Smart MarketerSMART MARKETER LIVE

SMART MARKETER LIVE 2026 · DAY 1 TAKEAWAYS

Build Your Own Media Buying Assistant With an AI Tool of Your Choice

Marianne White

Updated 2026-10-27 · 1 processed session

Turn your daily account review into an assistant that prepares useful checks and recommendations. Document your reasoning, verify its output and keep people in control of account changes.

01 / Marianne WhiteBuild Your Own Media Buying Assistant With an AI Tool of Your Choice

2026-10-20 · AI assistants · Media buying · Budget monitoring · Human review · Workflow documentation

Turn your daily account review into an assistant that prepares useful checks and recommendations. Document your reasoning, verify its output and keep people in control of account changes.

PUT THIS SESSION TO WORK

Welcome. This session with Marianne White can help you turn your daily account review into an assistant that prepares useful checks and recommendations. Document your reasoning, verify its output and keep people in control of account changes.

Turn your daily account review into an assistant that prepares useful checks and recommendations. Document your reasoning, verify its output and keep people in control of account changes.

01

Verify account access and the first budget report.

Confirm the intended account, distinguish campaign and ad-set budgets, and manually check the first output. Configured budget and actual spend are different.

02

Teach your review process, including its reasoning.

Walk through real decisions, define the desired report and test a simple version before adding detail.

03

Keep recommendations advisory and protect working campaigns.

The lesson’s assistant reports rather than edits. Include conditions that prevent isolated ad results from triggering unnecessary changes to a successful group.

04

Make every recommendation easy to verify.

Include a direct link, check recent results and refresh information that may have changed before taking action.

05

Carry the team’s decisions into the report.

Check recommendations against agreed plans, refine instructions as you learn and run the report manually if scheduling is unavailable.

Which of these would be most useful for your business?

Open a choice to see the first question and what you’ll create. Continue the work in your connected Coach.

Specify my budget checkA budget-check specification with account scope, totals, alerts and manual validation.

Define a read-only report that catches unexpected budget settings.

Useful inputs

  • Approved offer and persona
  • Authorized account scope
  • Budget structure
  • Review cadence and notification preference

Work through it

  1. Reuse the approved offer and persona; confirm which product, audience and decision this activity concerns.
  2. Separate configured budgets from actual spend and document campaign/ad-set treatment.
  3. Ask the user to choose an alert condition and timezone/cadence; do not reuse the speaker’s threshold.
  4. Specify read-only behavior and verify the available connector permissions.
  5. Plan a manual first-run comparison; do not claim a schedule is active until the host creates and verifies it.

If inputs are missing: Keep missing evidence explicit. Ask for the smallest useful input, or prepare a labeled draft for the user to complete; do not invent account results, customer facts or platform behavior.

Your working document

  • Account scope
  • Budget calculation
  • Alert rule
  • Cadence/timezone
  • Read-only permissions
  • First-run validation

Complete when: The user has reviewed the proposed output, confirmed its assumptions and chosen a next step.

Map my daily media-buying reviewA report blueprint with decision rules and protective checks.

Capture the decisions and exceptions an assistant should understand.

Useful inputs

  • Approved offer and persona
  • Current review process
  • Business KPIs
  • Lookback windows
  • Examples of decisions

Work through it

  1. Reuse the approved offer and persona; confirm which product, audience and decision this activity concerns.
  2. Capture each step and the reason for it using the user’s own thresholds.
  3. Add protective rules for successful campaigns/ad sets and recent changes.
  4. Draft a read-only report with evidence links and a clearly separate human decision.
  5. Compare a dry run with the user’s manual review before refining the next version.

If inputs are missing: Keep missing evidence explicit. Ask for the smallest useful input, or prepare a labeled draft for the user to complete; do not invent account results, customer facts or platform behavior.

Your working document

  • Review sequence
  • Decision criteria
  • Exceptions
  • Evidence links
  • Dry-run discrepancies

Complete when: The user has reviewed the proposed output, confirmed its assumptions and chosen a next step.

Review my assistant’s recommendationsA verification checklist and revised reporting instructions.

Find stale, unsupported or conflicting recommendations before acting.

Useful inputs

  • Approved offer and persona
  • Redacted report
  • Current decision rules
  • Relevant agreed plans

Work through it

  1. Reuse the approved offer and persona; confirm which product, audience and decision this activity concerns.
  2. Check each recommendation against current evidence and its timestamps.
  3. Identify conflicts with existing decisions, missing context or ignored protective rules.
  4. Write targeted instruction changes and a new validation check.
  5. Offer an on-demand workflow if scheduling or access is unavailable.

If inputs are missing: Keep missing evidence explicit. Ask for the smallest useful input, or prepare a labeled draft for the user to complete; do not invent account results, customer facts or platform behavior.

