跳到正文
原文
AdExchanger· <img src="https://www.adexchanger.com/wp-content/uploads/2026/07/34-120x120.png" alt="Photo of Allison Schiff" width="60" height="60" class="ai-author-image ai-author-image--circle ai-author-image--inline rounded-circle shadow-offset" loading="lazy"> Allison Schiff·· 5 天前AI 评分37

SWYM.ai 创始人谈用 AI 供给塑形避开程序化广告低质流量

Stop Buying Lemons, With SWYM.ai

文章要点
  • 1SWYM.ai 创始人在播客访谈中称 SSP 与 DSP 缺乏反馈回路,买方的效果数据无法回传给供给侧
  • 2Patel 估算头部 DSP 为控制云成本只看到约 40% 可用供应,较小 DSP 约 20% 到 25%
  • 3SWYM 的方案是以容器化方式嵌入 SSP 系统内的 AI 供给塑形,逐条判断竞价请求能否带来结果
  • 4播客提到 SWYM 曾帮 Georgia-Pacific 把 SSP 名单缩减 80% 而未损失触达(公司口径)
决策要点

程序化广告的浪费根源在 SSP 与 DSP 之间没有效果反馈通道,这家公司的思路是把流量筛选前移到竞价发生之前。

  • 影响:广告主的预算里有相当部分在买前无法分辨好坏;DSP 只看得到两到四成供应,采买决策建立在残缺信息上;精简 SSP 名单有望在不损失触达的前提下减少浪费。
  • 参考:可考虑复盘现有 SSP 名单,识别重复或低效的供应路径;值得在采购投放工具时要求对方说明效果判定口径,而不只看曝光和触达;先测算低质流量占预算的比例。
  • 依据:全部观点来自创始人 Ravi Patel 的播客访谈,DSP 覆盖率 40% 与 20% 到 25% 为其个人估算,Georgia-Pacific 缩减 80% 案例属公司自述,暂无独立验证。
正文

Wednesday, November 11th, 2026 – 1:00 am

SHARE:

Ravi Patel, Founder & CEO, SWYM.ai

A podcast interview with Ravi Patel CEO & Co-Founder

Running a programmatic campaign is like walking into a grocery store with a budget you have to spend before the place closes, and you have no way to tell the lemons from the cherries until you’ve already paid.

Oh, and if you end up buying a lot of lemons, the store automatically assumes you love ’em and keeps stocking more.

That’s a favored metaphor of Ravi Patel, CEO and co-founder of SWYM.ai, a startup that helps ad buyers make smarter decisions before spending their money. (In fact, “SWYM” is short for “Stop Wasting Your Money.”)

But avoiding low-value inventory (spotting lemons) and finding the high-quality stuff (pinpointing the cherries) is both harder than it sounds and harder than it should be.

The main problem is that SSPs and DSPs “talk at each other and not to each other,” Patel says, on this week’s episode of AdExchanger Talks. There’s no feedback loop between them, so all an SSP sees is which bid requests a buyer won, not whether any of them helped that buyer achieve its goals.

On top of that, DSPs throttle traffic to keep cloud costs down, so they only have eyes on a slice of what’s available. Patel estimates that the largest DSPs only see roughly 40% of available supply, while for smaller DSPs it’s more like 20% or 25%.

And all the while, SSPs just keep sending more of the same. “If you bought a lemon yesterday,” Patel says, “you’re getting a lemon today.”

That’s why the filtering has to happen earlier, he says.

SWYM’s approach involves AI-powered supply shaping, which it delivers through containerization, a fancy way of saying its software lives within an SSP’s own systems rather than sitting outside of them.

It might sound like curation, but Patel is quick to draw a distinction. Curation has historically meant packaging an audience, attaching a deal ID and adding a markup. Supply shaping, he says, judges each bid request on its own and only based on whether it’s likely to lead to an outcome.

“I don’t care that user 12345 lives in New York City and loves to snowboard,” Patel says. “I just care whether that potential bid request can lead this client to a sale.”

Also in this episode: The rise and fall of Patel’s previous startup Ad/Fin, how SWYM helped Georgia-Pacific cut its SSP roster by 80% without losing reach, why he thinks AI-generated sites aren’t necessarily bad if real humans are visiting them and whether the open web is really dying. (Hey, they said the same thing about linear, and it’s still here!)

For more articles featuring Ravi Patel, click here.

来源:AdExchanger · adexchanger.com