Découvrez Le Meilleur Générateur De Factures Gratuit Pour Les Entreprises

Best free invoice generator for businesses

what's good everyone welcome back to the channel and in today's video I will show you how with the help of AI and n8n you can scrape Google Maps for unlimited leads of local businesses for free and yeah that's actually pretty exciting um so as you will see I've have built two different workflows uh to get this uh done so I'm going to show you everything basically I'm going to explain what each note does how it works and how you can implement the exact same thing I will also put a link Down Below in the description where you can download um basically these workflows upload into your own n8n uh you know environment and run it you know as much as you want but first I have to give credit to uh Akram Kadri uh because I found this basically very similar build on n8n I just improved it for my own use case uh because um in this case previously was only extracting the emails and I noticed that it was extracting emails with weird symbols so it wasn't actually real emails but what I wanted I just wanted to know to exactly for which side the email uh is what search term was used to basically get uh that email and the location for which we scraped that email so just let me zoom in a bit so as you can see this is the data so I'm going to show how everything works um and now for example I can show you in real time how the workflow uh goes so for example we have the airor table trigger so I'm going to just hit uh test the workflow it's going to firstly generate different locations to scrape for uh with the help of AI and and now it triggers another workflow which is this workflow the scraper uh workflow uh that is responsible to actually extracting every single uh email from the website and then it adds the information into air table and right now you can see it running but for example if I go into uh executions over here you can see that uh the Google Maps scraper launch which is responsible for launching the scraper is running and then Google Maps scraper is running so uh again if I go back here you can see that already we iterated over third uh three different items and here as you can see it already added some of the emails for example uh in this case I'm trying to scrape HVAC companies uh and it's pulling out HVAC companies from downtown Los Angeles so I'm getting their email and I'm getting uh their you know website so for example know I can can see that I scraped support at top cooling guys and I have their website uh if I have a need I can go into the website you know do my own research and then maybe send them an email offering my services or whatever so now let's go get back and I'm just going to stop executing this because it's going to run for a long time but you can already see we scraped eight different locations for uh emails and this is the information that we got but uh now I want to explain what kind of approach we're going to be using uh for this scraper so if you go uh inside Google Maps again let me zoom in a bit you will see that for example if I type it in HVAC in Los Angeles Los Angeles and hit enter uh what it's going to do it's going to give me some of the results but as you can see it doesn't cover the whole area of LA and in total I'm getting like what around like 12 to 13 results which is definitely not enough if you want to script you know and get loads of leads or reach out to different businesses but for example if I write the same HVAC in uh West Lake uh I will get again I will get a smaller like region or sub region of Los Angeles but I will get uh additional 12 leads or 12 businesses in that region and that's why we're going to be using AI so we're going to be using AI to generate this exact search method for Google because as you can see HVAC invest L and it uh generates a search term for the Google um and this is how we're going to be basically scraping all of the districts regions and neighborhoods in LA in this case to generate let's say uh emails of HVAC businesses um and instead of getting you know like 15 leads from just all of the LA now we're going to scrape every single sub region and every single uh neighborhood and that way we're going to cover as maximum of area as we can so um now we can go into uh the Google Maps scraper Dash you know launch workflow so just to shortly explain this workflow is responsible for executing that previous workflow but how everything starts is that first I'm adding an air table trigger again oh it zoomed it so uh basic basically I'm triggering on the field created and getting like information from the grid uh view so let's say I'm going to fetch the test data um yeah and it fetches right now linked records but if I go into the location you can see I have a created field and I have a name and business HVAC so you can you know name these columns differently but uh this is scraping based on on uh these fields and this is what is used for the trigger but then from the trigger we pass the information to uh open AI so in this case uh I use 40 Mini model which is super cheap and it covers like the majority of areas that you want and now we can look at the prompt very quickly uh so it can look pretty scary but again I'm going to drop this in the link down below we can you download the whole workflow you can pull out the prompt but basically what it does here it uh tells the requirements how to format the output so we need business type Plus in plus sub region so HVAC Plus in Plus West Lake for example um again we give an example so HVAC and Los Angeles and then we say what kind of output we expect and then at the end I say please create search queries in and this pulls the name of a location from um the air table so this would be this name and City and business is going to be pulling out ex where is it this one so this uh variable so basically PR please create search queries in Los Angeles city and HVAC business and then important so I mentioned exactly how I want to have uh the output so I say uh provide me you know with an object which has a key of locations and array uh of this and then if we go into the