Cookiebees for marketing teams
Better signals in.
Better decisions out.
The measurement layer your ad platforms were supposed to give you: every order credited on real identity, audiences built from what the store already records, and every conversion sent back with the parameters bidding actually learns from.
A week in performance marketing
Same ad accounts. Different information.
Mon
Attribution by channel on four models, and Customer said as the fifth. The gap between Meta’s number and yours, explained by assists.
Tue
Build an audience: browsed twice, never bought, high predicted value. Previewed against what Meta can match before it syncs.
Wed
A custom signal: purchases over ₹10,000 by first-time buyers. Twenty fields, any condition, fans out to every connected platform.
Thu
Creatives: which ad, not which campaign, brought the buyers who came back. Funnels by campaign show where each one leaks.
Attribution
Four models, one set of numbers, and the customer’s own answer beside them.
Last click, first touch, linear and all touches, computed on identity-stitched journeys rather than on a pixel’s guess. After every order, one question over WhatsApp or email; the answer becomes a credit model of its own.
- Click ids kept past URL-stripping browsers
- Assists, and the line that explains the gap with Meta
- Click a campaign for its day-by-day trend
- Platform performance on each platform’s own traffic
Revenue
₹0
Attribution coverage
0%
Avg match quality
0/100
Google Ads
Paid search
Meta Ads
Paid social
Owned
Organic search
Earned
Direct
Unknown
Audiences
Segments from behaviour the store already records, sized against what Meta can match.
RFM, acquisition channel, cross-sell, cadence, location, negation and no-purchase conditions. Every segment states the identity wall beside its count, so you never sync an audience Meta will match at 12%.
- Ranked on cost per add-to-cart, which spend actually joins to
- Previewed before the sync, refused before the click when it cannot help
- Who buys next: a propensity score calibrated nightly per store
- Playbooks per audience: suppress buyers, seed lookalikes, nudge the rest
Signals back
44 parameters on every conversion, and the payload open before it leaves.
Which order this is for that person, lifetime value, prepaid or COD, the discount, every channel in the path, how long they took. Attached to the Purchase, so custom conversions inherit the full history. Shadow mode shows exactly what would be sent.
- Meta CAPI, Google Ads via Data Manager, GA4, server GTM
- Value on the wire as what they paid, or profit after costs
- Deduplicated on your own order id, pixel-safe
- Match quality scored per event, missing field named
Your pixel sends the order. This is what we know about it.
Built on one parameter, in Meta
Purchase where is_first_purchase = true → CB First-time buyers · cost per genuinely new customer, per campaign
The reports
Twenty reports, one date window, every one ending in a decision.
Channel roles, funnels by campaign, prospecting new against returning, creatives, timing, retention by channel. One date picker in the top bar governs all of them, prior periods are counted rather than projected, and a comparison that cannot be made is said, not substituted.
- Every column in plain English, every explanation findable
- Drill from channel to campaign to ad set to ad
- Spend has full depth; revenue says where it does not
- Every number checkable from another screen
brand search
₹33,480 spent · ₹1,44,720 back
cheapest growth
retargeting · DPA
₹37,350 spent · ₹1,43,460 back
well past the line
magnesium search
₹55,890 spent · ₹1,37,430 back
profitable
pmax · bestsellers
₹1,13,130 spent · ₹2,33,280 back
just above it
prospecting · advantage
₹1,62,180 spent · ₹1,57,320 back
under break-even
generic supplements
₹1,46,610 spent · ₹39,420 back
loses every order
And the rest
What else the team gets
Everything reads the same identity, so a segment built on Tuesday is the audience the conversion on Thursday was scored against.
Signal health
- Click-id coverage per platform
- Where the lost ones landed
- A signal-issue verdict with cause, impact, action
- Alerts the day a tag breaks
Event stream
- Every delivery and what happened to it
- Per-destination shadow switches
- Retry rules taken from Meta’s contract
- Resend what was held while sending was off
Custom signals
- Twenty fields across four groups
- Any condition, one-click recipes
- Fans out to every connected platform
- Trial-run suffix for Meta event names
First-party that survives
- A path on your own domain, no CNAME
- Server-set cookies, 400 days
- Meta and Google cookies re-issued at their own lifetimes
- Traffic-integrity scoring on every page view
Catalog and product
- Categories as reported, per product
- Product IDs mapped per destination
- Top products with share of revenue
- Frequently bought together
Engagement signals
- Email opened and clicked, sent as real events
- WhatsApp replies and button taps
- Every tracked link tap with its real destination
- Never a faked page view
Every claim on this page has a 'show me the actual thing' behind it. Open one from the homepage, or connect a store and open your own.
Send the ad platforms something worth learning from
Better signals in.Better decisions out.
Connect Meta and Google in two popups, install one snippet, and watch the first enriched Purchase leave with its payload open.