Demo • Metadata — Agentic Paid Media in Action

1722
05:00:52.430 –> 05:00:56.649
Julia Nimchinski: Agentic Faid Media! So excited to see this in action.

1723
05:00:57.530 –> 05:01:09.830
Lisa Sharapata: Yeah, alright, so I’m gonna actually just start with a couple slides, because what I have discovered in talking to a lot of CMOs is that

1724
05:01:10.150 –> 05:01:12.520
Lisa Sharapata: There’s still a lot of questions about…

1725
05:01:12.800 –> 05:01:29.709
Lisa Sharapata: Like, the conversation that we just had isn’t the normal conversation that I’m having, even at Inbound a couple weeks ago with some of, like, the brightest, most leading-edge marketers I know. So, that… with that, I’m just gonna pull up

1726
05:01:29.930 –> 05:01:32.030
Lisa Sharapata: A couple slides,

1727
05:01:32.980 –> 05:01:41.340
Lisa Sharapata: this concept of prompts to pipeline, I’ve actually seen a couple of companies talking about it, and…

1728
05:01:41.830 –> 05:02:00.619
Lisa Sharapata: So I’m gonna kind of unpack that, and then I’ll show the actual use case with metadata. So, first, I want to just level set on this MCP server, because if you don’t know what it is, pretty quickly, the conversation, it’s like, what are you even talking about? And I just joke around because

1729
05:02:00.780 –> 05:02:02.410
Lisa Sharapata: like, 5…

1730
05:02:02.450 –> 05:02:12.560
Lisa Sharapata: 6 months ago, if you asked ChatGPT what MCP was, when I was asking it, it kept saying Model, or Marketing Content Portal.

1731
05:02:12.560 –> 05:02:23.709
Lisa Sharapata: So I didn’t even know, right? It’s such a new concept, but I think of it like a bridge, and it stands for Model Context Protocol, but it, like, that doesn’t even describe it for me.

1732
05:02:23.710 –> 05:02:29.550
Lisa Sharapata: So, I made a little picture, because I am very visual, so…

1733
05:02:29.640 –> 05:02:40.980
Lisa Sharapata: Take this idea of where we just were talking about all these agents, and agents can do a lot of different things, and you can create, like, handoffs and workflows from agent to agent, but

1734
05:02:41.270 –> 05:02:49.929
Lisa Sharapata: when you put them all together, and you build out an MCP server connector, and then connect that to

1735
05:02:49.960 –> 05:03:03.229
Lisa Sharapata: another platform, in this case Metadata, but, like, HubSpot has an MCP server connector, so whatever that larger platform is, then you’ve got this bridge between

1736
05:03:03.290 –> 05:03:09.600
Lisa Sharapata: what all of your agents can do, what the platform can do, and what the LLM

1737
05:03:09.600 –> 05:03:25.299
Lisa Sharapata: that you’re in can do. So, just, I mean, imagine how powerful that is, right? So, in the case of metadata, metadata’s doing digital advertising in all the different advertising channels.

1738
05:03:25.530 –> 05:03:30.010
Lisa Sharapata: And right now, you build all that inside of metadata.

1739
05:03:30.470 –> 05:03:35.410
Lisa Sharapata: And you can… you get your reports, and you can analyze stuff, but,

1740
05:03:36.040 –> 05:03:47.429
Lisa Sharapata: you know, not at the power of what an LLM can do. So, now I can analyze information across my CRM, my metadata platform.

1741
05:03:47.590 –> 05:04:00.449
Lisa Sharapata: everything that’s going on outside of that, like, what are my competitors doing, you know, in the ether? And I can even bring in, like, ask for best practices in the market.

1742
05:04:00.570 –> 05:04:09.969
Lisa Sharapata: To analyze my campaigns and what’s working, and get recommendations on how to build a new campaign.

1743
05:04:10.110 –> 05:04:25.650
Lisa Sharapata: what audiences should I use? What channels? What… what language, going to… getting into creating, creating those assets? What copy should I use? If my competitors are saying this, and I’m saying that, and this is what’s performing best for both.

1744
05:04:25.650 –> 05:04:31.000
Lisa Sharapata: like, how do I marry that up and then create a new campaign? So…

1745
05:04:31.920 –> 05:04:45.499
Lisa Sharapata: just, I mean, the ability to do that in 5-10 minutes is, like, a miracle in itself. And then you can actually deploy it from the MCP server, and then back to, like, agents.

1746
05:04:46.110 –> 05:04:52.880
Lisa Sharapata: can optimize for you, or you can give parameters and tell those agents, like.

