How to Use Customer Data to Increase Sales

Most Nigerian businesses are sitting on the cheapest sales opportunity they will ever have and cannot reach it, because the record of who bought what is spread across WhatsApp threads, a POS terminal, a delivery notebook and a receipt book.

Acquiring a new customer costs money every time. Selling again to someone who already bought, liked the product and trusts you costs a message. This article covers how to assemble the data, which plays to run, how to prove they worked, and how to stay within your obligations while doing it.

What counts as customer data

Customer data is any record connected to an identifiable buyer. Most businesses have far more of it than they think, sitting in places that are not databases.

• Transaction data — what was bought, when, at what price, in what quantity.
• Contact data — phone number, name, sometimes email, delivery address.
• Channel data — where the customer came from and where they prefer to be reached.
• Behavioural data — pages viewed, carts abandoned, enquiries that never converted.
• Service data — complaints raised, returns, how the issue was resolved.
• Preference data — sizes, flavours, brands, delivery windows, payment method.

For a typical Nigerian SME the useful sources are the payment gateway, the POS statement, the WhatsApp chat list, delivery records, the receipt book and the website if there is one. The work is not collecting new data. It is joining what already exists onto one identifier.

The minimum customer record

You do not need a rich profile to run the plays in this article. You need these fields, and nothing else is required to start:

The phone number is the key. In Nigeria it is more stable than email and more reliable than name. Normalise the format on entry so the same person is not stored three ways, and decide one rule — for example, always store the full international format — and apply it everywhere.

Step one: build one customer table

1. Export what you can. Payment gateway transactions, website orders, POS statements. These usually carry phone numbers or can be matched to them.
2. Type in what you cannot export. Receipt books and delivery notes for the last 12 months. Tedious, but this is usually where the most valuable customers are hiding.
3. Normalise phone numbers. One format, consistently applied, before anything else.
4. Collapse duplicates. Merge rows sharing a phone number, summing purchases and spend and keeping the earliest first purchase and the latest last purchase.
5. Fill in the channel field where you know it, and mark the rest as unknown rather than guessing.
6. Fix the capture point going forward. Every new sale, on every channel, must add or update a row. If this step is skipped, the table is out of date within a month and the whole exercise is wasted.

A spreadsheet is a perfectly good home for this at small volume. A CRM becomes worthwhile when several people update records at once, when follow-ups need to be assigned and tracked, or when the table passes a few thousand rows.

Six plays that increase sales from existing data

1. Reactivate lapsed customers

Take everyone whose last purchase is older than your normal repurchase window but who bought more than once before. These are proven buyers who have simply drifted.

Send a short, specific message referencing what they bought, not a generic broadcast. A reason for the message — new stock in their category, a restock of the item they bought, a seasonal relevance — performs better than a discount, and protects your margin.

2. Remind repeat buyers when they are due

For consumable products — skincare, supplements, pet food, printer supplies, cooking ingredients, medication — calculate the typical gap between purchases and message shortly before the customer runs out. This is the highest-return play available to most product businesses, because it meets a need the customer already has.

The calculation is simple: for customers with three or more purchases, take the median gap between orders. Send at roughly 80% of that gap.

3. Recommend based on what they bought

Group your products into categories and note which pairs are commonly bought together by the same customer over time. Then message customers who bought one and not the other. This requires no algorithm — a sorted spreadsheet reveals the obvious pairs for most SMEs.

4. Follow up enquiries that never converted

Every business has enquiries that received a price and then went quiet. In many Nigerian SMEs nobody follows up more than once, and a single well-timed second message recovers a meaningful share. Log unconverted enquiries with a reason and a follow-up date, and work the list weekly.

5. Treat your highest-value customers differently

Rank customers by total spend and identify the top group. Give them something concrete: early access to new stock, a named contact, priority delivery, or a call rather than a broadcast. The aim is to protect revenue that is already yours, which is far cheaper than replacing it.

6. Ask your happiest customers for referrals

Filter for customers with multiple purchases and no complaints. Ask directly, make sharing easy with a prepared message or link, and give them a reason to do it. Referral is a major acquisition channel in Nigeria and it is almost always left to chance rather than prompted deliberately.

How to segment customers without analytics software

You do not need clustering algorithms. A simple recency, frequency and value split does most of the work and can be built in a spreadsheet in an afternoon.

Score each customer 1 to 3 on three dimensions:

• Recency — 3 if they bought within your normal repurchase window, 2 if within roughly twice that, 1 if longer.
• Frequency — 3 for four or more purchases, 2 for two or three, 1 for a single purchase.
• Value — 3 for the top third by total spend, 2 for the middle, 1 for the bottom.

