App for Advanced search on Facebook Marketplace – Need Your Feedback

[paul_white]

Hi everyone,

I’m building an app for advanced search and management of car listings on Facebook Marketplace. It is different from other similar projects by having all listings in my database. You don't need to setup search by params or intervals - just search through the catalog like you normally do, but with improvements like advanced search.

The app is at an early stage and only covers California. It’s free to use, and no registration is needed.

Here is what it already can do:

* Tracks all vehicle listings, which appear on the webpage 15–120 minutes after becoming available on FB (more frequently during working hours).
* Filters out non-car and irrelevant listings (boats, bikes, auto detailing, parts, etc.).
* Allows filtering by additional fields (private/dealership, mileage).
* Your FB account is not used

These are my ideas for what could be useful (let me know if that is true):

* Evaluation of listings (and finding listings priced below market value).
* Showing new listings as soon as possible.
* Notifications about new listings matching search criteria.
* Listing management (bookmark, don’t show, export to file or API).
* Better filtering (by mileage, transmission, keywords; improved location selection).
* Better separation of private sellers and dealerships.

I’d appreciate it if you could try it out, leave a review, and let me know how I can make this app useful to you. Thanks for your time.

The app lives here: cleanttl.com

 

[paul_white]

Bumping my topic with a bunch of updates.

The app does a lot of listing processing to fix titles, determine car condition, identify recent repairs, and detect signs of high motivation to sell. I aim to help buyers find undervalued cars based on their parameters.

Having all listings in my database (as opposed to crawling on demand) enables a lot of useful features: finding underpriced cars, tracking price drops, exclude offers from dealerships, description analysis of different kinds, etc. I really need feedback from dealers, flippers, and other industry leaders to choose the right direction for my project.

Please give it a try: cleanttl.com

Still free and still California only.

Thank you.

 

[carguyfiles]

Pavel, a few notes from the dealer side. Disclosure, I build posting software for dealers, so I'm biased toward the seller's view.

Private vs dealer is going to be your hardest filter. A lot of dealer inventory on Marketplace goes up from reps' personal profiles, so the profile type lies. Stock numbers, a VIN in the description, ten-plus photos and studio-style backgrounds are better tells.

Price history beats a "below market" score. In our dealer data, 22.6% of listings changed price after they went live. A buyer seeing "dropped $1,500 last Tuesday" learns more than a green badge.

Crawl on the dealer calendar. Thursday is the busiest day for new dealer listings, about 3.3x Sunday, so a flat interval wastes weekend calls and runs behind on Thursday.

Watch for relists. Plenty of dealers delete and repost instead of marking sold, so the same VIN comes back with a new listing ID and a fresh "listed 1 day ago." If you key on VIN, you can show the real age.

One honest caution: read Meta's terms on automated collection before you scale past California.

What are you seeing on relists so far?

 

[paul_white]

Joe, thanks for your reply. It is a huge help.

At this moment, I have no re-list or "Private vs dealer" detection at all. These are my next priorities. Indeed, they are going to be tricky, but I have some ideas (yours included).

 

[KennySmithAutoUnite]

Pavel, looking at this from the used-car desk, finding something below market is one thing. Figuring out whether it’s actually worth buying is another.


A car can look like a great deal on Marketplace until you factor in recon, the title situation, transportation, and what you can realistically retail it for.


I’d be interested in a way to build an acquisition watchlist by year, model, trim, mileage, radius, and maximum buy number. Then show which listings are new, which have changed, and which might actually be worth making a call on.


I’d also separate what’s confirmed from the listing versus what’s being estimated. That matters when somebody is making a buying decision.


Are you looking to build this more for consumers hunting deals or dealers sourcing inventory? I’d approach those two pretty differently.

 

[paul_white]

Kenny, thanks a lot for your input.

Implementing watchlists is definitely on my list.

Accurate estimates accounting for recon, transportation, etc. - I am not sure if I can do this based on just the description.
What sounds more realistic is a rough estimate based on Year/Make/Model + mileage: "A typical 2018 Toyota Prius in good condition costs $11,000–15,000." Adding trim, mileage, and location will give more precise numbers. Would such an estimate be helpful?

My target audience is dealers and flippers. Here is my thinking. Some of them look for an underpriced car in good condition, while some look for cars needing repairs (because they can fix them cheaply). I want to give each a tool to find what they need. And I assume everyone wants a convenient search, notifications, and listing management.
Is that what dealers need when buying? Am I missing something important?

Thank you.