Skip to content

CFIA Recall Effectiveness Checks: Proving a Recall Worked

What the SFCR and the CFIA ask you to record to show a food recall worked, who to check first, and a filled-in effectiveness record for a bakery's recall.

By Koa Sterling. Product specialist at TaroStack and small business owner. Yes, I am a real human, and I actually sit in front of a computer and write these articles. Reviewed November 16, 2026 · 7 min read

A recall effectiveness check shows that every customer who received a recalled food was told and acted on it. Under section 84(4) of the Safe Food for Canadians Regulations, you document a recall "including any information that substantiates its effectiveness" and keep that record for two years. The CFIA's recall guide says what goes in it. For each customer, record when and how you reached them, who you spoke to, how much they had, what they did with it, and whether it worked.

The short version

  • One row per customer who received the food, from your distribution list.
  • Call distributors and customers who serve vulnerable people (hospitals, care homes, daycares) first.
  • Record the contact, the person, how much they had, what they did, and the result.
  • Chase everyone who hasn't confirmed. "No answer" is not a result.
  • Add it up: what you made has to equal what you hold, what customers had, and what was sold or used before the recall.
  • The CFIA may check some customers itself, and may ask you to repeat the recall if it wasn't effective.
  • TaroStack records each contact on the recall as your team makes it. More at the end.

What the CFIA asks you to record

The CFIA's guide lists, for each customer on your distribution list:

Field What to write
Date contacted And the time, for the first contact
How Phone, email, in person, letter
Who you spoke to A name, not "the store"
What they had Quantity, and the lot code if they can read it, at their site when the recall started
What they did Destroyed, returned, held and marked recalled, or told their own customers
Result Effective or not
How you know Their signed acknowledgment, a photo, the returned goods counted
If it wasn't effective What you did next

The guide doesn't set a percentage or say how many customers to check. It asks you to follow up with anyone who hasn't confirmed, and it says the CFIA may contact some of your customers itself.

This is general information, not legal advice. Your CFIA inspector has the last word on what applies to you.

Who to contact first

When you can't reach everyone at once, the CFIA's guide says to put distributors and customers who serve vulnerable people first. A distributor has customers of its own to tell, so every hour you save there is an hour saved for all of them. A care home or a hospital serves people who get sicker from the same food.

A bakery's recall, checked

A 14-person bakery in Ontario recalls its sesame burger buns after a seed supplier's peanut warning. It made 1,500 bags across three lots. When the recall starts, 210 bags are still at the bakery and held, 40 were sold over its own counter, and 1,250 went to eight customers. The recall coordinator calls the care home first, then the distributor, then everyone else.

Customer Received Contacted Spoke to Had Did Result
Care home, Hamilton 30 Nov 10, 2:40 p.m., phone Kitchen manager 12 Destroyed Effective
Distributor, Montréal 300 2:55 p.m., phone Buyer 140 Held; told its 6 customers Its customers checked by Nov 11
Grocer A, Ottawa 300 3:10 p.m., phone Store manager 120 Destroyed, photo sent Effective
Grocer B, Hamilton 250 3:25 p.m., phone Buyer 90 Returned Effective, 90 counted back
Café, Gatineau 60 3:40 p.m., phone Owner 0 None left Effective
Grocer D, Kingston 200 3:50 p.m., phone Store manager 75 Held, marked recalled Effective, picked up Nov 12
Grocer F, Toronto 80 4:20 p.m., voicemail and email; Nov 11, 8:15 a.m., phone Manager 30 Destroyed Effective, a day late
Deli, Toronto 30 4:35 p.m., phone Owner 10 Destroyed Effective

The customers had 477 bags between them when the recall started. Add the bakery's side, and every bag is accounted for:

  • 210 held at the bakery
  • 40 sold over the bakery's counter
  • 477 at customers, destroyed, returned or held
  • 773 sold or used by customers before the recall

That's 1,500 of 1,500. The 773 is the number the CFIA will care about most, because it's food already with the public. Whether that needs a public warning is part of the conversation with the CFIA, not the bakery's decision alone.

Grocer F is the row that matters. A voicemail at 4:20 p.m. isn't a contact. The coordinator called again the next morning, and the record says so.

When a recall isn't effective

If a customer can't be reached, still has the food on sale, or can't say what they did with it, the row isn't effective. Write down what you did next: another call, a visit, a call to the customer's head office. The CFIA's guide says it may ask you to repeat the recall if it finds the recall wasn't effective.

Then come the last steps. Keep returned food apart and counted, agree with the CFIA how it's disposed of, and record the disposal. Afterwards, hold a review: what worked, what didn't, and what changes in your recall procedure.

Mock recall effectiveness

A mock recall tests the same thing without the food coming back: whether you can account for every unit and reach the people on your list. In a drill, phone one or two customers to check that the contact on file still reaches someone who could act, and tell them it's a drill. An out-of-date phone number is exactly what a drill should find.

Where this stops working

On paper, the effectiveness record is a sheet the recall coordinator fills in while making calls. The quantities come from invoices on another desk, and the contact names from a spreadsheet nobody has updated. By the end of the day there are notes on three sheets, a voicemail log in someone's phone, and an email thread. The record the regulation asks for gets written up afterwards, from memory, which is the moment details go missing.

How TaroStack does it

In TaroStack, the recall starts from the lot and its record fills in as the recall goes.

Starting a recall holds every affected lot still in stock, and the recall's trace lists every customer who received those lots, with dates and quantities by lot. That's the distribution list and the "Received" column, written as orders shipped. Returned goods come back into the lot they left, so the returns are counted on the same record.

As your team reaches each customer, the contact is recorded on the recall: when, how, who answered and what they had left. The effectiveness record is written while the calls happen, not reconstructed afterwards.

The recall report carries each customer's name and address, and it downloads as one CSV file for the CFIA.

Paper TaroStack
Who received it Invoices, matched to lots On the recall, by lot
Each contact Notes on a sheet Recorded on the recall as it happens
Returns Counted on a pallet Back into the lot, counted
Two years A folder Kept

Lots, recalls and traceability drills are on every TaroStack plan, from $9.99 a month. It's free for 14 days. If you'd like a hand setting it up before you need it, ask, and we'll do it with you.

Questions people also ask

What percentage counts as an effective recall?

The CFIA's guide sets no percentage. It asks for a record of each customer's contact and action, and follow-ups for anyone who hasn't confirmed. A recall where you can't say what happened at one customer isn't finished.

Who does the effectiveness checks?

You do, for your own customers, and the CFIA may check some of them itself. The guide doesn't say how many.

How long do I keep the effectiveness records?

Two years, under section 84(4), along with the rest of the recall's documents.

Do I need to check customers who sold everything already?

Yes. "None left" is a result, and it goes on the record with the date and the person who told you.

Sources

  1. Safe Food for Canadians Regulations, SOR/2018-108, s.84(4) Recall implementation · read November 16, 2026
  2. CFIA: Recall procedure, a guide for food businesses (modified 2023-02-27), steps 6 to 12 · read November 16, 2026

We use AI to help with the research for these articles. Every one is read, checked against its sources and edited by Koa before it's published. Spot a mistake? Tell us and we'll fix it and say so. How we write these.

Let’s get your stack in order.

Get TaroStack

Sign up now for your free 14 day trial. No credit card required.