Then act on the combinations:

Recalculate monthly. The value is in the movement between segments — customers sliding from "best" to "at risk" are the ones worth a phone call this week.

How to prove a play actually worked

Without a comparison, any increase after a campaign gets credited to the campaign, including increases that would have happened anyway.

The method:

1. Define the audience — for example, all lapsed customers who last bought between 90 and 180 days ago.
2. Hold back a portion. Randomly exclude around 10% to 20% and send them nothing.
3. Run the play on the rest.
4. Measure both groups over the same window: orders, revenue and margin.
5. Report the difference, not the total. If the contacted group bought at a higher rate than the held-back group, the gap is the genuine uplift.
6. Count the cost, including messaging fees, staff time and any discount given, and compare it with the uplift in margin rather than in revenue.

A holdout group feels wasteful the first time. It is the only way to know whether a play is worth repeating, and it prevents a business from spending years on campaigns that do nothing.

Consent, privacy and the NDPA

Using customer data for marketing in Nigeria carries obligations under the Nigeria Data Protection Act 2023. This is not legal advice, and requirements change, so verify current expectations with the Nigeria Data Protection Commission or a qualified professional. In practice, sound habits are:

• Collect for a stated purpose and tell customers what you will use their details for at the point of collection.
• Record consent — when it was given, through what wording, and for which channel.
• Make opting out easy and honour it immediately across every channel.
• Keep only what you need. A marketing list does not need payment details or full addresses.
• Restrict access. Not every member of staff needs the full customer table.
• Do not buy contact lists. Beyond consent problems, response is poor and it damages the reputation of your number and domain.

Messaging channels have their own rules on top of the law. Broadcast messaging through the WhatsApp Business Platform involves opt-in and message template requirements set by Meta; check its current documentation before designing a campaign around it.

What changes for Nigerian businesses

The phone number is the identity. Email-based customer data models do not transfer well here. Build everything around a normalised phone number and treat email as optional.

Most history lives in chat. Past purchases recorded only in WhatsApp threads are recoverable but require manual work. Do the 12-month backfill once, then fix capture so it never has to be repeated.

Message channel choice matters. WhatsApp is read; SMS reaches people without data; email is weakest for consumer audiences and strongest for B2B. Match the channel to the segment, and respect the cost difference — SMS is charged per message and adds up quickly on large lists.

Timing follows money. Consumer response is stronger around salary periods and weaker at month end. Seasonal peaks around December, Ramadan, Easter and school resumption are pronounced. Schedule reactivation and replenishment campaigns accordingly.

Trust is fragile. Nigerian consumers are alert to scams. Messages must clearly identify the business, reference a real past purchase, and avoid urgency tactics that read as fraudulent. A verified business profile and consistent sender identity improve response materially.

Data quality is the constraint. Incomplete phone numbers, inconsistent product names and missing dates will break every play above. Budget time for cleaning; it is usually the longest part of the project and the one that determines whether the rest works.

Example (hypothetical): a Lagos pharmacy chain

This is an illustrative scenario, not a Linestech client result.

A three-branch pharmacy takes payments by card, transfer and cash, and keeps a paper record of customers who order repeat medication by phone.

The team spends two weeks building one customer table: gateway and POS exports for card and transfer customers, plus manual entry of twelve months of phone-order records. Duplicates are collapsed on normalised phone numbers. Roughly half the rows turn out to belong to customers who had bought more than once.

They then run three plays:

1. Replenishment reminders for customers buying regular monthly items, sent a few days before the expected reorder date, with a holdout group excluded.
2. Reactivation for customers who bought two or more times but not in the last four months, referencing the category they previously bought.
3. High-value recognition for the top spenders, who are given a direct WhatsApp number for a named pharmacist rather than the general line.

Measurement is built in from the start: each play has a holdout, and results are reported as the difference between contacted and held-back groups, with messaging costs and staff time deducted.

The replenishment play produces the clearest uplift and becomes a standing weekly routine. Reactivation performs moderately and is repeated quarterly rather than monthly. The high-value play produces little measurable short-term revenue but a noticeable fall in complaints from that group, so it is retained on retention grounds and judged on repeat rate instead.

Two of three plays are kept, one is repurposed, and the business knows which is which — because of the holdouts.

A 30-60-90 day implementation plan

Days 1 to 30 — assemble. Export every available source. Backfill twelve months of manual records. Normalise phone numbers, collapse duplicates, and produce a single customer table with the nine minimum fields. Fix the capture point so new sales update the table automatically or as part of the checkout routine.

