Welcome to the easiest way to create onboarding and activation experiences.
Say hi to BentoAI →
Resources /

How we fight spam at Bento

Emily Wang
5 min read

Bento is embedded into our customers' products. Done well, an end user can't tell where the host app ends and Bento begins. So we know there's a high bar to looking and feeling integrated and not stick out or feel spammy.

We knew this from the earliest days of building Bento because we had experienced the spam of closing one modal only to have another pop in your face. We’ve personally faced the bombardment of clicking around an app only to have an 8 step product tour forced on us in each page. 

But the tricky part is that no app intentionally spams their users. Everyone’s intentions are to be helpful, and to be appropriately disruptive.

The real challenge we faced was: how do we help our Bento users be successful in not spamming their users?

Hindsight being 20/20, we actually tackled this in 4 phases (so far):

  1. Design in friction
  2. Promote form factors that are less spammy
  3. Help stakeholders easily see the spam
  4. Expose information to help users proactively design less spammy experiences

Phase 1: Design friction

From day 1, Bento’s onboarding guides were “sequential”, meaning that a user who qualified for 3 onboarding guides would only get guide 2 after completing guide 1. 

When launching a new onboarding guide, authors have to stack rank it against other onboarding guides.

Call it “progressive disclosure” if you want, but our basic thinking was: if your end user doesn’t even want to complete the first guide’s tasks, what makes you think they’ll do more advanced tasks?

End-users see success states after completing a guide, before their next guide is unlocked.

Then, as we introduced elements that could exist in parallel, we added two more points of friction:

  1. Product packaging: Our core pricing metric is the number of guides launched. We made the decision to make Each tooltip in Bento count as a guide. Egregious? Maybe. But it also makes it a lot more expensive to build a 8 step tour and forces more thoughtfulness on where pop-ups really need to occur
  2. Feature announcement modals are throttled per user session. Even if a user is eligible for 5 feature announcements, they'll only get 1 per session.

These experiences saved us from some of the worst offenses, but as we offered more embedded experiences and more form factors, users could still be eligible for many onboarding and everboarding triggers per session.

Phase 2: Offer less spammy form factors

No new user clicks into a bunch of empty states and says: “oh! There’s such helpful content in these empty states – that’s so spammy!”. 

Nor do users generally hate on info-tooltips that offer details on hover.

Spam is created when users are interrupted, disrupted, and when the actions they’re trying to take are blocked by an overlay.

Embedded contextual guides live in the page that is loaded.

To empower our customers to offer relevant calls to action and guidance without spam, we released:

  1. Embedded contextual guides. These inline banners, cards, and carousels live in a page and replace the need to build an empty state. As such, they don’t disrupt and can be revisited.
  2. Visual tags. Small UI elements that open a guide on hover in contrast to disruptive or blocking tooltips.

Phase 3: Help stakeholders see the spam

Many of our customers would tell us that their users experienced spam, but not intuit how their own decisions contributed to this. Frequently, it comes from different teams each promoting their own feature and unknowingly creating a 

So we looked at ways we could make these spam-moments more visible to our customers.

For example, we realized that in order to see “targeting logic” in Bento, you had to click into each guide, then tab over to the targeting tab, then scroll down to see the logic. 

Enter in any user's email address and you'll instantly see the guides they qualify for.

To make this easier, we added a view where you could select a saved audience or individual user and see their flight path. Over time, we plan on adding more real-time details like which guides a user has seen and what could be next on their path.

Next: Help users proactively make the right choices

The phase we’re currently in is all about creating awareness and encouraging smarter decisions during the guide design phase. For example, most modals are used for feature announcements. But for brand new users, every feature is new! 

Therefore when creating a modal, we could automatically set the targeting to only target users created > 3 days ago. Our customers can easily change this, but having this as a default state has prompted more thoughtfulness around targeting logic.


Ultimately the best fight against spam is to increase the relevance of content, nudges and triggers. Yet by building Bento in a way that considers the interactions across our guides and components, we aim to be the easiest platform to build without having to over-engineer a process to wrangle possible user-overload.

BentoAI - Auto-create guides from existing articles and recordings | Product Hunt