Current answers to the marketing questions consumer businesses ask. Sign in free to read them all.1,898 sources, at least 3 behind every answer.
Phone tag
Meta's new device default from Thursday's issue now has an answer behind it, and the answer is to stop fencing Sales ad sets to one phone. Also: a measurement vendor's Black Friday survival guide, a lead-gen account paying close to $6,000 per sale, Google's AI labels reaching California, and new answers on Google lead scores and opening a new location.
By Scott Stroud, Editor
Our answer on limiting Meta Sales ad sets to iOS or one device is new, and it says don't, by default. When your own breakdown shows a device or operating system doing worse or better, bid it down or up with a value rule. Keep a hard limit only for a business, legal or policy reason that needs zero delivery elsewhere, and then switch off "Reach more people across other mobile devices", which Marketing API v27.0 turns on for new Sales ad sets that target devices unless the tool says otherwise (Meta for Developers). Confidence is medium on the setup and low on performance, because nobody we read compares iOS-only delivery with open delivery, and Meta shows no performance data for the new default (PPC Land).
Social Media Today landed on the same advice I gave Thursday: if a third-party tool launches your campaigns, check that its settings match what you meant to run, since both the device expansion and Product showcase can be switched off (Social Media Today). Ask your tool vendor. Then ask again.
Survival of the fittest bid
Northbeam, which sells attribution, published its Black Friday survival guide, and Bryan Bumgardner's best line is about bidding: rising auction prices "will stall out cost cap and bid cap campaigns who simply can't operate at these prices", so he copies a stalled capped campaign into an identical lowest-cost one (Northbeam). Our answer on Black Friday timing already has you build an uncapped copy of your capped campaigns before the sale opens, so this is a reminder to do it now. He also wants "up to" out of your discount copy, because "it sounds like you're cheating people", though he shows no click-through numbers. And he warns that Meta's Product Browsing enhancement "has a tendency to drift your prospecting campaigns toward retargeting", so confirm it's on or off on purpose for every ad, the feature-by-feature check our answer on Meta's creative enhancements asks for anyway. His 20% to 35% CPM spike for last year's big weekend comes with no source, so I'm filing it under weather report.
Barry Hott, in Foxwell's recap of his New York talk, says he ran "multi-million-dollar campaigns" on evergreen ads, changed only the on-site offer, and watched performance rise with no sale creative at all (Foxwell). He gives no figures, and our answer on Black Friday creative still wants a separate, bigger batch of sale ads on CTC's data. Hott's other line deserves a sticky note on your monitor: a low CPA "often just means an ad set is picking up easy wins from warm audiences."
Shop talk
- Bumgardner again, on Xhe says he analyzed $6 billion of ad spend across 1,000 Northbeam customers and found Meta conversion rates at record lows this year while "ROAS is better than ever" (Bumgardner). No figures for either, and his fix runs through a Northbeam product.
- Bumgardner again, on XOur answer on landing page conversion rates plans on about 1.5% of Meta visitors buying and puts page speed first, so his "landing page optimization is suddenly existential" lands closer to home than his sales pitch.
- Jordan West, who runs a TikTok Shop agency, relays Prescient AI, which sells mix modeling: across $126 million of TikTok spend, "63.2% of modeled TikTok-driven revenue happened OUTSIDE TikTok for the median brand" (West). No brand count or method, and it's West's second relay in two issues pointing the same way after Thursday's Fenty study.
- Our answer on measuring TikTok judges TikTok on total sales across your site, Amazon and TikTok Shop, never on TikTok's own reporting.
- Eric Seufert, writing about AppLovin, calls CPM industry checks "actively misleading", because "CPM is a synthetic metric" and ROAS is what you care about (Seufert).
- Our answer on whether your CPM rise is normal does compare you with Meta's own price trend and with peers, but it sends you to your own account first, which is where Seufert would start too.
- Menachem Ani, an agency owner, shared a lead-gen account getting leads at $117 each: "204 leads and 4 sales" in 90 days, "close to $6,000 per sale", with deals taking about 100 days to close (Ani).
- Our answer on optimizing for lead quality says judge every campaign on cost per qualified lead and cost per customer from your CRM, never cost per lead. Ani's account is the horror-movie version of why.
- Meta's out-of-cycle changes page, in an entry dated November 5, adds a creator breakdown to the Ads Insights API that splits partnership ad results by creator handle, with impressions, reach, link clicks and video completion only, no spend or conversions, and data from September 24 (Meta for Developers). Handy for seeing which creator gets the clicks. Which one sells is still your problem.
Our answer on Google's lead intent scores is new, for anyone who sells through leads. The scores, still a pilot, label each lead high, medium or low intent so sales can prioritize outreach (Google). Use them to decide who gets called first, keep working the low-intent leads, and check each band against your own qualified and closed leads before routing on it. They don't feed bidding, so bid on a qualified lead or another CRM stage that predicts a sale and keep the form fill as a secondary conversion. Medium confidence on what the feature does, low on what it's worth. See Ani's account above for what a form fill alone can hide.
Search Engine Roundtable reports that Google now puts visible AI labels on ads whose assets are marked AI-created or AI-edited when they target California, South Korea or Vietnam, joining the EU, India and New York (Search Engine Roundtable). Google's help page lists all six, warns that the label setting "doesn't guarantee your compliance with specific regulations", and suggests keeping your own labels out of image corners and edges and opting out of image enhancements, which can crop them (Google Ads Help). Our answer on labeling AI-made ads covers Meta, where your own duty is a clear disclosure on any realistic AI-made person, and Google now draws its own label on top.
Ani's Black Friday housekeeping, 22 days out: don't launch a new Performance Max campaign; add promo assets or a new asset group to a campaign with history (Ani). That fits our answer on Google campaign structure, which runs as few campaigns as your goals allow. He also says submit Merchant Center promotions now, since they "go through manual approval" (Ani). Our answer on free listings and promotions gives Google's floor of 24 hours before a promotion starts, and I'd take Ani's clock over Google's.
New in the book
Two more new answers. WhatsApp Status ads, and advertising a new location before it opens: start 2 to 3 months before it can take bookings, aimed at awareness and a waitlist, then switch on conversion campaigns and "now open" messaging when booking opens and keep it about 3 months (Cardinal Digital Marketing). Cardinal sells paid media to healthcare groups, and the other sources are fitness businesses, so confidence is medium on when to start and low on how long "now open" keeps working.
One changed call, and it's a callback to Thursday's new-customer answer. Our answer on excluding existing customers now says test Meta's "Acquire new customers" setting on a duplicate prospecting ad set first, with its window cut from the 730-day default to 180 days or shorter, and keep that window if the test wins.
Meta's on-device counting starts November 12, in 6 days. Black Friday is 21 days out.
Every fact links to its source. Opinions are the writer’s, built from the sources linked beside them.