Your working document

  • Recommendation checks
  • Stale/missing evidence
  • Conflicts
  • Revised instructions
  • Next validation

Complete when: The user has reviewed the proposed output, confirmed its assumptions and chosen a next step.

The activities below are Coach-designed applications of the reviewed teaching. Before personalizing them, reuse approved business context or gather the business name and website, prepare Offer and Persona Profiles, obtain approval of both, and then resume the chosen activity. Session summaries do not require intake. Coaching paths and output templates are Coach-designed applications. This guide does not submit or store answers.

Read all takeaways ↓ · You can also ask the Coach a question about the session.

Questions about the session
Explore the full session index · 15 entries

Index of reviewed teaching from the supplied recordings. Repeated rotations are consolidated; attendee exchanges and private details are excluded.

Two assistant jobs

The speaker describes two practical assistant jobs: a budget check and a daily media-buying review. Both depend on successfully connecting the relevant account data.

Talk · 04:37

Verify intended account access

Before using a connected assistant, ask it to confirm access to the intended account. A budget check should show the configured budgets so unexpected changes can be noticed.

Talk · 05:33

Budget levels and alert conditions

For a budget report, distinguish campaign-level budgets from ad-set-level budgets, total the applicable amounts and flag changes above a threshold chosen for the account.

Talk · 07:27

Manually validate the first report

Check the first report manually against the account before relying on the recurring output.

Talk · 09:16

Read-only intent versus connector permissions

The described assistant checks and reports rather than changing budgets. An instruction not to change the account expresses the intended behavior; it is not the same as removing a connector’s ability to make changes.

Talk · 09:54

Configured budget versus actual spend

A configured budget report is not a statement of actual spend. Changes made during the day can affect the relationship between the configured amount and what is spent.

Talk · 11:02

Explain the human review process

Teach the assistant by walking through the decisions you make when reviewing an account, rather than expecting one generic prompt to capture your method.

Talk · 12:57

Define the requested decision report

Define the report you want, including which campaigns or ads need attention and the reasoning for possible increases, decreases or pauses. The decision rules should fit the account.

Talk · 14:07

Start with a simple dry run

Explain decisions campaign by campaign, then run a simple dry run and compare what the assistant sees with your own reading. Add more detailed review steps gradually.

Talk · 17:23

Protect successful groups

Include protective rules before recommending that an ad be stopped. The speaker checks whether its containing ad set or campaign is meeting the objective rather than judging the ad in isolation.

Talk · 18:42

Recent performance and direct links

Check recent performance before recommending a stop, and include direct account links in the report so a person can inspect the relevant ad.

Talk · 21:35

Refresh and spot-check results

Spot-check the assistant’s reasoning and refresh stale results before acting. Conditions may have changed between the report being produced and the person reviewing it.

Talk · 22:36

On-demand reporting fallback

When scheduling is unavailable, the same documented report can be run on demand. The review method does not have to depend on an automatic schedule.

Talk · 25:14

Check recommendations against decisions

Keep the review instructions with the relevant business context and agreed decisions. Compare recommendations with recent decisions so the assistant can flag conflicts rather than overriding them silently.

Talk · 25:56

Refine instructions in natural language

Refine the instructions by describing what should change in natural language. The context and report can be improved through repeated review.

Talk · 26:56

Key takeaways

  1. Verify account access and the first budget report.

    Confirm the intended account, distinguish campaign and ad-set budgets, and manually check the first output. Configured budget and actual spend are different.

    Talk · 05:33; Talk · 07:27; Talk · 09:16; Talk · 11:02

  2. Teach your review process, including its reasoning.

    Walk through real decisions, define the desired report and test a simple version before adding detail.

    Talk · 12:57; Talk · 14:07; Talk · 17:23

  3. Keep recommendations advisory and protect working campaigns.

    The lesson’s assistant reports rather than edits. Include conditions that prevent isolated ad results from triggering unnecessary changes to a successful group.

    Talk · 09:54; Talk · 18:42

  4. Make every recommendation easy to verify.

    Include a direct link, check recent results and refresh information that may have changed before taking action.

    Talk · 21:35; Talk · 22:36

  5. Carry the team’s decisions into the report.

    Check recommendations against agreed plans, refine instructions as you learn and run the report manually if scheduling is unavailable.

    Talk · 25:14; Talk · 25:56; Talk · 26:56

Context to keep in mind

  • Product plans, connectors, prices and scheduling availability mentioned in the talk are not verified current instructions; check the chosen host before implementation.
  • This session specification does not connect accounts or create a monitor by itself.
  • The assistant’s output needs human verification; no automatic account-editing workflow is supplied.