1747
05:04:53.290 –> 05:04:59.539
Lisa Sharapata: what to do next, so that they’re 24-7, like, optimizing on your behalf. So…

1748
05:04:59.900 –> 05:05:08.489
Lisa Sharapata: With that, I’m gonna switch over to a different screen. Actually, I can’t have this pulled up here, so…

1749
05:05:08.740 –> 05:05:18.570
Lisa Sharapata: Inside the metadata platform, There’s all these different experiments going on, all these different campaigns running, you know, there’s…

1750
05:05:19.040 –> 05:05:31.229
Lisa Sharapata: Budgets and audiences and reports and all this different information that has been built out and has been running from the metadata instance.

1751
05:05:31.300 –> 05:05:46.330
Lisa Sharapata: So today, metadata already within that, you can turn on, you know, you can decide if you want auto-bid or manual bid, or you can run, a bid agent example. And then…

1752
05:05:46.550 –> 05:06:05.059
Lisa Sharapata: you know, there’s also these experiments happening behind the scenes, so you’re putting a threshold, you’re saying, this is the outcome I’m trying to get, here’s the max number of spend I want, but here’s, you know, 3 different campaigns, and 3 different offers, and 3 different,

1753
05:06:05.410 –> 05:06:19.919
Lisa Sharapata: creatives, so multiply that, and I’ve got, say, 16, 12, 16 different experiments I’m running, and it’s being optimized. So that’s within the platform. But then, when I go…

1754
05:06:20.420 –> 05:06:21.640
Lisa Sharapata: to…

1755
05:06:22.470 –> 05:06:36.350
Lisa Sharapata: connect this to the MCP server, and I’m just gonna show how it’s done. So in this case, within metadata, you get this access token here, and you’re just cutting and pasting it, and I’m going to put it into Claude.

1756
05:06:36.400 –> 05:06:48.159
Lisa Sharapata: It’s very similar with HubSpot and some of the other ones I’ve seen. So, I mean, it’s really as simple as, like, you get this code, and now you’re connecting it to your LLM.

1757
05:06:48.340 –> 05:06:54.090
Lisa Sharapata: So, I’m gonna stop sharing here, and I’m just gonna pull up… Claude.

1758
05:06:55.390 –> 05:06:56.480
Lisa Sharapata: So…

1759
05:07:02.580 –> 05:07:14.530
Lisa Sharapata: So, inside Claude, which I’ll say is running better than ChatGPT today, but I know ChatGPT is, you know, building this out too.

1760
05:07:14.640 –> 05:07:15.950
Lisa Sharapata: I…

1761
05:07:16.540 –> 05:07:26.459
Lisa Sharapata: have my metadata MCP server connected, and then these are all the different agents that are built out, and…

1762
05:07:26.650 –> 05:07:37.810
Lisa Sharapata: turned on within it. So I have control, like, I can turn on and off any agent I want. I could say, oh, I don’t want to be able to launch a campaign from here, that’s

1763
05:07:38.030 –> 05:07:48.319
Lisa Sharapata: I don’t know, out of my comfort zone, I’m gonna turn that piece off. I’m gonna turn it back on, but I don’t want it touching my budget, right? So I could turn that…

1764
05:07:48.830 –> 05:07:51.840
Lisa Sharapata: piece off, but also when I’m looking at

1765
05:07:51.960 –> 05:07:56.670
Lisa Sharapata: like, budget allocation, I want that on for analysis.

1766
05:07:57.140 –> 05:08:02.259
Lisa Sharapata: So, again, you can see there’s all these different,

1767
05:08:03.020 –> 05:08:06.499
Lisa Sharapata: agents turned on here, and I’ll just also show

1768
05:08:06.640 –> 05:08:10.070
Lisa Sharapata: Like, to get that set up,

1769
05:08:11.110 –> 05:08:16.539
Lisa Sharapata: to get it turned on, you just go into Connectors, and then

1770
05:08:16.700 –> 05:08:29.740
Lisa Sharapata: This is a custom connector, so that code that I showed you in metadata, I’m just gonna add… I would add the custom connector, and then that’s how it got turned on there. And again, this is, like, all of the different agents that I just showed you.

1771
05:08:29.950 –> 05:08:38.009
Lisa Sharapata: And here you’ve got even another level of control, so I can say, allow it to do something unsupervised, or…

1772
05:08:38.110 –> 05:08:46.430
Lisa Sharapata: ask permission before you do it. So, in some of these cases, like, anything with my budget, I’m saying you need to ask permission.

1773
05:08:46.660 –> 05:08:57.459
Lisa Sharapata: So again, just to give you an idea of, like, the amount of control, and how this is built out, and so now I’ve got all these agents, and I’m connected to Metadata’s instance.

