SocialMapped Panel Review
This page is private. Enter the password to view.
That's not it. Try again.
Round 2, overnight: two more reports, and what shipped
You asked for reports on evolutiondermatology.com and 4everyoungantiaging.com, then asked to panel them for more engine fixes. Both are done, deployed and tested, 361 tests passing.
Read the new reports
- Evolution Dermatology, Boulder medical and cosmetic dermatology
- 4Ever Young Anti-Aging Solutions
Both sets of quiz answers were filled in by me from their public websites, not by the owners. The real research and my assumptions (capacity, ad budget, past failure) are marked separately in the code. Worth knowing before you send either one anywhere.
Shipped from the first panel
- The bad facts are gone: Meta Business Suite as a Pinterest scheduler, Idea Pins, repin quotas, and the "engaging warms the algorithm" ritual. The 30-day plan now publishes in week 1 instead of stockpiling to week 3.
- Cadence has a real ceiling and the report prints what the plan costs in hours. Pinterest is 3, 6 or 10 pins a week by capacity, never a daily quota.
- An owned platform can no longer rank below a cold start. This took two attempts: points and a tie-break were not enough, and Evolution's real report still put a net-new TikTok above the Facebook page they actually run. It is a hard ordering rule now, and Facebook took the top-3 slot on regeneration.
- LinkedIn is no longer erased for a visual creator whose buyers are businesses, and TikTok is demoted for lead-gen service businesses (your call, over your own July 24 ranking; it still leads for product businesses).
- A Google Business Profile section, shown only when the customer's own website proves they have a location, so an online coach never gets it.
- The AI avatar pitch is gated: suppressed for anyone who said no, and for anyone comfortable on camera.
- The fabrication gate catches invented client results like "a financial coach who doubled her inquiries", which carried no number and slipped past every existing check.
What the second panel found, verified against the live reports
A defect showing up for two unrelated businesses is a pipeline bug, so the harness now aggregates across reports. It found real damage, and I checked every claim before acting on it:
- Nine bracketed placeholders shipped in one paid report: "[Clinic Name]", "[city]", "[state]", "[Patient name]", "[Link to lead magnet]". All were already banned in the prompt.
- The fabrication gate is leaving wreckage. "skin cancer is a very common form of cancer?S., but early detection can save your life" is what remains after it deleted an unsourced statistic mid-sentence. The false claim went; the broken grammar shipped.
- Evolution Dermatology was told to name its Pinterest account after "Peak Dermatology". I checked whether that was another customer's data leaking. It was not: the model invented the name to fill an empty slot. Still the worst thing in either report.
- Plus an empty interpolation ("At, we help adults 35+"), a prompt instruction echoed to the reader, a 2019 Pinterest daily-cap myth, "tag a friend" engagement bait, and Instagram's sends signal printed on YouTube and Pinterest pages as though one algorithm ran everything.
Fixed: a new output lint catches the mechanical defects (both reports score 11 blocking each, which is how I know it works), and the four factual leaks are banned globally. The lint logs rather than blocks, deliberately: a scuffed report still beats no report, and you should see the real defect rate before deciding to fail a generation over it.
Waiting on you
- Regulated-medical content. Both panels flagged that these reports tell medical practices to publish before-and-afters and patient stories with no consent language. That is real exposure for a clinic and it needs your judgment on how far the product should go, so I left it alone.
- 4Ever Young is a franchisor (4Ever Franchisor LLC, many locations, separate per-location accounts), not a single business. The report treats it as one clinic with a brand name on top. Worth deciding whether multi-location is a product you want.
- Their website blocks automated reading with a CAPTCHA, which I did not attempt. So that report had no website research behind it, and the pipeline did not admit that. It should.
- The Teal Moon duplicate is still there for you to delete with the new button.
The full cross-report analysis is in the repo at research-notes/panel-runs/ENGINE-CHANGES.md, and adding a new business to the panel is one entry in research-notes/panel.py.