Days 31 to 60 — segment and run the first play. Build the recency, frequency and value scores. Pick one play — replenishment if you sell consumables, reactivation otherwise. Define the audience, set aside a holdout, write one message, send it, and record results.

Days 61 to 90 — measure, refine, add. Report uplift against the holdout and deduct costs. Keep the play if it paid; adjust timing or message if it nearly did; drop it if it did not. Add a second play, and decide whether the table has outgrown a spreadsheet and needs a CRM or a custom system.

Resist running all six plays at once. Overlapping campaigns to the same customers make results unattributable and risk fatiguing your list in the first month.

Mistakes to avoid

• Broadcasting the same message to everyone. Untargeted blasts produce opt-outs, and a customer who mutes you is harder to reach than one you never contacted.
• Leading with a discount. Discounts train customers to wait for the next one and erode the margin the campaign was meant to produce.
• Skipping the holdout group. Without it, you will keep repeating whichever campaign happened to coincide with a good week.
• Measuring revenue instead of margin. A campaign that lifts revenue while giving away more margin than it earns is a loss.
• Letting the table go stale. If new sales do not update it, the data is wrong within weeks and staff stop trusting it.
• Ignoring consent and opt-outs. Beyond the legal exposure under the NDPA 2023, ignoring an opt-out is the fastest way to lose a customer permanently.
• Collecting fields you will never use. Every extra field slows capture and increases your data protection burden.
• Waiting for a CRM before starting. The spreadsheet version teaches you what the CRM should do, and costs nothing.

Conclusion

The revenue is already in the building. Consolidate every past buyer into one table keyed on a normalised phone number, with last purchase date, purchase count, total spend and last category. Score customers on recency, frequency and value, then run one play at a time — replenishment reminders for consumables, reactivation for everyone else — with a holdout group so you can tell what worked. Keep consent records, honour opt-outs immediately, and measure uplift in margin rather than revenue. Most Nigerian SMEs can do all of this in a spreadsheet before spending anything on software.

If your customer history is spread across chats, a POS terminal and a receipt book, Linestech builds customer databases, CRM systems and the integrations that keep them current for Nigerian businesses.

Frequently asked questions

What is the fastest way to increase sales from existing customer data?

For businesses selling consumable products, replenishment reminders timed just before the customer runs out. For everyone else, reactivating customers who bought more than once but not recently. Both target people who have already bought and trust you, so they need a message rather than a marketing budget.

How do I build a customer list if all my sales happen on WhatsApp?

Work through twelve months of chats once and record phone number, name, what was bought, when and how much in a single sheet. Then fix capture going forward by sending a payment link or short order form so every new sale creates a row automatically. The backfill is tedious but only needs doing once.

Do I need a CRM to do this?

No. A spreadsheet handles every play in this article for a few thousand customers. A CRM becomes worthwhile when several people update records simultaneously, when follow-ups must be assigned and tracked, or when the volume of messaging requires automation rather than manual sends.

Is it legal to message customers who bought from me before?

Contacting your own customers about related products is generally more defensible than marketing to purchased lists, but the Nigeria Data Protection Act 2023 sets expectations on purpose, notice, consent and opt-out. Record when and how consent was given, honour opt-outs immediately, and confirm current requirements with the Nigeria Data Protection Commission or a qualified professional.

How often should I message my customer list?

Frequently enough to be remembered and rarely enough to be welcome. For most Nigerian SMEs, a relevant message every four to six weeks per customer is a reasonable starting point, with replenishment reminders timed to individual need rather than to a calendar. Watch the opt-out rate as your guide.

What is RFM segmentation?

A simple method of scoring each customer on how recently they bought, how often they buy and how much they have spent. Combining the three scores produces practical groups — best customers, at-risk spenders, lapsed buyers, promising new customers — each of which deserves a different action. It needs no software beyond a spreadsheet.

How do I know whether the sales increase came from the campaign?

Hold back a random 10% to 20% of the target audience and send them nothing. Compare the purchase rate of the contacted group with the held-back group over the same period. The difference is the real uplift, and it is the only figure worth using to decide whether to run the campaign again.

What if my customer data is incomplete or messy?

Start anyway with the rows that are usable. A clean list of 400 customers is more valuable than a messy list of 4,000, and running one successful play on the clean portion builds the case for cleaning the rest. Fix capture first so the problem stops growing while you work backwards.

Sources and further reading

Figures, platform rules and regulations change. These are the primary references behind this article and the places to check before you act on it.

• Nigeria Data Protection Commission (NDPC)
• WhatsApp Business Platform — Meta for Developers
• PostgreSQL documentation