1774
05:08:57.980 –> 05:09:01.990
Lisa Sharapata: And then, I can come in and…

1775
05:09:02.480 –> 05:09:06.830
Lisa Sharapata: basically just start prompting, like I would with any…

1776
05:09:06.970 –> 05:09:14.339
Lisa Sharapata: other time that I’m prompting to get, questions answered. So, in this case, I just…

1777
05:09:14.490 –> 05:09:31.720
Lisa Sharapata: I started just to make sure, you know, everything was working before we started, so I just asked it, are you connected to the metadata MCP server? It answered me yes, and saying, you know, I have access to… and it just kind of breaks down into categories.

1778
05:09:32.030 –> 05:09:35.909
Lisa Sharapata: Some of these different agents that are built out, so that

1779
05:09:36.110 –> 05:09:46.770
Lisa Sharapata: It can tell me, like, I can help you analyze campaign performance, create new campaigns, optimizing existing ones, etc. So, then, I’m…

1780
05:09:47.060 –> 05:09:57.229
Lisa Sharapata: I’m gonna… now I’m gonna start prompting to try to create a new campaign. And I’m basically giving…

1781
05:09:57.570 –> 05:10:00.970
Lisa Sharapata: Claude, a time frame, I’m saying, okay.

1782
05:10:01.290 –> 05:10:09.289
Lisa Sharapata: Starting with today, you’re a marketing AI assistant with access to our MCP tools.

1783
05:10:09.480 –> 05:10:19.220
Lisa Sharapata: I want you to look at last month’s top performing LinkedIn campaigns based on this criteria. And again, there’s lots of criteria, right? So.

1784
05:10:19.250 –> 05:10:31.040
Lisa Sharapata: I’m gonna be really specific in this case, because otherwise it’s gonna try to crunch all that information, and I’m also giving it a time frame, last month. So, I put a start date.

1785
05:10:31.140 –> 05:10:34.650
Lisa Sharapata: Go into the last month, And then…

1786
05:10:34.860 –> 05:10:52.189
Lisa Sharapata: analyze this information specifically, and I’m also telling it to be specific to LinkedIn campaigns, so there’s lots of other channels that we’re running, but keep it to this one channel, and then make suggestions to improve

1787
05:10:52.780 –> 05:11:12.300
Lisa Sharapata: my past high-performing offers and ads, and then helped me create a new optimized offer, LinkedIn copy, a campaign strategy, and clearly label all that. And, by the way, also had competitor strategies in there. So.

1788
05:11:12.560 –> 05:11:15.839
Lisa Sharapata: No, this is then pulling from

1789
05:11:16.060 –> 05:11:32.010
Lisa Sharapata: the web, we have a competitor agent built that can go and scrape and see, like, here’s what the other campaigns are that are being run in your competitive set in LinkedIn, and so that now you’ve got all this different information.

1790
05:11:32.160 –> 05:11:39.919
Lisa Sharapata: to… look from. So… Claude is now going in and starting to analyze, and

1791
05:11:40.020 –> 05:11:46.889
Lisa Sharapata: You can see, you know, what it’s looking at. It’s going into the metadata instance.

1792
05:11:48.020 –> 05:11:50.679
Lisa Sharapata: Pulling out all this information.

1793
05:11:51.210 –> 05:11:59.079
Lisa Sharapata: It’s also looking at trends. So again, going into…

1794
05:11:59.500 –> 05:12:15.509
Lisa Sharapata: The internet, and… so funny even saying the word internet now, but looking at other articles and things that have been posted in 2025 about the best tips and tricks for optimizing your campaigns.

1795
05:12:15.660 –> 05:12:26.750
Lisa Sharapata: And it’s even looking at videos about it, and watching thought leadership content, and then diving more into my performance metrics.

1796
05:12:26.900 –> 05:12:28.960
Lisa Sharapata: And then it’s saying, okay.

1797
05:12:29.290 –> 05:12:35.320
Lisa Sharapata: Here’s what your top performing audience is, and looking at all of our different audiences, etc.

1798
05:12:35.600 –> 05:12:39.090
Lisa Sharapata: And, you know, it’s gonna keep going and going.

1799
05:12:39.300 –> 05:12:50.699
Lisa Sharapata: it’ll actually, like, give me links to offers and things. This would take me back into the platform, so I won’t click it right now. And then it says, okay.

1800
05:12:50.850 –> 05:12:55.730
Lisa Sharapata: The LinkedIn ad has been created successfully. Here’s your URL.

1801
05:12:55.890 –> 05:12:58.120
Lisa Sharapata: Now let me create a budget group.

