Take these 3 products: Ridehailing, Food Delivery, and Groceries.

As the PM who is the owner of the product funnel, you will have an interest in MAU -> MTU. You want to know how many people enter your funnel with the intention of doing their job, and how many end up getting their job done.

But MAU -> MTU is easy to game.

 If a person enters your funnel 30 times in a month but only completes once, you still have 100% conversion.



So you need to focus on conversion in a smaller duration. Let’s look at at a session level. A typical session is 15 min long.

Session conversion is a hard thing to game. Your funnel can’t have a high leakage. Giving a one time high value voucher to get the user to convert one time to meet your MAU -> MTU goal is not going to move the needle when it comes to session conversion.

And as the title of the post says: all conversions are not equal.

 The strength of the user’s intent is important. Use case is important. Product type matters.



An office commuter looking to book your ride will have high intent. They don’t have many alternatives.

While intent for someone ordering snacks in the evening might not be as high.

The alternative is to just prepare Maggi at home.

At the weekend, however, the same ridehailing user’s intensity of intent changes. Instead of the office, it could be a trip to the shopping centre. They might check the price. They might see a high surge, get off and try again later. If the distance is greater, they might even want to check prices on multiple ride-hailing apps.

If it’s the weekend and rush hour, driver supply could be affected. They may drop off on the fare estimation screen when they see the price and ETA. There may also be no voucher (demand-side subsidy) if supply is low and demand is high.


Take the example of food delivery.

It is evening. Your user would like to order tea. But only tea might not raise the AOV above the minimum point below which free delivery does not kick in. So the user does a search for snacks to order. Spends some time on the restaurant’s page. Perhaps they can’t find what they are looking for. They might go over to their flat mates and ask them if they would like to order something. They might minimise the app. Get on Twitter. Their session might end.

Now there are 3 possibilities:

  1. They come back and order.
  2. They might delete the food item, find another restaurant that serves both the snack they want and tea, and place their order.
  3. They don’t open the app again that day, open it the next day, and place the order for the items in their cart.


Session conversion is 50% in all these cases. It took 2 sessions for the user to get their job done. Conversion happened on the 2nd attempt.

The Home -> Restaurant -> Checkout -> Order -> Delivered funnel is messed up because different events are triggered for all 3 cases. Home doesn’t even fire for the 2nd session in the first case. So if you look at the funnel conversion as the absolute truth as a PM, you’re not going to uncover these nuances.

If the user orders at the end of the day, you are still helping the user to complete their JTBD. But it took them 40 minutes to complete instead of 15. Do you consider that you’ve failed as a PM? Or you say that for the JTBD of ordering evening snacks, your product was successful, so a further obsession with the session won’t lead to a better result, at least in the first case. What about the second case? If someone is back and they have items in their shopping cart, but they are still on the fence, maybe you can do some recommendations for other restaurants? Maybe.

For groceries, the person adds items, comes to the checkout and notices that the delivery charge as a % of the basket AOV is very high and drops off. The next day they come back, add more items and finally order. They have completed their task of ordering groceries. But it took 2 sessions, 2 days. The JTBD has been completed. Is this a failure?



It is raining in the evening. The user is trying to get home from the office. They order a cab. It is cancelled three times. On the 4th attempt they get their ride. The booking to job completion rate is 25 per cent. Your CEO is in a panic, because normally in Bengaluru this metric would be 60% on a non rainy day. You are asked to head a task force for the improvement of this metric.

 Now there are 2 things in a 2-sided marketplace:

  1. Demand shaping.
  2. Supply shaping.

You can shape demand by giving vouchers. Or through a change in user behaviour patterns. Maybe you send a notification to the user that there will be rain soon and they should book their ride. Perhaps to reflect the lower supply you have during rainy periods, you add a rain platform fee to discourage demand.

 You can manage the supply by

  1. Geting new supply in (difficult).
  2. Supply positioning (moving supply to high demand areas by nudging supply).
  3. Offering one-off flash incentives to supply.
  4. Collecting the rain fee and passing this fee on to supply to incentivise them.
  5. Implement more aggressive surge pricing (again, to increase the chances of supply getting picked up).

You can do all this. But what if you looked at the data and realised: most users stick to the platform and keep trying until they get a ride. They get a ride by the 3rd or 4th time anyway. And the cost of new supply and more supply initiatives is far too high to justify trying to improve the Booking to Job completion rate.

 A data-driven PM, hell-bent on hitting their metrics, might use all the tools at their disposal, including more budget, to solve this problem. During ZIRP, this was the standard approach. And you had to do it. You have to justify why this metric, that is your responsibility, is down.

A good PM is going to do more than that. They are going to look at how many orders are actually lost when it rains. How many people are actually going to drop out and never come back. What if you find you lose only 5% of the orders because even if conversion is low people just order a couple of times more and end up getting their ride. And the cost of saving that 5% of the orders is far higher than the GT that you are going to make from those orders.

What I am getting at is that the conversion rate is a guard rail metric. It is a lagging indicator of the value that your product is delivering to customers. And not all sessions are the same. And you need to dig a little deeper.