Five failures of a marketplace - lessons from running a babysitting platform
Between 2014 and 2020 my wife and I set up and ran an on-demand babysitter marketplace.
Think Uber for babysitters.
The platform ran on a mobile app and allowed parents to search for, select and pay for their choice of interviewed and security checked sitters.
COVID brought an unplanned pause to the business - but with six months of clear air to think we made the decision not to restart the business.
Running a marketplace is hard - and here are five lessons that will be useful if you are considering a marketplace element to your own business.
Marketplaces are margin constrained businesses
Marketplaces have more than one customer.
In our case we had the parents who paid for the care, and the sitters who provided it.
(In some cases there are three or more customers - Deliveroo has the eater, the restaurant and the rider).
The money the parent pays needs to provide enough for the sitter to feel it is worth their while compared to the other options to earn money, as well as providing enough for us to develop and manage the service, and make a profit while we do it.
To use simple round numbers - we charged a parent £10 an hour, the sitter was paid £8 an hour and we made £2 an hour. An average booking was 4.5 hours long giving us an average margin of £9 per booking.
The challenge comes when parents start requesting non-standard services - a pick up from school, a last minute booking or an overnight stay.
Now the amount a sitter rightfully should expect to drop everything, drive 15km to a school and pick up a child, or to stay overnight so parents could have a night away starts to become larger than what a parent is prepared to pay.
The parent’s anchor point is throwing £20 cash at the neighbour’s teenage daughter - a far cry from the hourly rate plus extras that the platform would require.
Non-homogenous supply drives against network effects
Homogenous supply sounds complicated, but it means that each supplier on the marketplace is offering the same or similar service.
Consider Uber - you really don’t care which driver is taking you. They drive you from A to B and you get out.
Even with Airbnb - you really don’t worry about your host - it is the building and the location you are interested in.
Childcare providers are not homogenous.
Once your children trust someone and like playing with them, then even though the parents found them through the platform they will disregard the other sitters available.
Parents would run a search:
“You have sitters available” we would say on seeing they hadn’t chosen.
“I’m waiting to see if Emily accepts”
“Emily is already booked that night” we would say
“Oh alright, we’ll change the night we go out”
This was frustrating to see as it devalued the entire point of the platform. Parents trusted it, but they wanted a specific person.
We’d often then get a text: “Emily can do Tuesday, can you book us in for then please?”
Local network effects did not drive regional network effects
Within our local area (which stretched across a couple of counties in the UK) word spread amongst the two segments - parents talked to other parents at the school gate about the fancy new app, and sitters talked to each other at university about this neat way to earn money quickly.
But as we tried to expand into a larger area, and specifically into larger cities it was as if we were starting from day zero again every time.
There was no word of mouth, and the fact we had hundreds of sitters available just 15 miles away meant nothing - they would not be suitable for parents in the new region.
With Uber and Airbnb - their demand side (riders and travellers) by their very nature travelled a lot.
Both platforms could see a lot of people downloading and trying to use the app in a new location and this meant Uber or Airbnb could invest in growing supply in that region - they had the chicken, they just needed to get the egg.
Parents don’t move around like that - each new market was a new chicken and egg problem.
People are unpredictable
When you book an Uber to go to Hanover Street it is highly likely you will go to Hanover Street. When you book an Airbnb for three nights it is highly likely you will be there for three nights.
With babysitting when you book for six hours it is actually very unlikely you will be out for exactly six hours.
You’ll either come back early (because the party or event wasn’t great) or you’ll come back late because it was amazing.
Come back early and your babysitter is frustrated because they thought they were going to earn £60 but only earned £30,
Come back late (or in some cases not at all) and the sitter is stressing at 4am because they have a university lecture or other babysitting booking at 9am in the morning.
On the sitter side you would be amazed how many grannies or pets fell ill on Saturday afternoons. “Sorry I am not going to be able to make the booking later”.
Both of these scenarios put responsibility on us to fix the problem.
We implemented a number of platform features to try and control both of these:
Cancellation fees
Emergency on call sitters (that we would pay for)
Early return minimums
Late return fees
Rating systems
But humans at 3am at a great party are unpredictable and we found money was not an effective lever.
For six years we were up until 3 or 4am on a Saturday night fixing human issues.
The biggest spenders required the most support
As parents learned that they could hit a button and sort their childcare, some started to rely on it heavily.
Evenings out, school pickups, homework support, overnight stays - even going on holiday with the family.
On the face of it this was positive as our average spend per parent was increasing.
However these parents started to move off the app to communicate with us directly to arrange these more custom requirements - pick up from this school, then a pick up from a second school, then home for some homework support, cook this meal….
The highest volume users moved into more of a consulting relationship.
When we tried to push them back to the app, this resulted in them agreeing these bookings with the sitter directly and then letting us know what to book in the platform.
Regularly we would spend a lot of time fixing a complicated multi-date booking only for the parent to text us “Actually its OK, my mum’s going to do it”.
Very frustrating.
Suggestions for building your own marketplace
Many companies add a marketplace element to their business.
Most professional services firms are in some ways a marketplace - hiring a bench of consultants to match to their customer’s requirements.
True marketplaces require as close to homogenous supply as possible - think how you can standardise the service from each supplier.
Focus on either very high volume low transaction cost (Uber) or lower volume high transaction cost (AirBnB) - we were stuck in the middle.
Build additional platform services that add value to the core service. These help differentiate and earn your margin.
Ensure that the more your customers (demand and supply) use the marketplace the better the experience gets to reduce the risk of de-platforming (going direct).
Further reading - this set of articles from Lenny Rachitsky is an excellent grounding on kickstarting and scaling a marketplace.
Get started
Whenever you are ready, there are three ways that I can help you accelerate your revenue.
Buyer Experience Audit - I’ll impersonate a buyer researching your segment and company and let you know what I find. Ideal for planning your Revenue Operations strategy.
Business Model Design Workshops - I’ll work with you and your team to design or refine a business model for a new or existing product.
RevOps Impact Playbooks - I’ll help you implement one or more tactical processes across your revenue teams - content, referrals, testimonials, adoption and more.