[$500][Due for payment 2025-08-01] [Due for payment 2025-07-30] Implement Ex50 - the feature modal into the Onboarding flow

[anmurali]

We ran an A/B test in the product by showing a modal with a clear set of features during onboarding.

Hypothesis: This would inspire more confidence that we offer what they need — and pull up selecting and engaging with workspace settings earlier

Success metric: More engagement, either in product, with sales, or via chat with the same or incremental levels of onboarding completion

The ultimate goal is increased conversion, but that takes 2+ months to measure. This test looked at early signals of engagement.

Onboarding Drop-Off by Group

Group	Start Onboarding	Finish Onboarding	Drop-off Rate	Difference
Control	709	599	15.51%
Variant	651	553	15.05%	🟢 -3.0%

Engagement by Group

• 20-30% of the variant group enabled a non default feature, with invoicing the most popular, followed by rules, distance and tags.

• 5% of the variant group disabled a default feature.

Group	Engaged Users	Total Users	Proportion
Variant	305	516	59.1%.
Control	288	602	47.8%.

✅ The difference in proportions is statistically significant at the 99.95% confidence level (p = 0.0005), suggesting that contacts in the Variant group were significantly more likely to have associated deals.

We also had statistical significance on Corporate Upgrades
Corporate Upgrade Rate by Group

Group	# with Upgrade	Total Contacts	Proportion
Variant	119	516	23.1%
Control	33	602	5.5%

✅ The difference in upgrade rates is statistically significant at the 99.9999% confidence level (p < 0.000001), indicating a very strong effect of the variant on corporate upgrade behavior.

While the billing card added data is not yet statistically significant, it is slightly lower on variant.
That said, we don't think this should be a blocker because:

(a) that's not what this test directly looked to address
(b) there's no argument that increased engagement causes lower conversion, and therefore we should aim for less engagement

For transparency here is the data so far:
Add Billing Card Rate by Group

Group	# with Billing Card	Total Contacts	Proportion
Variant	39	516	7.6%
Control	51	602	8.5%

❌ The difference is not statistically significant (confidence level: 73.6%). That means this slight edge for Control could easily be due to chance.

Overall conclusion: There is no incremental drop off at this step, there is incremental engagement in the variant, and having access to better data around a user's preferences on features is accretive as we fine tune our email campaigns to them based on this. So let's 🟢

Issue Owner

Current Issue Owner: @dylanexpensify

[dylanexpensify]

Taking this across the line:

• Post for C+ volunteer. Done here
• Share email w/Victor for BitBucket view-only access. Done here
• Wait till access is given
• Work with C+ to understand what the workload looks like for bringing this code into our repo and updating it to our standards/style

[dukenv0307]

@dylanexpensify The PR is ready for review

[dylanexpensify]

Beautiful @dukenv0307!!