Method
Site screening vs sales forecasting: what each model should decide
Screening filters a long list of candidates quickly. Forecasting sizes the survivors before you commit. A working expansion pipeline runs both, in that order.
Quick answer
Screening and forecasting do different work. Screening moves fast across many candidates and drops the ones that fail your gates, leaving a short list worth deeper attention. Forecasting takes that short list and estimates how much each survivor should produce in sales, traffic, or membership. Geod screens with explainable scores, then forecasts the few sites that earn it and attaches a confidence level to every estimate.
Two jobs, one pipeline
Teams often treat site screening and sales forecasting as a single task, and the overlap costs deals. A screen filters: it scans many candidates and decides which ones deserve deeper work. A forecast measures: it takes the few that survive and estimates what they will actually do. One ranks fit across a long list. The other sizes performance for a short one. The pipeline holds together when these run in sequence.
What screening decides
Screening makes one decision across a wide field: which candidates deserve a closer look.
A screen runs for speed and breadth. Point it at a list of addresses or trade areas, and it ranks them on fit using drive-time access, demographics, competition, and your own network nearby. The output is a relative ranking plus a pass-or-fail against the gates you set, so a long list of maybes collapses into a short list of contenders. Nothing in that output is a sales estimate. Sizing comes later.
Screening rewards restraint. At this stage you are not pinning down any single site. The goal is to reserve the expensive work, the site visits and broker calls, for candidates that already clear the bar. In Geod the explainable score does this, ranking sites on traceable components so you can see why each one rose or fell.
What forecasting decides
Forecasting works on the survivors, estimating how much one specific site should produce.
A forecast trades breadth for depth. It takes a candidate that already passed the screen and estimates expected performance, sized for that location and its trade area, whether the metric is sales, visits, or members. It draws on the same geography the screen used and pushes further, accounting for the demand the site can capture and the share that competitors and your own units already hold, with a confidence level attached to the estimate.
A forecast is expensive to produce and defend, so it runs on a handful of sites. Its job is to support a real commitment, a lease or a capital request, which means it carries a number you are willing to put in front of a committee, along with the assumptions and confidence behind it.
Why you need both: the funnel
Screening and forecasting are stages in one process. Each handles a volume the other cannot.
At the top of the funnel you may have dozens or hundreds of candidates, far too many to forecast one at a time. Screening narrows that field to a short list worth deeper work. Forecasting then sizes the survivors, so you can rank them on expected performance rather than fit alone and commit to the best few.
- Top of funnel. Many candidates at a low cost per look. Screen all of them to clear the obvious misses and surface contenders.
- Middle of funnel. A short list of survivors. Layer on overlays for cannibalization, feasibility, and confidence to separate genuine contenders from the merely plausible.
- Bottom of funnel. A few finalists. Forecast expected sales or visits, attach a confidence level, and produce a brief that a committee can approve.
The cost of confusing them
Collapsing these jobs into one produces two failures. The first is forecasting too early, putting a precise sales number on every candidate at the top of the funnel. That approach is slow and expensive, and it spends your best analysis on sites a quick screen would have rejected. Precision on a doomed candidate is precision wasted.
The second failure is approving on a screen alone. A high score confirms that a site fits your pattern. It says nothing about how much the site will do, or whether the location simply pulls sales from your store two miles away. Sign a lease on the score without a forecast and an overlay, and you can open a site that scored well yet misses plan because the demand was already yours. Screen to shorten the list. Forecast before you commit.
Four objects, four jobs: score, forecast, overlay, recommendation
A clean pipeline keeps these distinct rather than blending them into one ambiguous number.
- Score. An explainable score orders candidates by how well they match your criteria, with every component visible. This is the screening engine, and it carries no sales figure.
- Forecast. An estimate of expected sales, visits, or members for a survivor. It answers how much, once a site has earned the deeper look.
- Overlay. Cannibalization against your network, feasibility constraints, and a confidence level qualify both the score and the forecast, so you know how much weight each number can bear.
- Recommendation. The decision itself: approve, reject, research, or revise. It is what the score, forecast, and overlays support together, written down so the reasoning travels with it.
Screening vs forecasting at a glance
| Dimension | Screening | Forecasting |
|---|---|---|
| Purpose | Filter many candidates and rank fit | Size expected performance for survivors |
| Inputs | Drive-time access, demographics, competition, your network, gate thresholds | Trade-area demand, competitor and self share, cannibalization, confidence |
| Output | A relative ranking and a pass or fail against gates | An expected sales, visit, or membership estimate with confidence |
| When in the funnel | Top: run on the whole list, cheaply, first | Bottom: run on a few finalists before you commit |
Frequently asked questions
- Is screening just a rough forecast?
- No. Screening ranks fit and filters candidates without estimating sales. A high screen score means a site is worth deeper work, not that it will hit a particular revenue number. Forecasting is the separate, deeper step that sizes the survivors.
- Can I skip screening and forecast everything?
- You can, but it is slow and expensive, and it spends your best analysis on candidates a quick screen would reject. Screen the long list first, then forecast only the short list that clears your gates.
- Can I approve a site on its screen score alone?
- It is risky. A screen confirms fit, not magnitude, and it does not catch cannibalization on its own. Before a lease or build, add a forecast and overlays for cannibalization, feasibility, and confidence so you commit on expected performance rather than pattern matching.
- How do score, forecast, overlay, and recommendation fit together?
- The score ranks fit, the forecast sizes performance, overlays add cannibalization, feasibility, and confidence, and the recommendation, approve, reject, research, or revise, is the decision those three support. Keeping them distinct is what makes the call defensible.
- How many candidates should I forecast?
- Only the survivors of screening, typically a small fraction of the original list. Forecasting is expensive to produce and defend, so reserve it for finalists headed toward a real commitment.
Related resources
See Geod on your next location
Geod is in a pilot program right now. Book a short walkthrough and we will score a candidate location with you: an explainable score, a drive-time trade area, competition, cannibalization, and a site brief.
Prefer the method first? Read the Geod methodology.