Round 1: the verdict on Living Potential
Accuracy
5.5/10
Platform mechanics are broadly right, undercut by one tool that cannot do what the report says, two dead Pinterest formats, and one invented algorithm rule.
Fit
4/10
The weakest axis. It benches a Facebook page she already runs, skips LinkedIn while half her revenue is B2B, and builds one funnel for a two-audience business.
Actionability
6.5/10
Scripts, hooks, bios and pin copy are genuinely paste-ready. The workload, roughly 41 publishing actions a week, is not survivable for someone whose stated problem is no time.
Would she feel the $49 was worth it? Marginally yes on the written assets alone, but the panel expects her to feel misled within three weeks, when the cadence collapses, the scheduling instruction fails, and the auto-DM she was promised does nothing.
Accuracy problems
1. Meta Business Suite cannot schedule Pinterest 4 of 8
Meta Business Suite handles Facebook and Instagram only. It has never scheduled Pinterest. A time-starved owner loses an evening finding this out.
2. Idea Pins no longer exist 5 of 8
Pinterest retired Idea Pins and folded them into standard Pin creation in 2023. There is no such option to select any more. Follower growth is also the wrong objective for a business chasing bookings.
3. The prescribed volume is unsurvivable 8 of 8, unanimous
That is 35 pins plus 6 videos a week, agency pace, for someone who answered "medium capacity" and "no time." The report correctly identifies inconsistency as her core problem and then prescribes the exact volume that recreates it. One reviewer called it "a relapse plan." "No exceptions" is the rigidity that ended her last attempt.
4. The "warm up the algorithm" ritual is a myth 6 of 8
No platform has an account-warming signal. TikTok distribution is judged per video, not gated by a pre-posting ritual. Worse, the sequencing means she does not publish anything until day 15, which is how the last attempt died.
5. The comment-to-DM trick is not automatic 6 of 8
This needs ManyChat or a Meta-compliant automation, with setup steps and policy limits. Presented as a native toggle, the post will run and nothing will fire. Worth noting: the panel agrees the tactic itself is one of the strongest organic lead mechanics on Instagram right now. Only the missing tooling is wrong.
6. Fantasy benchmarks 3 of 8
"Bio link clicks: 1-3% of reach," "aim for 5%+," "at least 5 email signups from Pinterest in the first 30 days." Real profile-to-link-click rates sit well under 1% of reach, and Pinterest indexing ramps over 3 to 6 months. These manufacture a fake failure at her Week 4 review.
7. Invented client results, handed over as caption copy 2 of 8
"A financial coach who doubled her inquiries... a wellness practitioner who booked three retreats..." These are fabricated outcomes given to her to publish about named client types. That is an FTC endorsement exposure and a credibility risk in a referral business.
8. The AI avatar upsell contradicts her own answer 4 of 8
Her quiz answer is open_to_ai_avatar: "no". The closing upsell pitches "a lifelike, personalized AI avatar of you," inside a report about authenticity, to a photographer whose product is human presence. This is pipeline logic, not copy.
9. Third-party repins are dead advice 3 of 8
"2-3 repins (curated from others in your niche)." Pinterest deprioritized redistribution after the fresh-Pin shift, and repinning competitors sends her traffic away.
Where the panel disagreed
- Instagram's "first few hours." Four called the velocity framing outdated, arguing sends per reach, watch time and saves now dominate over days and weeks. Two defended it as close enough. The report also contradicts itself, correctly saying elsewhere to "create content worth sharing privately."
- Reviving the dormant account. Seven of eight endorsed keeping the existing handle. One argued that 228 followers dormant for 914 days are ghosts that drag initial distribution.
- The 1.5 second hook. Three called it accurate, two called it too rigid, preferring 1 to 3 seconds with completion rate as the real lever.
