A Field Playbook

Put it together. Whenever someone drops a "can you just track X" on you for a site with no clean dataLayer, run the same loop. It turns a vague ask into a sequence of small, checkable steps — and it keeps you honest about the difference between "it fired in Preview" and "it is trustworthy."

The loop

• Reproduce & watch — do the action once in Preview; read every event it fires.
• Inventory — list the candidate events and what each payload carries.
• Pick — choose the event that fires reliably once and is richest.
• Target — confirm a discriminator so you fire on this case only.
• Reach the data — confirm every value you need is readable at that moment (push, sibling, DOM, URL, cookie).
• Decide — patch in GTM, or request a clean push for anything high-stakes.
• Build — variables, trigger, tag, with guards on anything scraped.
• Validate — trigger the real action; check the values and that it fires once, including on navigation.
• Document — note the event you chose, the discriminator, the sources, and any fragility.

The mindset, in one line

A clean dataLayer is the goal, but it is a luxury you often won't be handed. The craft is getting a trustworthy number out of a messy page anyway — by investigating first, choosing the most reliable signal available, reaching the data where it really lives, and being honest in your documentation about how solid the result is.