Delivery

Goal setting

How to set a success criterion with a customer, encode it in the product, and track both outcomes and usage so the renewal conversation has evidence behind it.

A pilot without a written goal is a demo that lasts 90 days. This page is the checklist of what to track, module by module.

The one rule

No platform access before a written success criterion. It goes in the POC order, not in someone's notes. Every stalled pilot in the pipeline has the same root cause: nobody wrote down what it was supposed to prove.

What makes a usable goal

Baselined
1

We know the current number. If we do not, capturing it is week one of the pilot, not an afterthought.

Attributable
2

The change can plausibly be caused by us. "Revenue up 20%" fails. "Hours to produce the weekly report" passes.

Owned
3

One named person on the customer side agrees it is their number.

Time-boxed
4

Measured at a specific date, not eventually.

Write it in one line:

"By [date], [metric] moves from [baseline] to [target], measured in [where], owned by [name]."

Pick one primary goal per module and at most two secondary. More than three and nothing gets attention. The ★ rows below are the usual primary candidates.


The checklists

Track everything in the module you sold. Ignore the modules you did not.

AI Designer

Track these7 metrics · ★ = primary candidate
  • Adaptations delivered per month against the agreed capprimary candidate
    Baseline:
    Agency volume today
    Source:
    Library + delivery log
    Cadence:
    Weekly
  • Cycle time from brief to approved assetprimary candidate
    Baseline:
    Agency turnaround, usually about a week
    Source:
    Activity screen
    Cadence:
    Weekly
  • Average iterations to approval
    Baseline:
    Capture in week 1
    Source:
    Activity screen, tracked automatically
    Cadence:
    Weekly
  • Time to approve, and count of delayed assets
    Baseline:
    Capture in week 1
    Source:
    Activity screen
    Cadence:
    Weekly
  • Share of campaigns localised in-platform rather than outsourced
    Baseline:
    0% at start
    Source:
    Campaign list, manual tally
    Cadence:
    Monthly
  • Compliance flags caught before human review
    Baseline:
    n/a, new capability
    Source:
    Compliance Checker
    Cadence:
    Monthly
  • Assets pulled from Library by other teams or agencies
    Baseline:
    0 at start
    Source:
    Library
    Cadence:
    Monthly
Do not make creative quality the primary goal

It is real but unmeasurable in 90 days, and it invites subjective argument at exactly the wrong moment. Measure throughput and cycle time; let quality show up in the approval-without-edit rate.

Brand Moderator

Track these7 metrics · ★ = primary candidate
  • Median response time from inbound to replyprimary candidate
    Baseline:
    Their own SLA, and what they actually hit
    Source:
    Action History
    Cadence:
    Weekly
  • Coverage: share of inbound that gets any replyprimary candidate
    Baseline:
    Usually well under half. Ask, do not assume
    Source:
    Review Queue vs total mentions
    Cadence:
    Weekly
  • Approval rate without edit on AI-drafted replies
    Baseline:
    Week 1 of shadow mode
    Source:
    Review Queue
    Cadence:
    Weekly, should climb
  • Escalation accuracy: false positives such as appreciation creating tickets
    Baseline:
    Shadow-mode week
    Source:
    Action History + client feedback
    Cadence:
    Weekly in month 1
  • Tickets created and correctly routed by category
    Baseline:
    Current ticket volume
    Source:
    Rules Engine + CRM
    Cadence:
    Weekly
  • Channels live vs channels in scope
    Baseline:
    Agreed channel list
    Source:
    Connectors
    Cadence:
    Monthly
  • Queue backlog at end of day
    Baseline:
    Week 1
    Source:
    Review Queue
    Cadence:
    Daily in month 1

Consumer Insights

Track these6 metrics · ★ = primary candidate
  • Share of category voice and category rankprimary candidate
    Baseline:
    Captured at kickoff
    Source:
    Social, Share of Voice tab
    Cadence:
    Weekly, report monthly
  • Net sentiment, overall and per channelprimary candidate
    Baseline:
    Captured at kickoff
    Source:
    Social, Sentiment tab
    Cadence:
    Weekly
  • Time to answer a brand-health question
    Baseline:
    Often "we commission a study" or "wait for the monthly deck"
    Source:
    Ask the brand lead directly
    Cadence:
    At 30 / 60 / 90
  • Sentiment on a named problem theme the brand is actively fixing
    Baseline:
    Kickoff
    Source:
    Social, Topics tab
    Cadence:
    Monthly
  • Topic spikes detected and explained before the brand noticed elsewhere
    Baseline:
    0 at start
    Source:
    Social, Topics tab
    Cadence:
    Monthly
  • Sub-brand sentiment spread, best versus worst line
    Baseline:
    Kickoff
    Source:
    Sub-Brands
    Cadence:
    Monthly

AI Paid Media

Use the product's own Goals object for the first four. It supports CPA, CPC, cost per purchase, CTR, ROAS and purchase or lead volume, scoped to account, platform or campaign, and tracks each as on track, at risk or breached.

