Originally published on LinkedIn Pulse by Ankur Gattani.

From having picked up CRM as a function out of the blue back in Jan 2013 to having seen it evolve closely over the last 3 years, thought I’ll summarise my learning in a bunch of bullet points. Hope this is helpful to marketers and start-uppers thinking their CRM / Retention strategy.

1. You have to look at the two CRM work areas “Email marketing and Push Notifications” and “Retention and CLV optimisation” distinctly, even if the same team is responsible for both.

2. Emails/Pushes by themselves won’t be sufficient to impact your cohorts (re-ordering customer base). The funnel drops sharply through Opt-in rates / Open rates / click rates/ conversion rates and doesn’t ultimately offer a huge uplift.

3. Going multi-channel (specially using On-site CRM) are crucial, if you really want to have your CRM efforts impact your customer retention. Best-in-class now is to orchestrate multi-channel campaigns across email / push / in-apps/ display re-targeting/ FB custom audience / On-site overlays / on-site banners / SMS – all talking the same language, dealing with the customer as an individual, instead of each channel giving a different message.

Add predictive modelling for customer lifetime value / risk of churn and you have a compelling CRM mechanism that can actually help retention efforts. Check tools like Optimove.com / Agilone.com.

3.1 This may need a bit of organisational collaboration to end the silos / turf issues between different marketing channel owners. Best you think about it early in your org’s life. You can’t apply the same CPO targets showing display ads to your most and least valuable customers.

3.2 The segments here are floating. The audience of ‘those who placed their first order yesterday’ will have to change everyday, while the campaign remains the same. No way you can manage dozens of these floating segments manually.

3.3 The campaign content is again dynamic – you could show the ‘top recommendations for you’ based on your first purchase – similar / complementary products or ‘new arrivals in the brand /category of user preference.’ or something else, depending on your product portfolio.

4. Automation is the only way to do meaningful customer communication. You can never send thousands of personalised messages otherwise. To understand and plan for automation, you need to really understand how data is flowing in your systems, which data is available at user level and what additional data can be available at user level to base your campaign on and how this data can translate into a ‘relevant’ customer campaign.

4.1 Segmenting too finely will leave you with several small fragments of audience without a clear idea on how to treat them differently on an ongoing basis. Think more in terms of increased relevance via personalised content.

4.2. The lowest hanging fruit is to automate is the new user on-boarding – right from his sign-up to his placing his first order – educating him about your site / getting him to place his first order. He shouldn’t immediately start getting your regular promotions that are sent to all users. Best if you use dynamic content for ‘promotion-of-the-day’ as a section in the on-boarding journey mail.

5. What a user did (saw / spent time on / searched / abandoned…) on the site yesterday is way more relevant for the company than what he bought a month ago to target communication to me. The more communication (share of volume) you can move towards ‘triggered/automated’, the better ROI you’ll get.

5.1 Really worth looking at Tripadvisor’s emails to see how browsing behavior can be used.

TripAdvisor’s Unfair Email Marketing Advantage

6. In slim margin businesses, acquiring a customer via a steep discount, will NEVER generate a positive CLV at customer level. This is the most obvious point that surprisingly seems to be missed by SO many people. A customer with two consecutive orders with steep discounts is extremely unlikely to ever come back without a discount.

7. Your biggest retention problem would be to get the second order. After 3 orders, customers tend to be a lot more likely to repeat further. Funnily, across the businesses I’ve seen – a big chunk of the customers who ever repeat, repeat really close to their first order – same day / within 7 days or max within a month. It’s counter-intuitive for some industries, but true. Wallet cashback on the first order is the sanest monetary-incentive route to try for this.

8. It’s important to fundamentally set a singular user identity in your system. Unique verified mobile number is the most straightforward to handle and most difficult to cheat with. Specially important if you have incentives for first time customers. Kill the possibility for users to create multiple accounts, and engage with the coupon sites only for customer acquisition.

9. Kill the concept of voucher codes. As a user, one should see the offer(s) he’s eligible for through the multi-channel campaigns, then automatically get the offer using a back-end rule engine. Eg. A user coming back after a month is automatically eligible for a 10% cashback. He may not see your reactivation email / remember the voucher code – but if you push this offer over sms, show it to him throughout the checkout journey and automatically give the cashback after order placement, his chances of coming back for the next order are higher. If you are ok to send a reactivation offer to this guy, why not do all you can to get him to convert (multi-channel + on-site campaign + auto-apply offer).

10. Using a wallet that stores cashback, is going to offer much better ROI than giving upfront discounts with or without voucher codes. At least you don’t spend crazy voucher money on one timers who’ll never come back. Of course, this is a matter of what your competition is doing and what user growth your board is expecting. A cashback is indeed less attractive than an upfront discount. Eventually, transactions you ‘buy’ using heavy first-time upfront discounts, won’t result in loyal users.

