Half the year is gone. The patterns are clear.
Five months of 2026 across the firm’s account book tells a consistent story. The accounts that grew did the same handful of things. The accounts that flattened or shrank did the inverse, mostly without realizing it.
This is a mid-year read on what produced results between January and May, what produced the appearance of results without the bank account agreeing, and what the second half of the year looks like for the operators planning Q3 now.
What worked
Creative volume held the line. Accounts producing 8 to 20 new Meta variants per month maintained their CPA targets through the May attribution overhaul and the Reels safe-zone consolidation. Accounts producing fewer than that drifted upward 10 to 25% on CPA over the same period and blamed the platforms. The accounts that adjusted production capacity (in-house producers, AI-assisted variant generation, freelance partnerships) absorbed the platform changes without losing performance.
Conversion event hygiene paid off. Accounts that fixed their conversion event setup early in the year (cleaning up duplicate events, removing optimization against shallow conversions, adding server-side tracking) saw stable performance even as in-platform attribution numbers shifted. Accounts that optimized against page-view or button-click events saw their dashboards become increasingly disconnected from real revenue.
Owned-channel concentration beat diversification. Accounts that consolidated to two or three channels with depth produced 20 to 30% better blended CPA than accounts that spread across five or six channels with thin budgets. The pattern held across ecommerce, B2B, and services. The firm’s position on more channels survived another quarter unchanged.
Landing page work outperformed bid strategy work. Across audits this year, accounts that rebuilt a landing page or simplified a form saw 2-3x performance improvements 1. Accounts that spent the same hours on bid strategy adjustments saw single-digit swings. The leverage continues to sit upstream of the ad account.
What did not work
Channel additions without saturation testing. Most accounts that added TikTok, LinkedIn, or Pinterest in Q1 2026 found by Q2 that the new channel was producing thin returns and pulling attention from channels that were working. Several killed the new channel and reallocated. The pattern the firm has been writing about for two years: most accounts add channels before saturating their existing ones.
Always-on retargeting and branded search defense. Accounts that ran pause tests on these found 50 to 70% of the spend was non-incremental. Redirecting to prospecting produced measurable lift in new customer acquisition. The two categories of waste remain the most common.
Audience layering on Meta. Accounts running 15-plus lookalike or interest-layered audiences saw better results consolidating into Advantage+ broad with strong creative. Meta’s algorithm in 2026 has internalized most of what audience layering used to provide; the maintenance hours are better spent on creative.
Over-optimization against synthetic conversion events. Accounts that adopted QFC, engagement-based conversions, or other AI-derived signals as primary optimization targets often saw the signal volume rise while real revenue stayed flat. The signals are useful as bidding inputs on thin-volume accounts; they are misleading as reporting KPIs anywhere.
What is different about H2 2026
Three structural shifts that change the math for the next six months.
Universal Commerce Protocol expansion means ecommerce checkout is happening on Google’s surfaces, not just merchant sites. The brand experience that used to live on a landing page now has to live in the post-purchase email and the first-package unboxing. Brands that have not updated those two touchpoints will see reduced repeat-purchase rates on UCP-routed orders.
AI Mode placements (Conversational Discovery, Highlighted Answers) reward ad copy that reads as helpful answers. The accounts that rewrite their RSAs and product descriptions in the next quarter will earn 6-12 months of placement advantage before the rest of the market adjusts.
Meta’s March-to-May attribution overhaul is now stable enough that dashboards reflect real performance more honestly than they did in 2024-2025. Operators who anchor reporting on downstream revenue rather than in-platform attribution will find that the reported numbers and the bank account agree more often this half than last.
The Q3 planning shape
Three concrete moves before July ends.
First, audit the creative production rate against the per-channel-spend floor. If you are spending $20K monthly on Meta, you need 8 to 12 new variants per month. If you are below that production rate, fix the pipeline this quarter, not in Q4 when the holiday creative load triples.
Second, run the pause test on your single most expensive non-incremental candidate (retargeting, branded search defense, or whichever LinkedIn campaign has been underperforming for six months). Two weeks of measurement reveals whether the spend is earning out. Most accounts find at least one campaign to redirect.
Third, get the Q4 plan on paper now 2. Holiday creative production timelines now back up against the August-September production window; June is the planning month, not the production month. Holiday teardowns from the firm will continue through the summer, but the structural decisions (which channels, which offers, which audience strategies) want to be decided now.
The half-year frame that holds
The leverage continues to sit in the same four places it has sat for three years: offer, creative, landing page, conversion event quality. Every platform release this year has reinforced that pattern rather than changing it. The work that moved the number in January is the same work that will move it in December.
The accounts that win the second half will be the ones whose operators stop chasing the release notes and stay concentrated on the work the algorithms cannot generate. Spend less, on fewer things. Pour the hours into the inputs. Read the dashboards honestly. The half-year reset is mostly remembering what was always true.