How many landing pages should we run, how fast should we add new ones, and when should we drop one?
Our answer
At $200K a month or more, run a matching landing page for each main angle or persona. Add about 1 to 5 new pages a week, and drop one only after a fair run adds no spend at your target.
- Confidence
- Low
- Strongest proof
- 2 Meta rules
- 1 dataset, 16 practitioners, 2 opinions or claims
- Sources
- 13 sources
- 21 statements
GNCPM
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentMore matching destinations let one winning ad spend more. Brad Ploch sent one top ad to 5 destinations in one campaign, and weekly spend about doubled with efficiency up (Scalability School).
Large accounts add pages every week. MarsMen launches about 2 to 5 new pages a week and Hollow 1 to 2, which sets the 1 to 5 range (Scalability School).
New pages look worse before they settle. Adding a new ad sends its ad set back into learning, when Meta says results don't necessarily predict later performance. Ploch's collection page looked bad for its first couple of days before it settled.
Our judgment
- The 1 to 5 new pages a week is our own range. It spans MarsMen's 2 to 5 and Hollow's 1 to 2. No source ties a page count to a spend level.
- Wait for a fair run before dropping a page. Don't judge it while its ad set is still in the learning phase, because adding the page's ad is a significant edit and Meta says results during learning don't necessarily predict later ones. This matches Ploch's collection page, Stuck's advice not to pause pages early and How to judge a creative test's rule of judging new ads over at least 7 days.
- SplitBase's test length applies to on-site A/B tests, not to pages Meta allocates spend to. Its 3 to 4 weeks and 100 conversions per variation cover on-site tests. Those are sized by Running A/B tests on the store, whose few hundred orders in each version sits above SplitBase's 100.
- MarsMen's "graveyard" campaign, run on a cost cap, works like the recovery campaign in New ads Meta won't spend on.
Our judgment, built from Scalability School (Zach Stuck and Brad Ploch), Meta, SplitBase and Intelligems: our reading of these sources, not a finding from any one of them.
What the sources say
21 statements from 13 sourcesMarin Istvanic says to test a new funnel with your 5 best existing ads so the destination is the only variable, and never run unproven creative through an unproven funnel. One client example, no results.
“Two unknowns in the same test = zero learnable signal.”
Pilothouse recommends opting out of Meta's product browsing before October 28 to keep the current ad experience through BFCM, warning that reported conversions may rise while AOV or margin fall. No test data.
“Duplicating an ad can turn the setting back on, so check every copy before it goes live.”
Relevant Audience, an agency, warns that Meta's product browsing can contaminate landing page tests on Meta traffic, and suggests comparing Ads Manager link clicks with GA4 Meta sessions to spot it.
“If you are A/B testing landing pages on Meta traffic, a change in what happens after the tap can contaminate the result.”
Meta says ad sets in the learning phase are less stable and usually cost more per result, so wait to edit until learning ends and avoid frequent budget changes, but do not try to avoid learning entirely.
“During the learning phase, ad sets are less stable and usually have a higher CPA.”
Meta says targeting, creative or optimization event changes and adding a new ad are always significant edits, while a budget change may or may not restart learning, depending on its size. No safe percentage is given.
“if you increase your budget from $100 to $101, that isn't likely to cause one or more ad sets to reenter the learning phase.”
Lorenzo Pravata's read of AG1's ads finds five customer personas, each sent to its own landing page so the ad's promise matches the page's first screen. It has no traffic or conversion data.
“Each persona gets its own landing page, so the promise in the ad matches the first screen of the page.”
SplitBase, a CRO agency, puts cold Meta conversion at 1.8% to 2.5% on product pages and 3.5% to 5.2% on dedicated landing pages (estimates it calls directional), and says to test for 3 to 4 weeks.
“the same traffic converts closer to double”
Brad Ploch says an apparel brand that sent Meta traffic only to product pages sent its top ad to listicles and collection pages too, in one CBO campaign; weekly spend roughly doubled and efficiency rose.
“the first couple days looked really bad. Like the collection page looked horrible.”
When traffic is too low for statistical significance, this podcast suggests choosing which landing page to back from engagement signals such as heat maps, as an editorial call.
“we're just getting more engagement on this new variation.”
MarsMen runs small page changes as normal on-site A/B tests, but tests a new angle, offer or page format by duplicating the Meta campaign to the new page and checking for added spend at target efficiency.
“incremental spend that would have not happened otherwise”
Zach Stuck says a new landing page takes about a week from idea to live; MarsMen ships about 2 to 5 a week and Hollow about 2, capped by the spend each page needs to prove itself.
Cody Plofker says his new sleep supplement brand starts on one channel, Shopify plus Meta, with about six reusable landing page templates, and has yet to find its beachhead persona, which is most of the early work.
“We don’t know our beachhead persona yet. Figuring out who we’re selling to is most of the early work.”
Intelligems studied 157 A/B test launches across 69 brands and 170M+ visits and found no systematic effect on paid traffic (paid social median +0%), including 29 redirect and theme tests. It sees no ad spend or bid data.
“The big-swing tests nudged paid no more than the quiet ones”
CTC builds several landing pages per creative concept so ads and pages match different personas, and keeps testing until conversion rate beats what its data says the order value needs.
CTC's hosts say frequent landing page testing is a key growth lever, and that their service puts no cap on how many landing pages get made.
Intelligems suggests a couple hundred orders per variant before trusting a result, about a month for a product doing 100 orders a week. With little traffic, make bigger changes and test one variant against a control.
“A loose guidepost is a couple hundred orders per variant before you trust a result”
On Scalability School, MarsMen runs 33 live landing pages across about 10 messaging pillars and adds about 3 new page funnels a week. A page wins if it adds net-new spend; misses are cut at about 48 hours.
“if we're not seeing the numbers that we want in like 48 hours, like we're usually killing that, that ad.”
Agency owner Nick Theriot ran the same ads to a listicle and a product page in one CBO. Over 45 days Meta spent $12,000 on the listicle at $62 per purchase vs $1,200 at $85. Conversion rate was about equal.
“the listicle actually got a much better cost per purchase of $62 versus the $85”
On Scalability School, Zach Stuck says to run many matching landing pages and let Meta split spend; one brand had about 29 pages spending in 14 days. The mix shifts over time, so don't pause pages early.
“it takes having these like different pockets to be able to get more scale.”
Grüns' paid media director says 74% of Huel's US ads point to one product page and 65% of its 684+ ads are static, and advises picking one hero product and sending most traffic to it.
“Pick one hero, make it exceptional, and then funnel the bulk of your traffic back to the same proven winner.”
Gruns' paid media director says to match the landing page to each new creative angle, even for the same product, showing two ads with different value props, each with its own page.
“Also, as you find new creative angles (even for the same product), match your landing page to the ads themselves.”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer