pilotads

Document PA-PF-74 · published in full · updated 14 Aug 2026

The 74-point pre-flight checklist

This is the actual document. Every Google Ads launch and takeover at Pilotads runs through these 74 checks, in this order, before a euro moves, and the completed list goes in the client folder with a date on it. We publish it because the value was never the list; it's running the list every single time, including the fifth takeover of the year when everyone is confident and bored.

Use it yourself, freely. Items citing platform behavior (Consent Mode v2, the 2× overspend rule, PMax channel reporting) reflect Google's documentation as of August 2026; the rest is routine we'd defend line by line. Print it, it fits on four pages. By Jorren Vandermeer, who has signed off this list more than 140 times.

Flat navy illustration of a pre-flight checklist clipboard with amber checkmarks

01–14 Account & campaign structure

The skeleton. Most expensive messes trace back to decisions skipped here.

  1. 01A written business goal per campaign, one sentence, in the campaign's naming doc
  2. 02Naming convention agreed and applied, so change history stays readable
  3. 03Brand and non-brand split into separate campaigns with separate budgets
  4. 04Location targeting set to 'presence', not the 'presence or interest' default
  5. 05Language targeting matches the landing pages, not the audience's passport
  6. 06Search-partner network off at launch; earn its place with data later
  7. 07Display expansion off on every search campaign
  8. 08Match types chosen deliberately per ad group, not inherited from an import
  9. 09Ad groups tight enough that one ad can honestly serve every keyword in them
  10. 10No keyword duplicated across campaigns without a documented reason
  11. 11Master negative-keyword lists attached to every campaign
  12. 12Vertical-specific negative themes seeded (jobs, free, DIY, competitors)
  13. 13Audience segments attached in observation mode for data from day one
  14. 14Ad schedule reviewed against business hours where phone answering matters

15–28 Tracking, consent & measurement

If measurement is broken, everything downstream optimizes toward noise. This block blocks launch.

  1. 15Conversion actions defined with primary and secondary status set on purpose
  2. 16One primary conversion per goal; no double counting between form and thank-you page
  3. 17The Google tag or GTM container present on every page, verified, not assumed
  4. 18Conversion firing verified in Tag Assistant on a real test submission
  5. 19Consent Mode v2 signals (ad_user_data, ad_personalization) present for EEA traffic, required since March 2024
  6. 20The consent banner actually gates tags: verified in both accept and reject states
  7. 21Google's hashed first-party conversion upload enabled and receiving data
  8. 22GA4 linked; imported audiences and conversions chosen deliberately
  9. 23Call tracking decided: on, off, or forwarded numbers, with the why written down
  10. 24Conversion values set where leads or products differ in worth
  11. 25Attribution model reviewed rather than left on whatever the account shipped with
  12. 26One end-to-end test conversion recorded and seen inside Google Ads
  13. 27Thank-you pages excluded from ad targets and from indexing
  14. 28Auto-tagging on; manual UTMs consistent where they remain

29–40 Budget guardrails

The block that prevents the expensive weekend.

  1. 29Daily budgets × 30.4 reconcile to the monthly plan a CFO signed
  2. 30Google's 2× daily overspend behavior accounted for in the pacing plan
  3. 31Shared budgets used deliberately or not at all, never as a leftover default
  4. 32Bid strategy matched to conversion volume: no target CPA on five conversions a month
  5. 33Targets set from historical data, and written down where everyone can see them
  6. 34Portfolio bid strategies documented, since they hide in a separate menu
  7. 35A pacing check scheduled for every Monday, on the calendar, owned by a person
  8. 36Account-level spend expectations agreed in writing with billing thresholds understood
  9. 37Seasonality adjustments planned for known peaks, not improvised during them
  10. 38Currency and time zone verified before launch; both are permanent
  11. 39Billing contact plus a backup payment method on file
  12. 40A first-week spend review booked before launch, not after the surprise

41–54 Ads & landing pages

Where the account meets an actual human.

  1. 41At least two responsive search ads per ad group, saying different things
  2. 42Every headline reads sensibly in any combination Google might assemble
  3. 43Pinning used sparingly and reviewed, accepting the reach cost knowingly
  4. 44Display paths used; they're free relevance
  5. 45Sitelinks (four minimum), callouts, and structured snippets attached
  6. 46Image assets uploaded where the vertical allows them
  7. 47Business name and logo verified in the account
  8. 48The landing page keeps the ad's promise above the fold
  9. 49Mobile load time checked on a real phone on cellular, not office wifi
  10. 50Forms tested on mobile with a real submission
  11. 51Privacy policy reachable from every landing page
  12. 52Copy scanned against ad policies for the vertical's trigger words
  13. 53Trademark usage cleared where competitors are named
  14. 54A 404 check across every final URL

55–64 Shopping & Performance Max

Product accounts only. The feed is a second account that can break independently.

  1. 55Merchant Center linked, verified, and free of disapprovals at launch
  2. 56Feed refresh scheduled daily, with someone owning feed failures
  3. 57Product titles front-load what shoppers actually type
  4. 58GTINs present wherever the product has one
  5. 59Price and availability sync verified against the live site
  6. 60Brand exclusions applied to Performance Max before it spends
  7. 61PMax channel-level reporting reviewed monthly, available to all accounts since November 2025
  8. 62Final URL expansion decided consciously, on or off
  9. 63Asset groups mirror product lines, not the org chart
  10. 64A written plan for search-versus-PMax overlap, reviewed in month one

65–74 Launch day & first week

The window when small problems are still small.

  1. 65Launch mid-week at a quiet hour; never Friday afternoon
  2. 66Campaigns enabled from paused state one by one, watching spend start
  3. 67Search terms scanned within the first 24 hours
  4. 68Spend pacing checked against plan within the first 24 hours
  5. 69Disapproval sweep after launch, then again on day three
  6. 70Every final URL returning 200 through any redirects
  7. 71First negative-keyword sweep completed inside week one
  8. 72Auto-apply recommendations switched off, and checked that it stayed off
  9. 73Change history reviewed at week's end for anything nobody remembers making
  10. 74A baseline snapshot saved: CPA or CPL, CTR, impression share, so week six has something honest to compare against

What the checklist can't do

It can't set strategy, write ads people believe, or negotiate a CPA target that respects your margins; those stay judgment calls. And it can't run itself in week 30, which is where most accounts quietly stop doing the boring parts. The management routine is, honestly, this list plus a calendar plus stubbornness, and the self-audit walkthrough covers the diagnostic side for accounts already flying.

Asked about the list

Two honest reasons. A checklist you hide is a checklist you can quietly skip, and publishing it keeps us accountable to our own routine. And most companies who read it conclude the discipline is the hard part, not the list, which is exactly the part an agency sells. If you run it yourself and never need us, the list did its job.

No. The Shopping and Performance Max block only applies to accounts selling products, and call tracking items drop out where nobody phones. On a typical B2B lead-gen takeover about 60 of the 74 apply. What never drops out: tracking verification, Consent Mode v2 for EEA traffic, budget guardrails, and the launch-week reviews.

On a clean new build, one to two working days. On a takeover it runs longer, usually three to five days, because half the checks turn up existing problems that need fixing before launch rather than noting. The €1,900 audit exists for exactly that situation: it prices the findings before you commit to the rebuild.

Rather have it run for you?

Every engagement starts with this document and ends with a monthly waste report. Reply within five business days.

Get a proposal