Snack shop franchisee - grand-opening surge fade early warning: the opening month's sales are rented from the promo, so split full-price from promo sales, watch the member second-visit rate, and call the base business real only when full-price weeks flatten - never run payback math on opening-month numbers

You are an opening-period early-warning analyst
for a franchised snack store. I will give you
week-by-week data since opening; judge whether
my real base business has formed and how far
the promo-borrowing risk runs.

Context: the launch promo (loss-leader pricing
+ stock-up bundles + top-up gifting) ran one
month and sales were high. I know that number
cannot feed a payback calculation directly; I
need the rented part separated from the real
base.

I will give you weekly data (from opening
week): 
- Weekly total sales, split: promo-price sales
  (through promo mechanics) / full-price sales
- Membership: new members this week, and
  members making their second in-store purchase
  this week
- Non-promo-day daily sales (where the promo
  had gaps)
- The promo mechanics: which price tiers, the
  bundle thresholds, the top-up gifting size
- If available: nearby competitor moves (a
  new rival opening in the same window)

Judgment criteria:
1. The full-price weekly trend is the spine:
  promo sales fade with promo intensity by
  nature and count for nothing; full-price
  sales holding within [settlement band] for
  [settled weeks] consecutive weeks is the
  only test that the base exists
2. Member second-visit share = members making
  a second purchase this week / the stock of
  members whose first visit was at least
  [repeat cycle] weeks ago - a more honest
  metric than sales: signing up for the
  opening gift is nearly free; paying to walk
  in the second time is the business
3. Promo dependence = promo-price sales /
  weekly total, listed by week - dependence
  falling while full-price rises is a healthy
  transition; dependence stuck and full-price
  dead is the double-weak case, the most
  dangerous
4. Borrowing signature: if late-promo sales
  ran clearly above early-promo (stock-up
  effect), the post-promo dip will run extra
  deep - flag that stretch; payback math
  should treat it as an outlier and drop it

Output:
- One weekly table: week / total / full-price
  / promo dependence / member second-visit
  share / verdict (settling / dependence
  stuck / double-weak)
- The current verdict: does the base exist;
  if not, is what's missing traffic or repeat
  business (based on which is weaker, second-
  visit share or full-price sales)
- When warning fires, "which lever first":
  second-visit share rising but full-price
  weak - the problem is pricing and
  assortment; do not cut the marketing spend
  yet. Second-visit share flat too - the
  problem is location or the trade area, and
  more spend fills a leaking bucket
- A payback-math correction at the end: how
  to treat the opening period (drop it
  entirely / annualize from the settled
  weeks), with the method stated - but the
  payback month itself is mine to compute

Constraints:
- No "snack stores typically pay back in N
  months" market talk - your data does the
  talking, the market does not
- The settled verdict requires [settled
  weeks] + 1 weeks of data; with less, say
  "observation window not yet full" and issue
  no conclusion
- Competitor moves are explanatory only - not
  an excuse for weak full-price sales unless
  the moves I gave clearly overlap the promo
  window
- No "keep watching" / "stay confident" -
  every verdict lands on "which lever first"

Why it is written this way

The spine of the whole warning tool is splitting promo-price sales from full-price sales. Launch promos are essentially borrowing volume from the future - stockpiled goods take weeks to eat through, prepaid cards months to burn down, all of it keeping post-opening weekly sales looking alive while the promo's afterglow pays down its own debt. Full-price sales are the only metric the promo never contaminated, so making them the settlement spine gives the question "is this business real or rented" an answer that cannot lie.

The member second-visit share is more honest than sales because signing up for an opening gift is nearly free (who refuses a gift?) - the second, self-paid visit completes the one full commercial validation. Defining it as second visits over the existing member stock, rather than a vague "repeat rate," cancels the denominator distortion of opening-week sign-up floods: more signups, bigger denominator, diluted share - which is exactly right, since a thousand cards with fifty returns deserves the alarm more than three hundred cards with fifty.

The "which lever first" branching wires the warning to action: second-visit share rising while full-price is weak means people stayed but found nothing to pay for - a pricing and assortment problem, where cutting marketing spend strangles a savable store; second-visit flat as well means people never stayed - a location or trade-area problem where further spend fills a leaking bucket, and the stop-loss conversation should precede any expansion talk. The final payback correction gives method, not conclusions, because the answer to "payback in how many months" must stay in the owner's own hands - this tool's one job is ensuring nobody computes a fantasy out of rented sales.

Variables