The Most Powerful Sales Tool at Lowe's: Satellites | Hacker News Reader

The Most Powerful Sales Tool at Lowe's: Satellites

businessweek.com

31 points1337biz39 comments

The actual information is less exciting:

Laura Champine - Canaccord

Good morning. Thanks for taking my question. Your close rate performance was impressive, how do you measure that?

Bob Hull - Chief Financial Officer

Laura, so a couple of different ways, so of late, we have been using satellite imagery. So we take pictures of parking lots throughout the course of the year. We max that up with actual transaction counts in stores. Of late we have been actually using some technology that involves traffic counters in the stores, which gives us close rate by day, by hour, which is going to further allow Rick and the team to optimize labor going forward. We have tested both methodologies for the same stores and got similar results. We are pretty comfortable with the methodology. It allows us to forecast and see actual improvement in close rates

From the earnings call:

http://seekingalpha.com/article/2051323-lowes-companies-ceo-...

>> "A little creepy? Yeah. Smart? Definitely."

Is it creepy? They're only looking at car park occupancy in aggregate. They're not tracking car number plates.

I don't get the creepy either. It seems like an ok plan to judge volume[1] of traffic compared to sales. Might even do some analysis of what vehicle types are going to the store.

1) I am with other posters and would have put up security cameras and done my count that way

As long as I don't start getting "we noticed you parked in our parking lot but didn't buy anything ..." e-mails, I'm totally ok with it ;)

I thought the story was going to be about Lowe's using satellite imagery of neighborhoods to decide where to target marketing about new roofs, outbuildings, lawn care, sidewalk maintenance, etc. So yeah, using it to watch their own parking lots is... not creepy at all.

I'm puzzled why they would go to the expense of using real time satellite imagery for this, versus mounting a camera on a light pole in the parking lot.

This is the exact same thought I had when reading the story.

Not only would it be cheaper, but you could follow each vehicle through the lot and figure out the traffic flow patterns better.

Plus, to tie in with this other front-page story ( https://news.ycombinator.com/item?id=7346629 ), you could track the plates of the cars, and tie that in with retail cellular tracking ( https://news.ycombinator.com/item?id=5154089 ). Facial recognition still needs a couple of years. Over time, you could tie the cellular phone, vehicle, and electronic payment (debit or credit card) to an individual.

Continual tracking.

And all brought to you by technology. In the words of Pogo: "We have met the enemy...and he is us."

That, and you can see the parking lot on cloudy days too.

> you could track the plates of the cars, and tie that in with retail cellular tracking ... Facial recognition still needs a couple of years. Over time, you could tie the cellular phone, vehicle, and electronic payment (debit or credit card) to an individual. Continual tracking.

No thanks. The day that happens is the day I stop shopping there. That would be the "creepy" mentioned in the story.

I would guess ease of roll out. It is a MVP, a test that could be run on a large batch of store without having to wait on a bunch of hardware being installed. It likely offers a similar picture to the alogrithyms, making the mining less difficult. Since it is proving worthy, parking lot cameras might be next.

They don't. It isn't real time imagery, they are correlating the number of cars in the parking lot with the transaction counts from the store to estimate how many of the people that showed up at that particular time actually bought something.

Because that's about twenty more fucks than anyone cares to give. The problem is "what is our occupancy in the parking lot"--the solution is as simple as just buying a few sat images.

Cameras imply installation, monitoring, data storage, analysis...it's just a colossal pain in the ass compared to buying some imagery and being done with it.

Because from space they can also take images of the Home Depot parking lots. (maybe?)

Or perhaps even just counting the number of people walking through the doors vs. the number at the checkout tills?

Why not track how many times the entrance/exit doors open or close? Or add a laser(?) that cuts across the entrance that counts each time it's line is broken. Directional but also simple to do.

you would need to account for employee traffic then, plus customers who run back in on the occasional oops I forgot

A lot of companies do this actually. Sure its not perfect, but it gives a good enough estimate.

It doesn't sound like it would make sense to do this given that stores generally have counters on the readers that detect stolen tagged items on exit. Surely it would be easier to use this data to figure out when the busy periods fall?

How about if a hardware chain used satellite imagery to detect new structures being built, like sheds, barns, etc, and mailed targeted coupons out to the property owners?

Isn't that a bit late in many respects though? The holy grail for a hardware company would be knowing when a person was thinking about building a new shed, barn, etc, not when they've already bought all the wood and nails to begin construction.

