The heat map you hand a client is two different things at once. Where your survey points sit close together, it is real data. Where they sit far apart, it is the software guessing between them. Same map, same colors, same confident gradient. The reader cannot tell which parts are measured and which parts are interpolation. You can, because you walked it. So the quality of that walk is the quality of the deliverable, and almost nobody talks about how to walk one that holds up.
I have scored real survey files to prove the point, and the numbers are blunt. Walk tight and the map is measurement. Walk loose and large parts of it are the software inventing whatever makes the gradient look smooth.
The map is interpolation, and you decide how much
Every survey tool builds its heat map the same way. It takes the points where you actually measured, then fills the space between them by interpolating. Close points mean short fills and a map that tracks reality. Wide points mean long fills and a map that invents whatever makes the gradient look smooth. That invented part is the software’s best guess, not your data.
This is the one thing you control in the field. Not the AP placement, not the client mix, not the building. The walk. You decide how much of the final map is real and how much is guesswork, and you decide it with your feet and your clicks.
I built a small scoring method to put a number on that decision. It looks at how you walked and grades the one thing that actually drives map accuracy: how close your points are to each other.
The math, in plain language
There are two honest halves to this, and I am only going to claim the one I trust.
The trustworthy half is spacing. Take every consecutive pair of points on your walk, measure the real-world distance between them, and look at the average. Then drop it on a simple ramp:
- Mean spacing of 3 m or less scores 100. A dense, continuous walk.
- Mean spacing of 8 m or more scores 0. Stop-and-go with wide gaps.
- In between, it slides straight down the line.
Those two anchors are not arbitrary. They come straight from my own field guidance on why you do NOT use a stop-and-go survey. A good continuous walk averages 3 to 4 m between points. A stop-and-go survey averages 8 m or worse. The ramp just turns that rule into a score.
A good average can still hide a few ugly gaps, so I also report the 90th-percentile spacing. That is the gap that 90% of your points beat. If your mean is a clean 3 m but your p90 is 12 m, you have sloppy stretches an average alone would paper over. The mean tells you the typical walk. The p90 tells you the truth about your worst stretches.
There is a second half, coverage density, that asks whether you actually covered the whole space or just walked tight lines down the middle. I am not going to dress that one up. It is not calibrated yet, so I leave it out and report spacing plus the distance you walked. When I say spacing is the trustworthy half, I mean the other half is honestly still in the shop.
Four real surveys, one floor, four very different walks
I ran four example survey projects through this. Same floor plan, same software, same physics. The only variable was the walk, and the scores split exactly the way the theory says they should.
| Survey | Walked | Mean spacing | Score |
|---|---|---|---|
| RTLS | 827.9 m | 3.67 m | 87 / Excellent |
| Mobile | 477.7 m | 6.37 m | 33 / Poor |
| Multi Device | 578.7 m | 8.15 m | 0 / Poor |
| Hybrid 802.11ac | 437.0 m | 8.56 m | 0 / Poor |
Read that and the lesson writes itself. The RTLS survey was walked tight, 3.67 m between points across 7 sessions and about half a mile on foot, and it scores Excellent. The other three were clicked 6 to 9 m apart, and the score flags every one of them as wide sampling, which is its whole job.
One thing I owe you before you react to those Poor scores. These four are Ekahau’s bundled demo projects, built to show off survey types, not to be careful production work. The wide spacing is exactly what you would expect from a demo, and the Poor scores are the method working, not a knock on anyone who touched those files. I am scoring the walks, not the surveyors.
Keith’s 7 rules, and the lever each one pulls
I have taught the same short list of survey rules for years. What I had never done until now is line them up against the scoring math and watch each rule move a specific number. Every one of them does. Follow them and the score takes care of itself.
Click when you start. That anchors the true beginning of the walk so the path does not guess its way back to wherever you last stood.
Click when you stop. That ends a walked segment cleanly, so the empty gap to your next starting point never gets counted as a measured stretch you didn’t actually walk.
Click when you change direction. Put a real point at every turn. Skip it and the tool draws a straight line through a corner you never walked, then interpolates data along a path your feet never took.
Walk both sides of anything you care about. Two passes double the real sampling around the things that drive decisions. That is coverage where it counts.
Check the known-important rooms. The CEO’s office. The boardroom. The one corner where a complaint will land on your desk by Monday. Walk those on purpose so the map shows measured data exactly where it would hurt most to be guessing.
The closer your data points, the more accurate the data. This one is the spacing sub-score stated out loud. Tighter points mean less interpolation, 3 m or under earns full marks, and there is no trick to it beyond clicking more often.
Click around a pause. This is the one that surprised me when I watched it in the math. Stop to talk, and click when you stop and again when you walk on. Here is why it works. When you stand in one spot and the tool logs several points within half a meter of each other, the method collapses that standing cluster into a single node before it does any spacing math. Standing still should never make your survey look denser than your feet made it. And clicking when you resume keeps your chat from fusing into one giant 30-meter leg that would read as textbook stop-and-go. The dwell-pause rule and the dwell-collapse math are the same idea seen from two ends, one in your hands and one in the score.
What a new surveyor actually needs to remember
You do not need the math. You need two habits. Click often and walk tight.
Click at the start, at every stop, at every turn, and the moment you pause or pick back up. Keep your lines close together and walk both sides of what matters. Do that and your average spacing drops under 3 m, the heat map becomes measured data instead of a smooth guess, and the score follows on its own.
The score was never the point. It is a mirror held up to your walk. A Wi-Fi survey is only as good as the walk that produced it, and now you know exactly how to walk one worth handing over.