Fit problems: the worst-scoring axis
Facebook is mishandled 8 of 8, unanimous
The report logs "Also found: Facebook (noted, not deep-audited in this report)," calls it "a proven lead-gen surface for group and local reach," and then demotes it to "worth adding later," behind two platforms she does not have. She already owns it. For local Boulder family and couples work, the neighborhood and parent groups are literally where "can anyone recommend a family photographer" gets asked, and cross-posting a Reel there costs nothing.
LinkedIn is missing entirely 4 of 8
X gets an explicit "do not bother" note. LinkedIn gets no mention at all. Half her revenue is branding photography sold to business owners, LinkedIn is the one place a bad headshot is publicly embarrassing, and it is the last major surface where text plus one image still reaches decision makers with no ad spend. A before-and-after headshot post is arguably the highest-intent asset this business can make.
TikTok is ranked too high at number two 7 of 8
TikTok is a global discovery engine with weak local booking intent, and the report never instructs geo-targeting beyond a single example. Building it from zero at three times a week, with its own daily routine, while her Instagram has been dead for 914 days, is the wrong second bet.
One B2B lead magnet erases half the business 6 of 8
"Every piece of content points to your Branding Photo Style Guide." A mother looking for fall family portraits will not download a branding guide, yet the entire 3-email nurture and every platform CTA is branding-only.
Google Business Profile omitted 3 of 8
Not social, but "Boulder family photographer" search converts better than anything in this report, and reviews plus geotagged galleries are a one-hour setup. A lead-gen report that prescribes 35 pins a week and skips GBP has its priorities inverted.
Pinterest: the panel split
Five reviewers say it is ranked too high and scoped far too heavy for a "leads now, no time" operator, since it is a 6 to 12 month compounding play that needs blog or landing pages she does not have yet. One argued it should be number two, ahead of TikTok, because booking intent there is stronger. Consensus resolution: keep it, move it to number four, cap it at 5 to 7 Pins a week from images she already owns.
An internal contradiction 2 of 8
"Your Situation" says she is "starting from zero" two paragraphs after the audit correctly reports 228 followers and a business 1 to 3 years old.
Actionability gaps, with rewrites
Week 1 versus its own time estimate 4 of 8
Week 1 asks for 6 Instagram posts, 6 TikToks and 30 Pinterest pins, estimated at "8 to 10 hours total." Thirty custom Canva pins alone is 4 to 6 hours for a non-designer. Realistic total is 16 to 20 hours.
The daily routine is unsustainable 4 of 8
The 30-5-2 habit on both TikTok and Pinterest, plus four separate daily check-ins, totals roughly 75 minutes a day.
"Content worth sharing privately" is never explained 4 of 8
The content buckets give no production method 3 of 8
Profile instructions are too loose 2 of 8
Review day is a non-instruction 2 of 8
"If on track: double down on what's working" and "no need to overthink, stick to this rhythm, and the leads will flow."
Sharp single catches
Points only one model made that are clearly right and worth acting on.
- MailerLite's free tier is understated by 4x. The report says "Free plan: 250 subscribers." It is 1,000. It also calls Kit's automated welcome emails "paid plans only," which is out of date. This makes the correct recommendation look worse than it is. (Claude Opus 5)
- The "limited-time discount for the first 10 spots" is the wrong lever. Discounting a premium branding photographer's first offer trains the exact price-shopper she does not want, in a report otherwise built on authenticity. Use a booking window or a session add-on. (Claude Opus 5)
- The Pinterest success metric contradicts the stated goal. The report names save rate as the key Pinterest metric, but the goal is lead generation, which makes outbound clicks the one that matters. The milestone list already puts clicks first, so the report argues with itself. (Kimi K3)
- The conversion post is assigned to the weakest format. The plan puts it in the static slot, which gets the least distribution. Conversion content should run as a Reel or carousel. (Grok 4.5)
What the panel says is genuinely strong
- The account audit is the best thing in the report. Real pulled numbers (228 followers, 0 posts in 90 days, dormant 914 days) and the line "closing that gap matters more than any tactic in this report" was called the truest sentence in the document by three reviewers.
