Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Programmatic Advertising › Advanced

Untangling Inherited Floodlight Activities for a Furniture Rental Firm

You inherit an invented furniture rental firm's Floodlight configuration, with duplicate enquiry activities, rental orders counted without revenue, silent activities and custom variables nobody reads, while live bidding in DV360 and SA360 depends on parts of it. The exercise practises building the register, finding every dependent before touching anything, deciding keep, fix, replace or retire for each activity and writing a staged, recorded change plan that protects the bid strategies.
ExerciseAdvanced levelModule 3: Running programmatic as a programme5 minUpdated

Why this matters

At the intermediate level you specified Floodlight for a company starting from nothing. Most marketing leads meet it the other way round: a CM360 advertiser built by people who have left, with live buying that depends on its quirks. The tutorial on cleaning up a Floodlight configuration sets the order: register, last tag dates, dependents, counting methods, the tags on the site, replace rather than rename, custom variables, staged retirement and a record. The Floodlight tags guide supplies the rules that make the order necessary. If activity design is not yet familiar, revisit the intermediate programmatic level first.

The rental firm and its inherited activities

Wrexquin Living is an invented firm renting furniture to people in flats and to small offices, on monthly contracts. Its new marketing lead inherits a CM360 advertiser linked to DV360 for display and, since last year's move of paid search into SA360, to Search Ads 360. Nobody left notes. The lead exports the configuration and looks for readers of each activity.

Illustrative example — not a real result

Wrexquin Living: the register after the first pass

ActivityGroup and typeCountingLast tagFires onDependents found
Enquiry form flatsEnquiries, counterStandardYesterdayFlats enquiry thank-you pageGoal of the DV360 automated strategy on the flats prospecting line item; audience list "Enquirers"
Lead form doneLeads old, counterStandardYesterdayBoth enquiry thank-you pagesGoal of the SA360 bid strategy for office furniture search
Rental orderEnquiries, counterStandardYesterdayOrder confirmation pageMonday "orders by channel" report; DV360 exclusion list "Customers"
Calculator usedEngagement, counterPer sessionYesterdayMonthly cost calculatorAudience list "Calculator users"
Activity AMisc, counterStandard11 months agoUnknownNone found
Student moveMisc, counterPer session7 months agoA student flats landing pageNone found
Brochure downloadEngagement, counterStandard14 months agoOld brochure pageA dashboard filter on its name

Custom variables: u1 "product line" (filled on every tag, used by no report); u2 "contract months" (filled on enquiries, used by no report); u3 "test" (filled with values nobody recognises).

Counts for the last 4 weeks against the firm's own records:

MeasureFloodlightFirm's records
Flats enquiriesEnquiry form flats: 520500
All enquiriesLead form done: 905860 (500 flats, 360 offices)
Rental ordersRental order: 226215, worth £41,300 in first-month rent

The late-summer moving season, the firm's busiest weeks, starts in 10 weeks.

Find out before you change

Each object above may look broken and still be holding something up: Chesterton's fence in a configuration. Two rules from the Floodlight guide set the limits. A group's type is fixed at creation, so the rental order cannot be moved into a sales group, only replaced. Tag strings sit in every deployed tag, so renaming a live activity to tidy it breaks every copy at once. And an activity that is the goal of a Bid Strategy is the most dangerous object in the configuration.

Task

  1. Read the register and say what each count comparison tells you.
  2. Decide keep, fix, replace or retire for every activity and custom variable, with the reason.
  3. For each replacement, write the new activity's group, counting method and order field.
  4. Order the moves of dependents so each bid strategy changes alone, after the new activity has been observed.
  5. Write the staged change plan, with retirements in batches and a record of every change, finishing before the moving season.

Worked answer

What the counts say. Both enquiry activities count the flats enquiry, so any report adding them double counts it. Both run slightly above the firm's records, which fits standard counting on a thank-you page that reloads. Rental orders run above the records and carry no revenue, because they sit in a counter group with no order field to deduplicate on.

Decisions.

  • Enquiry form flats and Lead form done: replace both with one "Enquiry submitted" activity in a new counter group named for the question, counted per session, with the product line variable telling flats from offices. Retire both once every dependent has moved.
  • Rental order: replace with a sales activity using transactions counting, the order reference in the order field so Deduplication stops a reloaded confirmation becoming a second order, and first-month rent as revenue, as finance decides.
  • Calculator used: keep.
  • Activity A: retire in the first batch once the tag check confirms nothing deploys it.
  • Student move: keep and ask the owner. Silence over the winter fits a seasonal page.
  • Brochure download: tell the dashboard owner, remove the filter, then retire in the second batch.
  • The product line variable: keep and document; it now does a job. The contract months variable: document and keep, since it is filled correctly. The test variable: trace through the tags before deciding, check in preview mode that it carries no personal data, and never reuse it.

Staged plan.

Illustrative example — not a real result

Wrexquin's change plan, each line entered in the change log

WeekChangeCheck before the next step
1Register finished; tags in the container matched to rows; u3 tracedNo personal data in any variable
2New "Enquiry submitted" and new sales "Rental order" deployed beside the old tagsNew counts within a few per cent of the firm's records over 2 weeks
4Audience lists "Enquirers" and "Customers" rebuilt on the new activities; Monday report movedLists collecting; report owner signs off
5SA360 office search strategy: new enquiry activity added as observed, then made the goalVolume steady; no other change that week
7DV360 flats line item goal moved to the new enquiry activitySame check; no other change that week
8Batch 1: tags removed for Activity A and both old enquiry activitiesNothing that read them changes
9Batch 2: old rental order and brochure tags removed; statuses changed after the Help Centre checkDashboards and reports unaffected

The sales activity's first fortnight: 4 weeks of orders at 215 suggests about 107 a fortnight, so a reading of 104 to 110 is the check. The last change lands in week 9, a week before the moving season starts in week 10.

Each change that alters a count or a goal is entered in the change log and marked as a break in trend reports, and the register is kept as the record the programmatic advertising audit will read. Each goal move follows the conversion goal change checklist, and each bid strategy is expected to relearn afterwards, which is why the moves sit in different weeks.

Common mistakes

  • Deleting the duplicate on day one. Each enquiry activity is a live bid strategy goal.
  • Renaming "Lead form done" to tidy it. Its tag strings are in every deployed tag.
  • Retyping the counter group. A group's type is fixed; build a sales activity beside it.
  • Moving both goals in the same week. Neither strategy's relearning can then be read.
  • Retiring Student move for its silence. It may be seasonal; ask first.
  • Reusing the test variable. Its history becomes meaningless.

Pages this lesson applies

Terms used on this page

Also related

Related