11. Think hard if you want to build the platform for deal hunters -who’ll always find the cheapest way to get something and whose loyalty is a fragile function of the best discount. If not, find alternate ways to keep the ‘convenience seekers’ engaged, who use your product for the ease it brings to their lives and not purely for the discounts. Every business will have some users like that.

12. If you’re not the ‘simplest’ way to fulfill that particular user need – you’re better off investing in product development / UX refinement, than blowing up money acquiring customers who won’t stick around.

13. As the irrational exuberance fades, improvement in user experience and competing via ‘real value provided’ instead of ‘deeper voucher given’ will have to become your mainstay.

14. Aggressive VC funding backed discounting, is at best a ‘covering fire’ which would give you time to organize a killer team and build up the real valuable / sticky product. Keep burning without building the fortress you’ll one day run out of ammunition and be left to die.

15. Having a ‘data science’ team is in fashion – but without the right mandate and goals, it won’t deliver what you need it to. You’d need to invest in roles like Chief Information Officer and Chief Retention Officer that span the data handling and customer touch-points across the organisation and have he mandate and authority to influence CXO group decision making.

15.1 Having display campaigns based on the NPS response / FB custom Audience campaign for customers who had to call more than twice for their order – seem like great ideas – won’t be possible to execute without the right data infrastructure.

16. It’s great if you understand the value of email marketing enough to employ an ’email marketing manager’, but if you’ve assigned his job to construct a few pretty looking emails every day, you need to fundamentally re-think.

17. As a small / early stage company – you might imagine automation to be not-so-relevant. On the contrary, it makes sense to automate early – so you’re not dependent on a staff member to diligently send out mails daily, and your consumers do see value in your content.

17.1 For segmented display re-targeting, you need to associate the user’s cookie IDs with the Customer ID in your DB. The sooner you start doing this, the more usable data you’ll have.

18. For those of the view that the email content should be engaging enough for the user to consider opening – keep in mind that you’re not running a magazine, but an e-commerce portal/app. Delivering consistent high quality editorial content that’s relevant to your entire audience, is nearly impossible.

19. If you’re worried about being spammy, do keep in mind that the promotions tab didn’t cause a big drop in open rates, but an average user gets dozens of marketing emails a day, so sending less emails is not going to improve your chances of engaging your audience.

20. If you’re able to keep it relevant, higher communication frequency doesn’t necessarily hurt. Enable a simple opt-out for the users who don’t want to hear from you, and let the remaining ones get your messages. At anytime, 2-3% users opt-out. If your friends tell you that you’re spamming people by too many emails / pushes – don’t let anecdotal evidence cloud your judgement about mail frequency. Just focus instead on the opt-out rates or engagement metrics like % of users still opening your mails a month after subscription.

21. Transaction attribution via the traditional routes will always be misleading. Even if you put a link in your voucher SMS campaign, the number of SMS orders showing in Google Analytics would be a tiny fraction of the number of people placing an order using voucher codes. This also messes up orders attributed to FB / Display / SEM in the last non-direct click attribution model.

Test and Control (where a random set of users is always excluded from the campaign) is the only real way to measure true uplift from your CRM campaigns, the wider you apply, the better.

22. Open rates at an overall level can be manipulated by the choice of audience you send to. Send only to people that have Opened-in-15-days and your open rates would be 3x of your average. If you can, track the open rates of this ‘engaged’ group as a KPI alongside the overall average.

23. It’s worth investing in the right data visualisation tools (tableau/sisense etc.) early in the business – the more you orient your organisation to be data driven, the better it is. All key people being able to take good quality base data for granted can have significant impact on their real understanding of the business and hence their ability to do their jobs more effectively.

24. I don’t get why companies advertise ‘voucher codes’ publicly. Most use cases can be covered by promoting the offer and applying it automatically to qualifying users. Unless your profit margins are solely based on eligible users forgetting or not applying the voucher codes anyhow.

25. There are straighter topics like having a notifications inbox (plus badge management for iOs) and delayed push permission for iOS – that help increase the reach for your pushes.

26. Loyalty points programs were setup by offline retailers to unify customer data across transactions. Restaurants can afford giving a 10th Sub free, with their >50% gross margins. Regular loyalty points scheme is nothing but proxy discount – so watch your margins before you go the platinum rewards route.

27. Wallet cashbacks or balance to make a transaction is the route to take for a referral program as well. It can be your growth hacking magic, or just another useless page on your site – depending on your reward structuring and enabling discovery for the referral program. Again, based on verified mobile number.

Hopefully, investors and boards will show maturity in accepting less-than-stratospheric numbers in W-o-W growth and drop the covering fire driven craziness in favour of a more sensible long term play. Else, players are destroying the ecosystem for themselves and for future players. Unviably high CPC bids owing to super aggressive competing funded players – don’t help any one except Google and FB.