ORGANIC
  9. Danny

    Peak: 182,782 new users, 10,700 paid leads, $1,520 spend.

    index.html Lot 02
  10. Danny

    Paid spend raised to $5,140. Leads stay at 10,600.

    index.html Lot 02; PAID
  11. Danny

    Spend cut to $1,140 after lead cost rose to $1.10. 1,040 leads.

    index.html Lot 03
09

The data, month by month

Two clear steps up in acquisition. A flat line underneath.

The June 2024 break: 4,538 → 32,014

Monthly new users. Reference-level content went live in June. Measured

Returning users stayed near one in ten

Returning ÷ total users each month. Peak 13.7% in Mar 2024, 10.2% in May 2025. Calculated

Organic impressions, millions a month

CTR held 1.7–2.5% as impressions grew. Measured

Tripling paid budget didn't triple leads

June 2025: $5,140 spend. August cut to $1,140. Measured

10

What I'd do next

SourcesMonthly GA4, Search Console and Google Ads series as compiled for the earlier EveryWatch case study. Raw exports were not re-checked for this page.Evidence labels: Measured, Calculated, Estimated, Assumption, Data not available.
← Previous caseForemost Public Adjusters0 / 247 · live keywords that were branded
Next case →Swift Restoration0 · service or location pages at audit

Work with me

Seen a leak in your own numbers?

Send me the symptom. I'll tell you which case is closest and what I'd check first. No deck, no retainer pitch.

danieldeepak3519@gmail.comLinkedInFree 12-point check