AMANYR
Use cases

Seven engagements, and the number each one ended in.

The client's own figures, one card per engagement. The big number is what the work returned. Open a card for how we got there.

$250k
incremental demand · first 13 days

Product walls sorted by a scoring model instead of by hand: $250k of incremental demand in the first 13 days.

+10%
add-to-cart rate
$8M
a year, estimated at scale
Daily
every wall reordered
420 h
returned every year

A legally required quarterly report rebuilt as one pipeline: 300 spreadsheets retired and 420 hours a year returned.

300
Excel files → 1 pipeline
310
buildings
1
definition per KPI
−50%
time per analysis

An AI agent on GA4 and Google Ads answers the team's questions directly: half the time per analysis, and a queue that stopped forming.

1
entry point for reporting
2
sources: GA4, Google Ads
Anyone
on the team can ask
Aug 2025
standing analytics partner since, still running

Analytical support and reporting for the build of new web and app features: measurement plans, ad-hoc analysis and strategy workshops.

3
tools under one plan: GA4, GTM, Amplitude
Go / No-Go
at the end of every A/B test
1
standing analytics forum for the whole team
1 holistic view
of how each campaign performed, every team and channel

Too many marketing teams on one campaign for anyone to say how it had performed. One shared tree of KPIs gave leadership a top-level read of every campaign, across channels and against the last one.

1
scorecard every marketing team reads
All
channels added up, none counted twice
Each
campaign judged against the last one
88.1K
devices on the World Cup Predictor, measured end to end

A video app turned fantasy app, with no shared way to judge it. One KPI pyramid from ESPN's priorities down to the app events, so every sprint and the World Cup campaign were read in the same numbers.

38.9K
peak active devices on predictions
20.4K
onboardings driven by the campaign
1
measurement plan, app and web
1 KPI tree
from strategy down to the levers

Scattered reports and one-off analyses, replaced by one hierarchy of KPIs the whole business reads the same way.

1
owner per KPI
1
place the numbers live
4
tree reviews a year

NikeHow we got there

The situation

  • Product walls ordered by hand, product by product
  • Good instinct, never measured
  • Only the walls one team could reach

What we did

  • Turned the merchandisers' own rules into scored KPIs.
  • Built a statistical model that ranks every product on every wall against them, daily.
  • Tested on live traffic for 13 days against hand-sorted walls.

The decision

  • Sort the full catalogue by the model
  • Merchandisers review exceptions, they no longer order
  • $8M of demand a year estimated at scale, on the test's uplift

Next lever

  • An agent pushing the sort to the site directly
  • Stock depth and margin added to the score

CorumHow we got there

The situation

  • One file per building every quarter, 310 of them
  • Occupancy counted differently by each team
  • The number changed with the person asked

What we did

  • One pipeline from Yardi, the property system, into Power BI.
  • One written definition per KPI, agreed across teams.
  • Numbers readable any week of the quarter, not only at the deadline.

The decision

  • Manual collection retired
  • Occupancy steered during the quarter, not discovered at the close

Next lever

  • The pipeline feeding the published report itself
  • A knowledge centre so every team reads the numbers the same way

HypersolidHow we got there

The situation

  • Reporting spread across GA4, Google Ads and one-off exports
  • Every question routed through one analyst
  • Support requests queued, people waited or guessed

What we did

  • A measurement plan first, so the tracking tells one story.
  • Claude connected directly to GA4 and Google Ads.
  • A pilot interface anyone on the team can log into and ask.

The decision

  • Support questions go through the agent first
  • The analyst keeps the ones that need one
  • No new source before the measurement plan covers it

Next lever

  • Answers shaped for the least technical user
  • Adoption decided in the interface

HypersolidHow we got there

The situation

  • Several sites and apps, each tracked its own way across GA4, GTM and Amplitude
  • Tests ran, then ended in a debate
  • Product teams launched and iterated without a shared definition of success

What we did

  • One tagging plan per site and app. Every implementation validated before go-live, data quality checked continuously.
  • A/B tests designed, run and read the same way, each closed with a written Go/No-Go.
  • Success and quality KPIs defined once, with a recurring report that guides launches, experiments and iterations.
  • Workshops and a standing analytics forum, so non-specialists read their own numbers.

The decision

  • Nothing ships without its event in the measurement plan
  • A test ends in a written Go or No-Go
  • Product reads the recurring report, the analyst reads the exceptions

Next lever

  • Group data strategy for a €1.6bn retail client: headquarters and countries on one set of practices
  • The AI insight agent as the front door for support questions

NikeHow we got there

The situation

  • One campaign, many marketing teams
  • No read across channels, none against the last campaign
  • Each team reporting its own slice, with its own definitions

What we did

  • Mapped the campaign objectives to a hierarchy of KPIs shared by every team involved.
  • Fixed one definition and one source per KPI, so channels add up instead of overlapping.
  • Built the dashboard that reads every campaign the same way.

The decision

  • Every campaign read through the same tree
  • The next one judged against the last, like for like

Next lever

  • A review each quarter as the channel mix moves
  • So the comparison stays honest

ESPNHow we got there

The situation

  • Goal app switched from video to fantasy football, reporting had not followed
  • Events tracked, but no view of which numbers mattered
  • No link between the numbers and ESPN's priorities

What we did

  • KPI pyramid from the strategic goals down to driver KPIs and events
  • Measurement plan written before each feature, events validated in Conviva, mapped into the dashboards
  • Every sprint reported on the same tree, World Cup 2025 Predictor included

The decision

  • One hierarchy of KPIs reflecting ESPN's priorities
  • Reviewed with product and tech every two weeks
  • Every feature judged by the behaviour it was meant to move

Next lever

  • Retention through the season as the fantasy game matures
  • The next KPI the tree points to, once the tooling can read it

Sweaty BettyHow we got there

The situation

  • Reports scattered across teams, analysis mostly one-off
  • One KPI, a different definition per team
  • Nobody could say which number was right

What we did

  • Broke the strategy into a hierarchy of the behaviours the business wants to see.
  • Defined how each one is measured and which drivers move it.
  • Built the pipelines and the dashboards the tree reads from.

The decision

  • One owner per KPI definition
  • One place the numbers live
  • The tree reviewed every quarter as the strategy moves

Next lever

  • Training built into the engagement
  • A framework the business understands gets used, one it does not becomes another dashboard

Delivered by Amanyr's founder, in-house at Nike and in consulting engagements since, with the method that now runs every Amanyr engagement. Figures as measured by the client at the time.

Your situation is probably on this page.

Thirty minutes to find out which line applies to your numbers, and what it would terminate in.

Book a free call See the services