Track these6 metrics · ★ = primary candidate
  • Cost per qualified lead, or ROAS at the scope that mattersprimary candidate
    Baseline:
    Their own reporting, reconciled first
    Source:
    Goals object
    Cadence:
    Continuous, reviewed weekly
  • Hours per week the team spends assembling reportsprimary candidate
    Baseline:
    Ask in discovery. Often the whole business case
    Source:
    Customer self-report
    Cadence:
    At 30 / 60 / 90
  • Detection speed: days from a problem starting to someone noticing
    Baseline:
    Often around two weeks. Nobody is embarrassed, everyone is slow
    Source:
    Signals timestamps vs customer awareness
    Cadence:
    Monthly
  • Goal achievement rate month over month
    Baseline:
    Month 1
    Source:
    Goals screen, computed automatically
    Cadence:
    Monthly
  • Signals actioned versus signals raised
    Baseline:
    0 at start
    Source:
    Signals + Tasks
    Cadence:
    Weekly
  • Spend reconciliation variance against their platform
    Baseline:
    Must be near zero before anything is shown to a stakeholder
    Source:
    Manual check
    Cadence:
    Weekly in month 1
Reconcile before you report

5paisa found a material spend discrepancy against Google Ads for the same period. Until variance is near zero, no goal number from this module is safe to put in front of a stakeholder.

Reporting

Track these5 metrics · ★ = primary candidate
  • Manual reports eliminatedprimary candidate
    Baseline:
    Count the decks rebuilt by hand today
    Source:
    Customer confirmation
    Cadence:
    At 30 / 60 / 90
  • Scheduled alerts active versus reports savedprimary candidate
    Baseline:
    0 active at start
    Source:
    Reports screen, "without alert" counter
    Cadence:
    Weekly
  • Deliveries confirmed as received, not just sent
    Baseline:
    First send
    Source:
    Alerts run history plus ask a recipient
    Cadence:
    Weekly
  • Recipients per report, and whether leadership is on it
    Baseline:
    Agreed distribution list
    Source:
    Alerts
    Cadence:
    Monthly
  • Dashboards opened per week
    Baseline:
    Week 1
    Source:
    Usage
    Cadence:
    Weekly

Brand Brain (foundation, every deal)

Brand Brain4 metrics · ★ = primary candidate
  • Graph built and signed off by the clientprimary candidate
    Baseline:
    n/a, binary
    Source:
    Sign-off record
    Cadence:
    Once, by end of week 2
  • Sources connected and fresh, no stale feeds
    Baseline:
    Agreed source list
    Source:
    Sources screen freshness timestamps
    Cadence:
    Weekly
  • Calibration outputs accepted by the client against their own rulebook
    Baseline:
    First calibration run
    Source:
    Client review
    Cadence:
    Once, then on drift
  • Skills created for rules the defaults keep missing
    Baseline:
    0
    Source:
    Brand Brain, Skills
    Cadence:
    Monthly

Tasks (adoption signal, not a goal)

Tasks2 metrics · ★ = primary candidate
  • Suggestions converted into owned tasks
    Baseline:
    0 at start
    Source:
    Tasks
    Cadence:
    Weekly
  • Overdue count
    Baseline:
    Week 2
    Source:
    Tasks counters
    Cadence:
    Weekly, it is the standup number

Track usage, not just outcomes

Where these numbers live

Every usage metric below sits on the product analytics dashboard that PM and SA build by the end of month 1. CS reads it before the weekly call so the conversation starts from evidence rather than from asking the customer how it is going. See Process & Timeline.

Outcome goals tell you whether the pilot worked. Usage tells you whether it is going to. Usage moves first, so it is the earlier warning.

Usage, every account4 metrics · ★ = primary candidate
  • Weekly active users against seats issuedprimary candidate
    Baseline:
    Seats issued in week 1
    Source:
    Admin
    Cadence:
    Weekly
  • Which surfaces get opened
    Baseline:
    Week 2
    Source:
    Usage
    Cadence:
    Weekly
  • Work created: briefs run, replies approved, dashboards built, reports saved
    Baseline:
    0
    Source:
    Per module
    Cadence:
    Weekly
  • Credit or usage consumption against the estimate
    Baseline:
    Estimate at kickoff
    Source:
    Credit balance and usage chart
    Cadence:
    Monthly
The tell that a pilot is dying

Goal metrics flat and usage falling. That is not a product problem yet, it is an adoption problem, and it is fixable in week three in a way it is not in week eleven.


Cadence

  1. Week 1Baselines captured and goals enteredSolution Architect

    SA captures and encodes them in the product. CS agrees them with the customer in writing.

  2. WeeklyGoal status at the top of the callCustomer Success

    CS runs the call, Product brings the numbers. Goals before features. A call that opens with the roadmap and runs out of time is how an account drifts.

  3. Day 30 / 60 / 90Formal checkpointCustomer Success

    CS runs the room against the pilot plan. The economic buyer joins at day 90.

  4. OngoingAdoption watchedProduct

    Product reads the analytics dashboard weekly and flags decline to CS. SA owns getting them enabled; Product owns whether it sticks.


Setting the goal on the sales call

Do not wait for onboarding. Ask in discovery: "What would you need to see to call a pilot successful?" and write the answer verbatim. Then reflect it back with structure:

"So the pilot succeeds if [their words], measured as [metric], moving from [baseline] to [target] by [date]. Is that fair, and is that the number your leadership will judge it on?"

That last clause matters. The metric the champion cares about and the metric their boss judges it on are frequently different, and finding that out at day 85 is too late.

See the pilot plan for what happens next, and Never Commit before agreeing to any goal that anchors commercials to a specific uplift.