That said, we already know they're trying their damnedest to track just that.

Isn't permitting info in the public domain? I wonder why they don't leverage that?

Here in the UK public domain does not equal easy to get.

Planning applications are available for viewing at the local planning office but they are not available online (though I know a guy who works at the local council who is pushing for them to put all their data online as part of the open government initiative so change is slowly coming here).

Good point. I wonder if that's a startup idea there? Go to local city halls and gather permitting information to sell to retailers?

Shameless plug - the company I work for is (sort of) doing this for tons of public data sources. http://enigma.io

A good question. My assumption is that even though it's considered public knowledge, there is still a cost for it and it's not electronic.

For instance, in my state and/or township one can go to township office and request a copy of all the property taxes for any piece of land, no matter who owns it. It costs something like ten dollars and in return you get a hardcopy with a bunch of numbers on it.

Assuming the process for getting copies of new permits is similar, the biggest downside is cost for the information and paying for some intern or whatever to type it in. Even if it's five dollars plus intern is decent chunk of money (I have no idea what the profit margin is on lumber, etc). And then you have to factor in the cost of the promotion which will eat into yet more profits.

In Houston, it's electronic and searchable. See: http://www.cohtora.houstontx.gov/ibi_apps/WFServlet?IBIF_ex=...

when a person was thinking about building

Such as when they file a building permit request. Which I suspect may be public information in many jurisdictions.

They might want to monitor building permits, so they can get in ahead of when actual construction starts.

Also - the gold coupon is the 10% off one from Lowes that comes in the US Postal Service change of address packet.

Monitoring just issued permits may only useful if the home owner is the one doing the work (which is often, but in large metro areas, may be a very small subset of the actual permits being issued). Contractors tend to have their own suppliers for their materials. Not to mention, that often, by the time permitting process is started, much of the materials, appliances, etc. have already been chosen, so coupons only work if they're relevant to the chosen materials/appliances/etc. (Anecdotally, I'm going through this right now. The permitting process takes so long, that nearly every thing is already decided on by the time the permits are issued.)

I think useful permit indicator is to look at recent past permits. Neighbors tend to be inspired by their neighbors to do long-needed projects. As soon as I started talking about doing work on my house, other neighbors started asking for the name of my contractor and started talking about how maybe they should "fix up that garage, or re-do the kitchen."

That's not a bad idea. I think addresses can be matched to postal delivery routes, so they could be a little more targeted with their mailing.

Neat.

Satellite imagery isn't terribly expensive and Lowes has 1830 stores.

I'd bet it's a lot more efficient in the end to just buy a consistent set of images for X stores of interest, X times a year than worry about the installation, maintenance of location specific concerns of cameras/sensors at every store.

Kroger uses infrared near the checkout lines to similar effect: http://www.timesdispatch.com/business/retail/infrared-techno...

really cool? yes. Impractical? probably?

Scanning earth using satellites is really cool. Maybe they're outsourcing this to a central company, and paying a hundred bucks per store per month. fine. But to be honest, there are some more effective low tech ways to catch these numbers, for example by looking at the number of transactions, or installing sensors on every entrance to count the number of people entering.

Seems like they're shooting a pigeon with a cannon.

>But to be honest, there are some more effective low tech ways to catch these numbers

Heck, just have someone count cars and write down the time whenever they retrieve shopping carts from the parking lot.

You think any of that is as cheap as a hundred bucks a month per store?

"Satellite imagery" is the most powerful Sales Tool?

I understand the need for "link-baity" headlines but this is flat-out wrong. The very same article goes on to state some numbers - 1% improvement vs 2% reduced labor.

If they can attribute it ALL to Satellite Tracking as opposed to say other promotions/general economy & barring other accounting tricks, it is still too little to be even attributable.

At the hospital I work at we correlate 'sales'in a similar way. Just we found a cheaper way. We noticed that at night less people come in. So we reduced staffing at night.

I don't understand why they don't correlate EPOS, door and aisle counters, put it in their data warehouse, and have precise close rate based on 1) visits 2) store departments

I'd have thought this was totally standard practice!

Still though, really cool that they have confidence that satellite imagery is a good proxy for this.

Hm, what the article doesn't say is that Lowes are typically located near Home Depot. Definitely close enough to get a snapshot of competitors parking lot too.

Reply on news.ycombinator.com