Compare each month with the same month last year, and split ad visitors from everyone else. On the first Saturday of each month, when you fill in the WeScale scorecard, write two numbers next to last year's: the share of visitors who bought among people who came from Facebook/Instagram ads, and the same share for everyone else. Worked example, August: ad visitors 2.00% (Aug 2025) → 0.74% (Aug 2026); everyone else 2.84% → 3.19%. The data is in the store database (G:/iris/igl-ops-sage/data/igl.db, table fact_traffic_channel_daily: purchases ÷ sessions, filtered by channel_group = 'Paid Social' and by everything else).
Before you make any of the "simplify and remove" store changes from mastermind call #43, find out what changed in your ads between summer 2025 and January 2026. Ad visitors have bought less often than in the same month a year earlier for eight months running, January through August 2026. Search visitors buy more often than last year. So the store page is not the first suspect; who the ads reach and where they send people are. Three checks, all in Meta Ads Manager:
1. Landing page: for the 10 highest-spend ads of July 2025 and of July 2026, write down where each one sends the visitor (a product page, a collection page, or the home page). 2. Ad type: for the same months, what share of spend went to catalog ads (the automatic product-feed ads) versus single-image or video ads you built yourself. 3. Audience: did any campaign switch from a set audience to broad Advantage+ targeting between December 2025 and January 2026?
Whichever of the three changed around the turn of the year is the lead to test first.
Claude Shannon ignored daily price charts as noise and plotted earnings over years. Read IGL's own store data that way and the problem shows up as a year-over-year collapse in ad-visitor buying, which a one-week "below the cohort median" comparison never showed.
Claude Shannon beat almost every fund manager using a two-part habit. He threw out price charts as "a very noisy reproduction of the important data", then plotted each company's earnings over several years and asked what would make the trend continue or stall. The key was comparing years, not days. IGL's conversion problem has only ever been read the price-chart way: one self-reported week against the mastermind's other members (week 34 of 2026: 1.5% vs a 2.6% median).
Read over the full record, the gap against peers is real but modest: across 33 reported weeks IGL sat below the weekly cohort median in 26, with a median of 2.04% against 2.49% (58th of 69 members). The larger finding is IGL against itself. In the live store database (synced to 2026-10-04), Shopify orders per visit fell from 2.55% in July–August 2025 to 1.56% in July–August 2026, down 39%, while visits rose 28%. Sales per visit fell from about $1.08 to $0.79 even though the average order rose from about $42 to $50.
Operator addition, verified tonight: split by where visitors came from, the drop is mostly in ad traffic. July–August, Facebook/Instagram ad visitors bought at 2.09% in 2025 and 0.98% in 2026 (21,170 vs 25,351 visits). Search visitors went up, 5.5% to 7.9%. If ad visitors had held last year's rate, July–August 2026 would have had about 280 more tagged purchases; all other channels together account for about 80 more. Month by month, ad-visitor buying is lower than the same month a year earlier in every month from January through August 2026 (Jan 1.70%→1.09%, Feb 1.75%→1.04% … Aug 2.00%→0.74%), so the slide started by January. September 2026 is unreliable: "Direct" traffic spiked to 3,128 visits at 0.58%, which looks like a tracking or bot problem.
No single knowledge base gets here. founders-kb has the method (read the multi-year trend, not the tick). wescale-kb has the room's habit of reading curves and its standard cure for conversion, "simplify and remove" on the store. igl-brand-kb has Wren's February 2026 store review, which asked for 60–90 day evaluation windows. Only IGL's own monthly data, split by channel, shows that the store is losing to its own past, and that the loss is in ad traffic. That moves the first question from the product page to the ads.
founders-kb/founders/claude-shannon — seed: Shannon's investing record (about 28% a year for ~30 years), treating markets as a puzzle.founders-kb/transcripts/092-ed-thorp-claude-shannon-transcript — the practice, verbatim: price charts are "a very noisy reproduction of the important data"; "how the earnings have changed in the past few years"; earnings plotted on log paper with a trend line. (Transcript cited as the only verbatim source of the quote.)founders-kb/episodes/428-claude-shannon-worked — episode 428 (2026-08-09): the investing record and the habit of reducing a problem to its core.wescale-kb/calls/call-43 — call #43, 2026-08-20: the room reads members' curves; 2.5% conversion benchmark; "simplify and remove rather than add".wescale-kb/scorecard/wescale-scorecard.db — all 33 IGL conversion weeks (2025-W43 to 2026-W34), self-reported, pre-W30 values unit-normalized per nugget #032: 26 below the weekly median, IGL median 2.04% vs 2.49%, rank 58 of 69.igl-brand-kb/methodology/website-review-feb-2026 — Status: REFERENCE, Wren review dated 2026-02-28: a proposal, not a live commitment. Priority 2 (end the permanent crossed-out sale price) and priority 6 (swap bundles for a $75–85 free-shipping threshold), each to be judged over 60–90 days.G:/iris/igl-ops-sage/data/igl.db (outside the knowledge bases; live, synced 2026-10-04) — fact_traffic_channel_daily (sessions, analytics-tagged purchases, by channel) and fact_shopify_daily_financials (Shopify orders, gross sales). Every year-over-year figure above was recomputed by the operator tonight.