1802
05:12:58.690 –> 05:13:01.040
Lisa Sharapata: And so it’s starting to build this out.

1803
05:13:01.220 –> 05:13:07.490
Lisa Sharapata: Let me create another variation of the copy, so you have options, and then it’s just giving me…

1804
05:13:07.640 –> 05:13:09.680
Lisa Sharapata: Here are 3 different…

1805
05:13:09.970 –> 05:13:18.800
Lisa Sharapata: Campaign offers with the optimized offer are actually two different creatives, and… and a summary for it.

1806
05:13:19.000 –> 05:13:33.469
Lisa Sharapata: And on top of that, it’s telling me the expected performance of the campaign, how much pipeline it will likely generate based on my past campaigns, and telling me what to do next.

1807
05:13:33.700 –> 05:13:46.230
Lisa Sharapata: So… Because with that, I’m gonna ask it to actually go and create the new campaign, and

1808
05:13:46.750 –> 05:13:53.330
Lisa Sharapata: I’m gonna have it, create it based on this audience.

1809
05:13:54.210 –> 05:13:57.560
Lisa Sharapata: So, I… I know that…

1810
05:13:57.800 –> 05:14:07.469
Lisa Sharapata: and it’s… it’s in here somewhere, but this audience is… is the one I want to use, and I just wanted to just go in and build this campaign up for me, so…

1811
05:14:07.890 –> 05:14:09.290
Lisa Sharapata: I’m gonna hit…

1812
05:14:11.310 –> 05:14:29.409
Lisa Sharapata: enter on that, and, you know, now it’s gonna be thinking for a little while, so this is why I already, like, typed this other stuff out. And then, again, it’s, like, working on this in the background, and so it’s building this, so I’m just gonna let it think for a minute.

1813
05:14:29.940 –> 05:14:35.270
Lisa Sharapata: And… I’m just gonna look back up here, too. It’s talking about…

1814
05:14:35.370 –> 05:14:43.090
Lisa Sharapata: how I should configure my budget groups, and I’m looking at the spend, and I’m also looking at, like, how much…

1815
05:14:43.160 –> 05:14:54.170
Lisa Sharapata: pipeline, it’s saying that I’ll probably generate with this, and so I’m thinking, you know, I have a little bit more to spend than what it was recommending.

1816
05:14:54.170 –> 05:15:13.030
Lisa Sharapata: So, I always, like, you know, I’m gonna, like, just test it out here, and this is my curveball, so I’m just gonna throw it in. I’m gonna say, you know what, I have $20,000 to spend, and, I need to generate as much pipeline as possible in the next 30 days, so can you make adjustments? So I’m just gonna type that, so…

1817
05:15:13.320 –> 05:15:14.510
Lisa Sharapata: have…

1818
05:16:04.350 –> 05:16:05.220
Lisa Sharapata: Okay.

1819
05:16:05.420 –> 05:16:11.540
Lisa Sharapata: So, just, again, letting it run, just so you can kind of see the power of this, and how…

1820
05:16:11.710 –> 05:16:26.289
Lisa Sharapata: this will allow you as, you know, to be strategic, to really think through things, to kind of work with your LLM, but still with all this power at your fingertips and, like, information at your fingertips.

1821
05:16:26.390 –> 05:16:37.710
Lisa Sharapata: to make informed decisions and build out campaigns and strategies. Again, I think we’ve been maybe 15 minutes here at the most, so…

1822
05:16:38.290 –> 05:16:44.659
Lisa Sharapata: That’s pretty cool. And then… I’m gonna ask it to create a new campaign

1823
05:16:45.540 –> 05:16:54.469
Lisa Sharapata: So, revised the campaign strategy. Now I’m gonna ask it to actually create the new campaign.

1824
05:16:55.020 –> 05:16:56.470
Lisa Sharapata: Based on…

1825
05:16:57.260 –> 05:17:15.119
Lisa Sharapata: this recommendation. And, you know, obviously, I would normally review it, make sure I agree with it, and, see if I had any other tweaks before I asked it to create the campaign. But for the sake of this, as soon as it’s done.

1826
05:17:15.370 –> 05:17:22.139
Lisa Sharapata: I’ll ask it to create the new one, and then we will deploy it, and I’ll show you how that works, and then we’ll be all set.

1827
05:17:24.150 –> 05:17:31.110
Lisa Sharapata: So, I’m just gonna let that through… so it’s actually… oh, it’s actually creating it. Now let me create the high converting.

1828
05:17:33.200 –> 05:17:40.250
Lisa Sharapata: Oh, yeah, it’s doing that, so I won’t do that. I’m just gonna ask it to… deploy it.