structure at output so I just again wrote output and then again array so again this is not exactly perfect I should probably update this uh but this is how it's set up right now so for example we can test it test run this uh node and what you'll see is going to happen is going to generate just a a huge list of different uh neighborhoods and regions inside of Los Angeles because this is the specific um area that we're pulling so as you can see locations let's go to Json or let's just look let's be in table but there's locations and again HVAC in Crest View HVAC in downtown HVAC in Hollywood HVAC in in Echo Park HCK in vest lake so on and so forth so basically you know pulling um all of the neighborhoods or as much neighborhoods as we can again this can depend on that prompt then we are simply splitting out uh because when we're getting this data we're getting it into one single array so I'm using uh split out node and as you can see and then it just splits into array of objects which has output. locations so in the fields I set basically this I just dragged this locations and it uh creates a separate item for uh each one of these then I have edit Fields um so in edit Fields I just basically uh set two additional Fields so first is going to be the output uh so this is going to be uh for the search term as you can see and there's going to be a location uh and this is going to be for the city and I'm using this just so for example when I add records uh in air table I know for which big city this was and of course then I know for which uh search term it was as well uh so this is how that works and then once we have those 36 items what it does uh it Loops over every single one of them so let's say we have um HVAC in Crest View and Los Angeles so I'm going to just copy this for later on and then this item is going to be first added into the workflow and this is going to execute this particular workflow that does the scraping just important thing to know that for example when setting up this um so select your respective workflow and don't forget to add void for subw work for workflow completion that means that um it doesn't launch workflow one after another and it just you know doesn't overwhelm uh n8n and it doesn't duplicate the same records over and over again because it's going to launch um this workflow and it's then going to wait uh until it finishes so now let's see once um this Loop triggers this one how does this work so for here what I'm going to do I'm just going to add uh some MOG data so from the previously added uh workflow so that way you can clearly see what's going on and I'm going to save it and we pinned it so for example yeah let's say I run this uh module or node I go in over here and you can see I have the out output and location so this is the exact information that is being passed over from uh this workflow then we have the HTTP request in the HTTP request we use get method and in the URL we basically have Google Maps search URL if you go here you can see google.com/ mapsearch and then there is our our query of course you could use coordinates if you want to but let's say the search uh you know query is just easier and I just add this particular output at the end and this how uh the full URL is going to look like when you're trying to fetch it from Google and in the settings uh yeah nothing nothing really magical I think you could add it um to continue on error but now doesn't make any sense yeah you shouldn't add that uh for this one because if this is going to error out then we're just going to look for another uh search quer so now for this one let's just uh submit it let's see what we get back and we get back a lot of HTML data so basically what this is this is an HTML of exactly what is in here so for example if I would go um into let's say the console in into the elements we're basically extracting all of this HTML uh that we have here with that request and that's how then we can use to extract emails that are inside the websites of these so now let's go back uh so here I have a custom code uh basically node again there's going to be a link Down Below in the description where you can download the workflow it's going to be with this code totally for free what you have to do is basically just upload it and it's going to have this code but it just takes the first uh item let's say of the data it has this reject function I wrote it with um open AI uh then it just matches uh all of the domains that have https inside here and then it just Maps through it and generates a list of uh URLs so let's say test um let's run this and you can see we got a bunch of URLs we got htfs maps.google.com and so on so at first uh I can seem a lot of these are irrelevant or these are Google uh domains that we don't need and you're totally correct that's why we have the Gmail filter in here uh I have the URL in here as an expression and I filter out by having the the uh condition does not match reex and again this is the reex that reex that I'm using so it basically excludes all of the Google domains gstatic domain ggpht domains so on and so forth which you know are not needed because this is not exactly what we're scraping so for example if I run this you can see that out of 163 uh domains we passed only 18 then uh this simply removes duplicates so for example if we have duplicated emails it's going to be removed so now out of 18 in a single area so for example let's say if it was West Lake out of let's say 18 uh we got a few duplicated and we're left with 12 items now once we have those 12 items we go over with a loop through each uh URL yeah so for example um if I go here or maybe you know what let's run it let's see Oho this is executed Let's test the step un let's maybe run this step reference node is unexecuted Loop over items so this is not executed so but I execute this one yes come on okay so now I have to maybe just test the whole workflow let's see it execute in real time and then we can go over continue going over it so as you can see it's working very smoothly passes over and adds an item so let's uh get back into here so we have a loop um in the loop uh the input items