01 / 11
Scroll

Attractions and leisure. Franchise food. Home services. Professional firms. Anywhere the money runs through more than one system.

Show me →

What we actually look at

Two files from one business. Same five weeks.

On the left, how many bookings came in each week. On the right, how much was spent on ads each week.

Bookings — per week

  • 04 Mar1,182
  • 11 Mar1,447
  • 18 Mar1,096
  • 25 Mar1,613
  • 01 Apr1,338

Ad spend — per week

  • 04 Mar$8,410
  • 11 Mar$9,255
  • 18 Mar$8,972
  • 25 Mar$11,040
  • 01 Apr$9,613

Both files are correct. Neither one can tell you which spend produced which booking.

What reconciling them showed

Cost to bring in one paying customer

$37Channel A · 4.1× return
$615Channel B · 0.27× return

Drawn to actual scale — channel B cost 16.6× more

Both numbers existed. Nobody had ever put them on the same page.

Why nobody could have known.

These are the column headings inside each of those two files.

bookings_mar.csv
booking_refcreated_atgrosssource_label
ads_export_mar.csv
campaign_iddatespendutm_source

● Not one heading appears in both files

Different identifiers, different date formats, nothing in common to match a booking to the spend that caused it. Every report either system produces is honest. The join is the part nobody built.

So we build the join.

One reconciled view where every booking can be traced to what it cost to win.

01

Find it.

We go through your own data — bookings, ad spend, card takings, CRM — and reconcile it. Fixed scope, agreed before we start. You keep the findings whatever you decide next.

02

Fix it.

Then we build what closes the gap. Sometimes an automation. Sometimes custom software. Sometimes AI does the heavy lifting; sometimes it has no business being there. The problem decides the tool.

03

Run it.

Then we operate it for you. Every month you get what ran, what it saved, and what we found.

A report that took four days now lands in four minutes.

Proof

A Texas adventure park

Found
$37 per purchase on one channel, $615 on another.
Fixed
Rebuilt the join so bookings, spend and takings reconcile.
Running
Monthly, since.

$605K – $1.15M identified

A drive-thru QSR franchise, Kentucky

Found
Staff acted on 31.6% of upsize chances. Said aloud: 36%. Unsaid: 5.7%.
Fixed
Built a pipeline that scores every order automatically.
Running
Yes.

90.6 F1 vs 87.1 human, ~1/20th cost

SnapAI — our own product

Found
HVAC technicians misdiagnose faults; the homeowner pays for it.
Fixed
Built the diagnostic platform end to end.
Running
Live, with beta users in Houston.

Next.js · FastAPI · Supabase

Delivered in

Attractions & leisureFranchise food HVACLegalEcommerce Real estateAmazon agencies Finance

Roughly, what's it worth finding?

$18,000 – $42,000 / year

Assumption: where channels are never reconciled, 15–35% of spend typically sits on channels that don’t return their cost. Applied to your figure over twelve months. This is an order of magnitude, not a quote — the audit produces the real number, and sometimes the real number is zero.

Who we are

A small practice that works on other people's numbers.

Ten of us, across analysis, automation, software development and marketing. Our team holds an MSc in Finance, Mathematics & Statistics, a BSc in Mathematics & Quantitative Analysis, and IBM Data Science certification. 3,400+ hours delivered across 26 engagements.

AnalysisAutomationSoftwareMarketing

Based in Karachi. Working with clients in the US, UK and the Gulf.

Send us the two files that never line up. We’ll tell you what we’d look for.

Send a note

Tell us what doesn't add up.

Most people reach us on WhatsApp, or just book the call. If you would rather write it down in a form, say so — we’ll build it.