1829
05:17:51.260 –> 05:17:58.039
Lisa Sharapata: So it’s still thinking, also… Okay, that’s it.

1830
05:18:07.090 –> 05:18:08.679
Lisa Sharapata: It’s creating… oh.

1831
05:18:10.250 –> 05:18:20.169
Lisa Sharapata: So… I’m gonna have to show another screen and go back to the platform to show you this, but…

1832
05:18:20.280 –> 05:18:28.619
Lisa Sharapata: If I wanted to check it, if I wanted to see what this cam… what this creative looks like, and how it actually looks.

1833
05:18:28.730 –> 05:18:38.630
Lisa Sharapata: I would go back into the platform, so I have to… Open the link… And… share…

1834
05:18:38.980 –> 05:18:44.820
Lisa Sharapata: Wish you could see this all at once, but… So that link…

1835
05:18:45.290 –> 05:18:57.180
Lisa Sharapata: took me to this page, and you can see here that this new offer header that was described by Claude,

1836
05:18:57.660 –> 05:19:07.019
Lisa Sharapata: this copy, everything that was written, it’s all in here. The landing page it’s recommending that we go to is here.

1837
05:19:07.270 –> 05:19:08.180
Lisa Sharapata: Ow.

1838
05:19:08.400 –> 05:19:10.739
Lisa Sharapata: the CTA…

1839
05:19:10.960 –> 05:19:20.540
Lisa Sharapata: Again, this is based on what’s been working the best, all the analysis that’s been done. It’s all selected for me here, and…

1840
05:19:20.850 –> 05:19:26.889
Lisa Sharapata: everything is, you know, built. I can update the offer, or I can save it as is.

1841
05:19:27.770 –> 05:19:37.629
Lisa Sharapata: So, from there, I’m gonna go back to Claude, I’m gonna try publishing it, hopefully it will work, and that will be…

1842
05:19:38.010 –> 05:19:44.580
Lisa Sharapata: the end. So, I’m saying, okay, it looks great, I like these recommendations.

1843
05:19:44.920 –> 05:19:50.769
Lisa Sharapata: I want to generate this much pipeline, I’m going to hit publish right up here.

1844
05:19:51.620 –> 05:19:57.839
Lisa Sharapata: And I… it’s right here to publishing, copy the link.

1845
05:19:58.010 –> 05:20:09.560
Lisa Sharapata: And… it’s been published. Now, it’s LinkedIn, so it still takes the standard, like, 24 hours for everything to get optimized on LinkedIn’s side.

1846
05:20:09.560 –> 05:20:21.609
Lisa Sharapata: If I wanted to, I could go pause it or cancel it because of that, but otherwise, it’s done. And this is all working for me in the background, and I can go do something else.

1847
05:20:24.040 –> 05:20:29.540
Julia Nimchinski: That’s amazing, Lisa. Super futuristic, and I believe you are the first Martech.

1848
05:20:29.920 –> 05:20:37.250
Julia Nimchinski: advertising the MenGen solution, with integration of this sort, but correct me if I’m wrong.

1849
05:20:38.230 –> 05:20:58.109
Lisa Sharapata: Well, I’ll say, I know that, HubSpot has an MCP connector, and there’s some things you can do with it, but not… I mean, I haven’t seen anything like this. This is by far the coolest use case, and I feel like the most powerful, that I have

1850
05:20:58.340 –> 05:21:02.290
Lisa Sharapata: seen anywhere in… I’ve been looking, so, yeah.

1851
05:21:02.770 –> 05:21:14.289
Julia Nimchinski: 100% blind. And for all of our community members who are watching, what’s the best next step to engage, try? Is there some freemium?

1852
05:21:14.570 –> 05:21:16.230
Lisa Sharapata: Yeah,

1853
05:21:16.240 –> 05:21:41.180
Lisa Sharapata: Well, I will put a link to the Metadata One landing page, and we’re… we are doing another webinar on it if you just want to see it again, and you don’t want to talk to somebody. So that’ll be on Thursday. So, I’ll drop a link to that in here. There’s a free trial for the bid agent piece of this, and that is just the piece that connects to your LinkedIn.

1854
05:21:41.180 –> 05:21:45.429
Lisa Sharapata: And optimizes your bidding for you, but we have a 14-day free trial for that.

1855
05:21:45.430 –> 05:21:51.250
Lisa Sharapata: So, I can drop a link to that in there, and if anybody has any questions or wants

1856
05:21:51.300 –> 05:21:57.279
Lisa Sharapata: to talk more with me, I will also drop my LinkedIn. It’s the best way to get ahold of me.