are going to be the URLs so we're going to be adding URL one by one because remove duplicates returned like a list of URLs and Now We're looping over each one of them so first thing we're doing is we're using get method and we're just adding that Json basically the URL from Loop over items in Here and Now what it does is extracting um the data basically yeah the whole HTML from that uh URL and then in this one what we're doing is we're basically setting uh the URL again in an edit field and also I'm adding the data um in here but as you can see for example this one returned an error so that that means that here it's going to be filtered out so as you can see we're looking if json. dat exists and and only then it passes through the next step but for example if we go back into this um here it's important to actually on error to uh add continue that means that it's if it's going to uh error out at least for one you know website or domain it's going to continue running and it's not going to stop because for example if there would be at least one error then the whole workflow would completely stop and the rest of us of them wouldn't be know uh checked for potential emails inside of those URLs so yeah basically this uh goes over each of the URL let's see yeah so for example for this we got the data and then here we've set it that oh no we didn't get the data let's let's let's check if we have any that we data looks like this one got data so let's select yeah so for example for this one maybe this not the best example but yeah for example call climat care.com we got the data we got all of the HTML and then again this passes over back to the loop once all of them are finished then we have a filter and in the filter we again simply filter filtering out those that returned errors um that means probably the business haven't checked their you know um website they didn't maintain it it returns you know 403 error or something like that um and we just can't access it so it's done no point you know in in trying to extract email from the business damn I'm talking so much I hope that whoever is watching this can can understand what I'm trying to say because this is actually really really interesting but then uh we passed through another filter and this again I said that um it just discards those that didn't have any data and then we continue another loop where we use another custom code node and again on this one on error it's going to continue and it's going to always output data but it has this code that it uh extracts only the domain of the URL so for example if in air table the URL is https a.com so the domain would be only the airtable.com so this extracts only this part because in the next uh step we are trying to use that domain for email search um and I'm using this approach here because I don't want to extract random uh email that actually don't exist or was um extracted by mistake and because for example if then you're uh sending out mass emails um from your personal let's say email that can you know impact the open rate and then you know Google can uh think that you're just sending bunch of spam so i' would rather you know extract only the correct emails and again this code is going to be included so you can uh not worry but um over here uh what happens is that we specifically are looking for um emails that match that particular domain so I think over here let's check there was only one item that passed yeah so it was fifth maybe maybe there were more items but uh let's say this was what we extracted raed based on the domain so info Ohan a.com or whatever Ohana a um and then once those all finish up so for example we extracted around 20 emails or or at least 20 items in here uh then we check if the emails are with valid domains so that means what we're going to be doing is we're going to be checking if the email was found because if it was not found it's going to match this and it's not going to pass through the filter but then we use the split function that again turns all of the lists yeah this is not going to be much of them but this is going to be basically turn yeah just going to split a list into bunch of uh items then what I'm doing is that I'm just turning all of the emails into lowercase so from here particularly um this again is going to be done to remove duplicates because if uh I noticed there was a use case for example there was email hello something.com from the small letter and there was hello something.com from the capital letter those would be uh perceived as two different um emails so what it does in this Set uh Fields node it is basically turns into the lowercase so then we can later on remove the duplicates and then we have the last if node so again it uses uh basically the Json uh. email and reject function uh to check if what we're want to add into air table actually has an email pattern and if it does have an email pattern we simply just create a record in the air table uh with the you know location with the search term and with the email so for example here what we added was info@ Ohana a.com search term was at track in Crest View location so this is going to tag the Los Angeles uh yeah as you can see here Los Angeles uh is going to be added um and this is pretty much the whole flow uh I'm using this uh just to prevent adding empty uh values into air table so as you can see we have only filled in values but this in my eyes is actually quite interesting and very exciting flow I hope you were able to understand what I was talking about if you didn't let me know in the comments down below and I might do this step by step I just didn't think people would actually want to see me build this step by step because I'm not sure how long this would take probably like I don't know 40 50 minutes I don't know if anyone has time to watch that but if you enjoyed watching this video hit the like hit the subscribe to my channel again there's a link where you can download the whole workflow and also there's a link where you can reach out to me if you ever ever want to work uh together on a project or whatever you have in mind so yeah thanks for watching and see you in the next one