September 15th. A delivery app in Riyadh calls.
"We're losing 60% of users between adding items and completing checkout. Our conversion rate is killing us."
They showed me their order flow.
User wants: Burger and fries.
What they have to do: Navigate 11 screens, switch keyboards 4 times, scroll through dozens of options.
Average time: 2 minutes 18 seconds.
Completion rate: 40%.
The 11-Step Conversion Killer
Here's what users went through to order food:
1. Open app, tap search 2. Type "burger" (switch to English keyboard) 3. Results page: 40 burger options 4. Scroll, find "Classic Burger" 5. Tap it 6. Product page opens 7. Tap "Add to Cart" 8. Go back to search 9. Type "fries" 10. Repeat steps 3-7 11. Navigate to cart 12. Tap "Checkout"
Most users quit before step 11.
Where They Lost Users
We analyzed 3,400 order attempts over 2 weeks.
Drop-off by step:
Total completion rate: 40%
The problem wasn't the food. It wasn't pricing. It was friction.
11 steps. 4 keyboard switches. 3 navigation loops.
Users were exhausted before they hit checkout.
What Users Said
We interviewed 60 users who abandoned orders.
*"I just wanted a burger. Why do I need to see 40 options?"*
*"I added the burger, went back, forgot what I was ordering."*
*"By the time I got to checkout I wasn't hungry anymore."*
*"The Arabic keyboard is slow. Switching to English for searches is annoying."*
Every user wanted the same thing: Fast. Simple. Done.
The Solution: Voice-to-Actions
October 3rd. They integrated Voqal.
Same order: Burger and fries.
New flow:
User says: "أبي برجر وبطاطس" ("I want a burger and fries")
Voqal returns:
{
"action": "add_to_cart",
"items": [
{"name": "Classic Burger", "quantity": 1},
{"name": "French Fries", "quantity": 1}
]
}App shows: "Classic Burger + Fries - 45 SAR. Confirm?"
User taps: "Yes"
Done. In cart. Ready to checkout.
Total steps: 2
Total time: 23 seconds
No search. No scrolling. No navigation loops.
Command → Confirmation → Cart.
Week 1 Results
They deployed to 10% of users in Riyadh.
| Metric | Typing Flow | Voice Flow |
|---|---|---|
| Average time to order | 2m 18s | 23s |
| Completion rate | 40% | 67% |
| Users who ordered again within 3 days | 34% | 58% |
Same users. Same food. Different flow.
Completion jumped from 40% to 67% because they removed 9 steps.
Why Not 100%?
Some users still dropped off. We found 3 patterns:
1. Ambiguous Orders (18% of drop-offs)
User: "أبي بيتزا" ("I want pizza")
App showed 8 pizza options (still friction).
Fix: Default to most popular. Let user say "No, the spicy one" to change.
2. Modifications (9% of drop-offs)
User: "أبي برجر بدون بصل" ("Burger without onions")
App added burger WITH onions (didn't parse "without").
Fix: Parse modifiers. "بدون" = remove ingredient.
3. Menu Gaps (6% of drop-offs)
User said: "شاورما"
Voqal heard: "شاورما" (correct)
But mapped to: "Shawarma Meal" (comes with sides)
User wanted: Just the sandwich.
Fix: Show what was selected. Let user adjust before confirm.
They fixed all 3 in week 2. Completion hit 84%.
The Numbers 8 Weeks Later
| Metric | Result |
|---|---|
| Voice adoption (all users) | 38% |
| Voice orders as % of total | 29% |
| Voice order completion rate | 84% |
| Typed order completion rate | 41% (unchanged) |
| Average voice order time | 19 seconds |
| Average typed order time | 2m 12s |
Voice orders completed 2x more often than typed orders.
Not because voice is magic. Because they removed the friction.
What Actually Worked
They didn't optimize recognition accuracy.
They didn't A/B test button colors.
They didn't add more menu options.
They removed 9 steps.
Before: Search → Results → Select → Add → Search → Results → Select → Add → Cart → Checkout
After: Command → Confirm → Checkout
Users don't want features. They want speed.
The Takeaway
If you're losing users in your flow:
1. Count the steps. If it's more than 3, you're bleeding users.
2. Measure drop-off at every step. Where do users quit? Fix that step.
3. Voice-to-actions removes navigation. Don't make users search. Execute the action.
4. Test with real orders. Not scripted demos. Real users ordering real food.
5. Completion rate > Adoption rate. 50% try your feature but 10% complete? You built friction.
They lost 60% of users in an 11-step flow.
They cut it to 2 steps with voice-to-actions.
Completion jumped from 40% to 84%.
That's not a voice feature. That's conversion optimization.