- Funnel coherence. One lead magnet, one CTA sentence, one nurture sequence, every platform pointing at the same opt-in. Rare discipline at this price, and it removes the decision fatigue that kills solo operators. The fix is to duplicate the funnel for the family side, not to scatter it.
- The written assets are paste-ready. Full TikTok scripts, hook formulas, bio copy, board names, keyworded Pin descriptions, 15 first-week post prompts. Multiple reviewers said this is why the product sells.
- Revive, do not restart. Seven of eight endorsed keeping the existing Instagram handle and skipping setup in Week 1.
- Correct current tactics, named plainly: the comment-to-DM keyword mechanic, the 50/30/20 format split, using TikTok search wording in speech and on-screen text, no hashtag spam, Pinterest 2:3 at 1000x1500, the batching method, and the owned-email-list argument.
Priority stack
ENGINE means it fixes every future report. ONE-OFF means it only affects Laura's.
| # | Change | Scope |
|---|---|---|
| 1 | Kill the Pinterest daily quota and publish a weekly hour budget. Replace "pin 5 times daily, no exceptions" with 5 to 7 fresh Pins a week, and state the plan's total cost as 4 to 5 hours in one batch day, inside the report itself. | ENGINE |
| 2 | Fix the three hard factual errors globally: remove Meta Business Suite as a Pinterest scheduler, remove Idea Pins, remove third-party repin quotas from every Pinterest playbook. | ENGINE |
| 3 | Add a hard rule that every platform the customer already runs gets an activation plan before any new platform is introduced. Here that means Facebook to number two, TikTok down to repost-only. | ENGINE |
| 4 | Add LinkedIn to the platform logic whenever the customer serves business owners, and give this report a LinkedIn slot at number three: personal profile, 1 to 2 posts a week, a before-and-after headshot post. | ENGINE |
| 5 | Split the funnel: keep the Branding Photo Style Guide with the STYLE CTA, add a Boulder family session prep guide with a FAMILY CTA. The dual-audience trigger itself is an engine fix. | ONE-OFF |
| 6 | Delete the warm-up week and the 30-5-2 ritual, delete the fabricated client results, replace invented KPI targets with absolute numbers, and hard-gate the upsell so a customer who says no to AI avatars is never pitched one. | ENGINE |
Where the fixes live. Most of the bad facts are hardcoded, not invented fresh each time, which is good news: they are cheap to fix and the fix reaches every report. Meta Business Suite, Idea Pins, the repin quota and the 30-5-2 ritual all sit in the report engine's prompt file. The MailerLite subscriber number sits in the report assembler.
How this was run, and what to distrust
Eight models reviewed the complete 32,225-character report in parallel, each given Laura's exact quiz answers alongside it so they could judge fit and not just prose. A ninth pass, Claude Opus 5, synthesized the eight into this document. Counts throughout are how many of the eight independently raised a point.
Two harness bugs were found and fixed mid-run, and they are worth remembering for the next panel:
- Empty reviews that looked successful. The first run gave each model a 2,000-token budget. Reasoning models spend that budget on hidden reasoning before writing anything, so three of the eight returned a completely empty review while still reporting success. Raising the budget and capping reasoning effort fixed it. Any panel that reports "all models responded" should still be checked for response length.
- The report was being cut before the models saw it. The harness truncated it at 24,000 characters, so all eight models saw a document ending mid-sentence and several reported it as a broken product with a missing Week 4. It is not broken. That finding was discarded and the panel re-run on the whole thing.
Also checked by hand rather than taken on trust: Week 4 is present and the report ends cleanly, and "Boulder" and "Laura" were correctly extracted from her live website rather than invented.
What is not covered here: nobody checked whether the report's advice actually produces bookings, because that needs Laura to run it. The panel judges accuracy, fit and executability, not outcomes. The scores are a weight of opinion across eight models, not a measurement.
Panel run July 25, 2026 · Private · SocialMapped internal QA · ← Hub