Open five browser tabs and search the same three words: Johns Creek home prices. Redfin will tell you homes are closing at a median of $700,000. Movoto will show a median list price of $769,000. Click into Zillow's main city page and you get a typical value of $651,407. Click into a different Zillow page for the same city and the number drops to $578,417, with a year-over-year change of 24.3 percent instead of the 0.3 percent the first page reported. Widen the search to a cost-of-living site and the median home value falls again, to $525,100.
Five sources. Five numbers. A spread of roughly $348,000 for a single Fulton County city in the same calendar year. None of these sites is lying. They are answering five different questions and calling all five answers "median."
If you are trying to price a listing, write an offer, or just figure out whether Johns Creek is getting more or less expensive, that gap matters more than any single number on the page. Here is what each figure is actually measuring, and which one you should reach for depending on what you are trying to decide.
Same Month, Same City, Two Very Different "Medians"
The clearest example of the problem sits in February 2026, when two well-known portals both claimed to report the median sold price for Johns Creek that month. Redfin put it at $700,000. Movoto put it at $873,215. Same city, same month, same stated metric, a $173,215 gap.
That is not a rounding error or a stale cache. It reflects how each company samples and cleans MLS data before it publishes a headline figure. One platform may include or exclude certain property types, apply different date cutoffs for when a sale "counts," or weight condos and townhomes differently than detached single-family homes. Johns Creek's closing volume in a given month is small enough, often somewhere between 45 and 130 sales, that these methodology choices can move the median by six figures without either company doing anything wrong.
Here is a plain look at what five commonly cited figures for Johns Creek actually represent:
| Source | Figure | What it measures |
|---|---|---|
| Redfin | $700,000 | Median price of homes that closed, per MLS-based sold data |
| Movoto (market trends page) | $873,215 | Also labeled median sold price, same month, different sampling |
| Movoto (July 2026 listings page) | $769,000 | Median asking price of homes currently for sale |
| Zillow (main city page) | $651,407 | Modeled "typical value" across all homes, not just those on market |
| Zillow (home-values page) | $578,417 | Same underlying model, restricted to a "middle price tier" snapshot |
Once you see the table laid out, the pattern is obvious. These are not five measurements of the same thing. They are measurements of five different things that all happen to be labeled "median."
List Price Is a Guess. Sold Price Is a Fact. A Model Is Neither.
The most useful distinction for a buyer or seller is the one between what sellers are asking and what buyers are actually paying. Movoto's July 2026 figure of $769,000 is a median list price, meaning it reflects what current sellers hope to get, not what anyone has agreed to pay. That same page noted list prices had fallen 7 percent from June and 6 percent from a year earlier, alongside a median 35 days on market. A falling list-price median in a market where homes still take over a month to sell is a signal sellers are adjusting expectations downward in real time, which is a genuinely useful thing to know if you are about to price your own home.
Compare that to Redfin's $700,000, which reflects closed transactions, meaning money that has already changed hands. If you are a buyer trying to gauge what a seller will realistically accept, the sold figure is closer to reality than the list figure, because list prices are opening offers in a negotiation, not outcomes.
Then there is the third category entirely: Zillow's modeled values. These are not based on current listings or even recent closings alone. They come from a statistical estimate, updated monthly, that tries to price every home in a market whether or not it is for sale. That is a genuinely different exercise than averaging what sold last month, and it explains why Zillow's own two pages for the same city can disagree with each other by nearly $73,000 and by a 24-point swing in year-over-year change. One page's snapshot restricts itself to a "seasonally adjusted middle price tier," which by definition excludes the top and bottom of the market and can produce a very different trend line than a raw median that includes everything.
None of these three approaches, list, sold, or modeled, is wrong. They are just not interchangeable, and most portal homepages do not tell you which one you are looking at.
Forty-Some Closings a Month Is Not a Lot of Data
There is a second reason these numbers jump around, and it has nothing to do with methodology. Johns Creek is not a market with thousands of monthly closings to smooth out its median. When the sample is small, a handful of transactions in either direction can swing the whole city's number.
Johns Creek's housing stock stretches from condos like those at Rivermont Village Condominiums to established swim and tennis neighborhoods like Sugar Mill, Medlock Bridge, Nesbit Lakes, Horseshoe Bend, Cambridge, Willow Springs, and Legacy Park, up to gated golf communities like St. Ives Country Club and Country Club of the South. That is an enormous price range packed into one city boundary. If three or four closings in a given month happen to land in one of the higher-end enclaves instead of being spread evenly across price tiers, the citywide median can shift by tens of thousands of dollars purely as a function of which subdivisions happened to close, not because the broader market moved at all.
This is the piece most headline numbers leave out. A median calculated across every price tier in Johns Creek tells you something about the city as a statistical entity. It tells you very little about what a three-bedroom traditional in Sugar Mill or a golf-course lot in St. Ives will actually sell for, because your specific subdivision's price behavior can move in the opposite direction of the citywide figure entirely.
Which Number Should You Actually Use
The right number depends on the decision in front of you, not on which site loads first.
- If you are writing an offer, weight the sold-price data more heavily than any list-price median. Sellers ask what they hope for. Buyers pay what the market bears. The gap between Movoto's $769,000 list figure and Redfin's $700,000 sold figure is roughly the size of that negotiating room in practice.
- If you are pricing a home to sell, a citywide median of any kind is close to useless. What matters is the median sold price for comparable homes in your specific subdivision over the past 60 to 90 days, a figure no portal homepage displays but one a local agent can pull directly from MLS records.
- If you are trying to gauge the general direction of the market rather than price a specific transaction, look at the trend within a single source over time rather than comparing one site's number to another's. Movoto's own July 2026 figures showing list prices down 7 percent from June and 6 percent from a year earlier tell a more coherent story than trying to reconcile that number against Zillow's model.
A Few Questions Worth Asking Before You Trust Any Number
Why did my home's Zillow estimate jump by double digits this year while the citywide median barely moved? Zillow's modeled values and a raw citywide median are calculated differently, and a model restricted to a "middle price tier" can post a very different year-over-year change than a broader sold-price average, especially if your home sits near the edge of that tier.
Is Johns Creek's market currently going up or down? It depends which measurement you trust. Movoto's July 2026 data shows asking prices softening month over month. Zillow's main city page shows a modest typical-value gain. Both can be true at once, since one measures what sellers are asking now and the other measures a longer-running model of overall value.
Should I set my list price based on what I see on a portal? Treat any citywide portal figure as a starting orientation, not a pricing tool. The subdivision-level, recently-sold comparable is the number that actually predicts what a buyer will pay for your specific home.
None of this means the portals are broken. It means "median home price" is shorthand for a specific calculation, and the calculation changes depending on who is doing it and what they are trying to show you. If you are weighing a move in Johns Creek and want the number that actually applies to your subdivision and your price range rather than a blended citywide figure, Stovall Properties Group can pull the real comparable sales for your specific street and walk you through what they mean for your offer or your listing price. Call or text Sarah for a personalized market consultation.